WO2010052303A1

WO2010052303A1 - Analysis of ventricular electro-mechanical activation data

Info

Publication number: WO2010052303A1
Application number: PCT/EP2009/064748
Authority: WO
Inventors: Kristoffer Russell; Otto A. Smiseth; Anders Opdahl
Original assignee: Oslo Universitetssykehus Hf
Priority date: 2008-11-06
Filing date: 2009-11-06
Publication date: 2010-05-14

Abstract

The present invention relates to assessment of ventricular dyssynchrony in relation to cardiac ^synchronisation therapy (CRT) using a combination of imaging modalities that display changes in dimension and left ventricular pressure to construct pressure-dimension loops for assessment of the left ventricle. A new way of defining onset of active force (OAF) is based on a pressure-length loop, C(t). Improved assessment of global left ventricular function, markers for regional active force and function, assessment of ventricular synchronicity including distinguishing between mechanical and electrical dyssynchrony and assessment of viability in infarcted myocardium would be advantageous for the about 30 % of non-responders that are selected for CRT on basis of QRS criteria.

Description

ANALYSIS OF VENTRICULAR ELECTRO-MECHANICAL ACTIVATION DATA

Technical field of the invention The present invention relates to assessment of ventricular dyssynchrony that has gained great attention after the introduction of cardiac resynchronisation therapy (CRT) which has shown to be a promising treatment option for patients with heart failure and ventricular electrical conduction delay.

In particular the present invention relates to a combination of imaging modalities that display changes in left ventricular dimension and pressure to construct pressure-dimension loops for assessment of the left ventricle.

Background of the invention

Assessment of ventricular dyssynchrony has gained great attention after the introduction of cardiac resynchronisation therapy (CRT) which has shown to be a promising treatment option for patients with heart failure and ventricular electrical conduction delay. However, in about 30 % of patients who are selected for CRT on basis of QRS criteria, there is no improvement in symptoms (non-responders) after implantation of the CRT device. Therefore a major challenge has been to implement new tools for evaluation of dyssynchrony.

Many echocardiographic (ECG) indices using tissue Doppler Imaging (TDI) and 2D imaging have been devised to improve selection of CRT candidates, but although initial studies were promising echocardiography has not demonstrated to improve patient selection at present.

Assessment of left ventricular (LV) function is fundamental in the evaluation of all cardiac patients. Invasive pressure-volume/dimension-analysis is regarded as the "gold standard" of measurements for LV function. Several simplified invasive- and non-invasive indices have been developed, and one of the most widely used is LV ejection fraction (EF). At present, EF is routinely measured or estimated using echocardiography. However, many of these indices are strongly operator dependant and have been proved to differ in accuracy. Another important tool in clinical practise is the ability to assess regional function of the LV. New imaging tools such as strain either by TDI or speckle tracking have been introduced for this purpose. However, these modalities only assess regional deformation and the ability to expand assessment by for example quantifying regional work would greatly increase its clinical potential.

Summary of the invention

Hence, an improved assessment of global left ventricular function, markers for regional electro-mechanical activation and function, assessment of ventricular synchronicity including distinguishing between mechanical and electrical dyssynchrony and assessment of viability in infarcted myocardium would be advantageous in the clinical setting. Furthermore this could directly aid patient selection for the about 30 % of non-responders that are selected for CRT on basis of QRS criteria.

Thus, an object of the present invention relates to assessment of ventricular dyssynchrony in patients that are potential candidates for cardiac resynchronisation therapy (CRT).

In particular, it is an object of the present invention to provide a method for evaluating an onset of active force (OAF) in left or right ventricular muscle segments.

Another embodiment of the present invention relates to a method for evaluating OAF in right ventricular muscle segments.

Previous attempts in this field have tried to define an onset of mechanical activation (OMA), in left ventricles as the onset of shortening determined from traces of fibre shortening, strain or deformation in left ventricular muscle segments. These have used time to peak strain or peak velocity or time to onset of systolic myocardial velocity for different regions compared to onset of QRS (Prinzen et al, and Chung et al.). The present invention applies a novel concept referred to as onset of active force (OAF) or onset of active force generation (AFG) in the ventricle as the determinative value for evaluating left ventricular function, including differentiating between electrical and mechanical dyssynchrony. Although OAF has some similarities to the onset of mechanical activation (OMA) known from the prior art, there are considerable differences

OAF according to the invention is preferably determined using a parametric curve showing ventricular pressure (or analogues) versus ventricular muscle segment dimension or length (or analogues). Such curves may be constructed using a combination of left ventricular pressure (or pressure analogues) and segment length (or analogues such as strain, strain rate or displacement) in left ventricular muscle segments. These curves typically form loops referred to as P-L loop or just "the curve". In the embodiments of the present invention, OAF is preferably defined as the point where the P-L loop curve deviates from passive behaviour during diastolic expansion/elongation.

On a side note, similar looking ventricular pressure - ventricular volume (P-V) curves are known from the prior art such as from e.g. WO 06/104869, WO 07/022505, WO 03/037428, or US 2008/0195167. These are considerably different from the P-L curves applied in the invention. The dimension or length applied in the curves of the invention is a regional parameter for individual ventricular muscle segments. The ventricular volume applied in the P-V curves of the prior art is a global parameter for the ventricle, and does not correlate with each individual segment dimension/length.

Onset of active force, defined as the deviation from passive behaviour, may result from either:

(a) ventricular muscle segment shortening, or

(b) a shift in ventricular muscle segment elongation.

Whereas (a) is comparable to interpretations of OMA in the prior art, (b) is an entirely new way of looking at activation of ventricular muscle segments. During filling, the flow of blood into the ventricle elongates the muscle segments. The "inertia" in the filling of blood means that the onset of active force in the muscle (b) does not momentarily result in a shortening of the muscle segment, but initially manifests itself as a deceleration in the diastolic elongation, i.e. a shift away from a passive elongation. First when this deceleration brings the elongation to a halt, will the shortening against increasing pressure (a) set in. In other words, (b) is where the muscle segment reacts to its activation, (a) is where this reaction results in a shortening of the muscle segment. In a strong, healthy heart, muscle segment activation almost immediately results in shortening of the muscle and (a) and (b) occurs practically simultaneously. In a weakened, old, or sick heart, (b) typically occurs before (a).

Thus, one aspect of the invention relates to a method for preparing data related to onset of active force in left ventricular muscle segments, the method comprising : A. generating, by means of a computer, a parametric curve C,(t) = (L,(t), P(t)) from concurrent values of a left ventricular pressure, P(t) and a segment length, L,(t), in a left ventricular muscle segment, i, as a function of time;

B. determining an onset of active force in the left ventricular muscle segment, Ci(t₀AF,i), as the first point on a diastolic part of the curve where the parametric curve C,(t) deviates, such as consistently deviates, such as over a period of at least 30 ms, from passive behaviour for the muscle segment, i.e. from the segment's passive elongation during the late diastolic filling phase, and determining the corresponding time, toAF,.;

C. comparing t₀AF,ι with a fixed time marker in the cardiac cycle, t_Cc, and/or with a time for onset of active force for another left ventricular muscle segment j,

In the present context, the terms "passive behaviour" and "passive elongation" designates the behaviour of a segment as it elongates during the late diastolic filling. This is considered normal terminology in the field, but for the sake of clarity, an illustrative explanation is given here with reference to Figure 1. During ventricular filling segments are stretched as volume increases. While filling during the late diastole, a segment is in an inactivated state and is therefore stretched passively (intervals indicated by thick lines). In early diastole a segment has started to relax, but may not be fully relaxed and we therefore concentrate on the late diastolic phase where the segment is known to be completely passive and inactivated. During this late diastolic phase a segments will therefore follow its passive-elastic curve (or passive-trend curve, which describes a segments passive properties). If a segment was completely passive, i.e. if it was not activated, it would continue to stretch along its passive-elastic curve with increasing left ventricular pressure. Therefore, for a segment to deviate from passive behaviour (and shift upwards from its passive-elastic curve/elongation phase) it must generate active force. When a segment is mechanically activated it becomes stiffer and will therefore either start to shorten (Figure IA) or it may continue to stretch but because it has become stiffer a higher LV pressure is now needed to stretch the segment and it will shift upwards from its passive elongation state (Figure IB).

In the aspects of the present invention P(t) may be a directly measured pressure, or may be estimated from secondary data, or may be any analogue thereto which are proportional to the left ventricular pressure (LVP). Different ways of determining P(t) will be described later in relation to Table 1. In a preferred embodiment, P(t) is measured non-invasively, several examples for such noninvasive determination of P(t) will be given.

The length L,(t) may be a measured length, strain, strain rate or displacement of one or more sections in the LV muscle segment, i. The chosen value depends on the imaging modality, e.g. echocardiography, MRI, CTI, ventriculography, sonomicrometry/implantable radio-opaque markers, changes regional left ventricular volume by conductance catheters as a length surrogate, etc. Different ways of determining L(t) will be described later in relation to Table 1. In a preferred embodiment, L(t) is measured non-invasively, several examples for such non-invasive determination of L(t) will be given.

The L and P values are preferably measured in an interval comprising the onset of electrical and active force of the left ventricle. A preferred interval is from at least the onset of the QRS complex in a simultaneously recorded ECG and at least the following 150 ms. Typically, however, L and P values are measured continuously over several heartbeats. Preferably, P and L are measured and values are stored in appropriate electronic memory or storage. Upon carrying out the method of the first aspect, P and L data can be retrieved from this storage by the computer. In alternative embodiments, the method is carried out during measurement of P and/or L. Here, these are preferably measured non-invasively so as not to involve surgical steps. QRS refers to the QRS complex, which is a structure on the electrocardiogram (ECG) that corresponds to the depolarization of the ventricles. A typical ECG tracing of a normal heartbeat (or cardiac cycle) comprises a P wave, a QRS complex and a T wave. A QRS complex refers to a Q wave, an R wave, and an S wave or any combinations thereof.

In a preferred embodiment, onset of active force is determined as the first point where the parametric curve C,(t) deviates from a passive-elastic curve PE(L) for the muscle segment. The passive-elastic curve PE(L) is a trend curve describing the segment's passive properties, or equivalent late diastolic trend or behaviour of C(t). A passive-elastic curve will follow the bottom section of C(t) during which the left ventricle is relaxed and multiple P/L coordinates can be extracted during late diastole (only interrupted by atrial filling) from which the passive elastic curve may be determined by mathematical regression or as a fit to part of this bottom section. Different ways of determining PE(L) will be described later in relation to Table 1.

When we refer to the first point on the parametric curve C(t); the first point that fulfils an indicated criteria on a diastolic part of the curve or in the preferred interval [onset of QRS + 150 ms] or equivalent is meant. When describing the development of C(t) here and throughout the description, we refer to the direction of increasing time, which is counter clockwise in all the depicted loops.

That the parametric curve C(t) deviates consistently means that it does not return towards PE(L) after the first point, e.g. by approaching PE(L) to take a value closer to PE(L) that it had at the first point or by becoming parallel to PE(L) after having broken off at the first point in the preferred interval or in the diastole of the same heartbeat.

Time marker t_Cc is a fixed time in the cardiac cycle, meaning that its position is constant and reproducible from beat to beat and from individual to individual. Such marker may be determined from several modalities, such as: ECG, a reference image-frame in a simultaneously obtained MRI, CT or ultrasound sequence, or the start of increasing pressure, i.e. first time where dP(t)/dt>0 after atrial filling (a wave in LVP) in each cycle. Other applicable markers may exist or be developed.

As the cardiac cycle is triggered by electrical activation, t_cc may preferably be a time marker, t₀EA, fixed to the onset of electrical activation. This is advantageous for use in distinguishing between primary electrical and primary mechanical dyssynchrony. Such maker, t₀EA , may be the onset of, or one or more peaks within (e.g. peak R, peak S), the QRS complex in a simultaneously recorded ECG.

Different ways of determining a t_Cc will be elaborated later in relation to Table 1. The comparing of t_cc and t_0AF,ι may be as simple as presenting values side by side, or may be implemented as a parameter based on the difference between t_Cc and toAF,ι, such as the value t_Cc - toAF,ι or any function thereof.

Alternatively, instead of comparing to a fixed marker, t₀AF,ι Can be compared to toAF_j of other segments, j, thereby evaluating the activation sequence of the different ventricular muscle segments. This provides the advantage that no fixed time marker is needed as only relative inter-segment activation times are compared.

A number of algorithms for determining OAF in accordance with the principles of the invention are presented later, and new ones may be developed. Other ways of defining or determining t₀AF than the one described in the first aspect may be conceivable. Therefore, in a first alternative, step B in the above aspect may be replaced by: Selecting a point in parametric curve C,(t) representative of an onset of active force in left ventricular muscle segment i, C,(t₀AF,ι), the selection being performed in accordance with the definition that OAF is where C(t) deviates from passive-elastic behaviour, and determining the corresponding time, t₀AF,ι-

As OAF from C(t) may preferably be determined by a human user by visual inspection, step B may, in a second alternative, be replaced by: presenting corresponding curves of P(t), L,(t) and C,(t) to a user; and receiving user selection of a point in parametric curve C,(t) for an onset of active force in left ventricular muscle segment i, C,(t₀AF,ι), and determining the corresponding time, t₀AF,ι- Also, if toAF is determined by a computer program using an appropriate algorithm, it may be preferred to have a human checking, approving, and possibly correcting or adjusting the OAF determined by the computer program. Hence, in a third alternative aspect the determined onset of active force is a suggested C,(t₀AF,ι), and wherein determining onset of active force comprises; presenting the parametric curve C,(t) with a marked up suggested C(t₀AF,ι) to a user; and receiving user input related to an optional adjustment of the suggested C,(t₀AF,ι) and an approval of the suggested or adjusted C,(t₀AF,ι)-

In the above alternatives, presenting the parametric curve C,(t) to a user preferably comprises also presenting P(t) and L,(t) with a marked up t_0AF,ι corresponding to the suggested C,(t₀AF,ι)-

Other aspects of the present invention relates to a computer program product, or a computer program product for updating a medical monitoring apparatus, for preparing data related to onset of active force in left ventricular muscle segments, the product comprising software applications which provides the following when executed by a processor or a computer:

A. generating a parametric curve C,(t) = (L,(t), P(t)) of concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in a left ventricular muscle segment, i, as a function of time;

B. determining an onset of active force in the left ventricular muscle segment, Ci(t₀AF,i), as the first point on a diastolic part of the curve where the parametric curve Cι(t) deviates, such as consistently deviates, such as over a period of at least 30 ms, from passive behaviour for the muscle segment, i.e. from the segments' passive elongation during the late diastolic filling phase, and determining the corresponding time, toAF,.;

C. presenting a comparison between t₀AF,i and a fixed time marker in the cardiac cycle, tec, and/or a time for onset of active force for another left ventricular muscle segment j, toAFj-

Dyssynchrony caused by electrical conduction delay is most likely the only etiology amenable for CRT. Introducing a marker that accurately reflect regional electrical activation is therefore of great interest for selecting patients for CRT. In yet a further aspect, the invention provides a medical monitoring apparatus comprising a unit for analysing and presenting data, the apparatus further comprising software means for performing the functions of steps A-D in the previous section.

Medical monitoring apparatus may be apparatus capable of measuring and analysing dimension/pressure changes from a patient, or apparatus capable of receiving and analysing dimension/pressure changes of a patient measured by other apparatus. Typical apparatus may be MRI apparatus, CT scanners, echocardiography machines, as well as image view workstations that may or may not be coupled to any such apparatus.

As mentioned previously, the invention may be applied to determine whether an electro-mechanical dyssynchrony can be characterised as a primary electrical dyssynchrony or a primary mechanical dyssynchrony. This may be of utmost important, since present CRT results in little or no improvement in patients with primary mechanical dyssynchrony, and the invention may thereby be applied to selecting patients with dyssynchrony symptoms for CRT.

Thus in a further aspect, the invention provides a method for determining whether a patient has a primary electrical dyssynchrony or a primary mechanical dyssynchrony, comprising :

- generating, by means of a computer, a parametric curve C,(t) = (L,(t), P(t)) of concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in two or more left ventricular muscle segments, i, as a function of time;

- determining an onset of active force in the left ventricular muscle segments, C,(t₀AF,ι), as the first points on a diastolic part of the curve where each the parametric curve C,(t) deviates, such as consistently deviates, such as over a period of at least 30 ms,from passive behaviour for the muscle segment, i.e. from the segment's passive elongation during the late diastolic filling phase, and determining the corresponding times, t₀AF,ι;

- determining intersegment differences in activation times using the determined times for onset of active force, t₀AF,ι, and evaluating whether the patient has a primary electrical or a primary mechanical dyssynchrony. In this method, if the determined intersegment differences in activation times are insignificant, for example less than 40 ms, less than 30 ms, or less than 20 ms, electrical dyssynchrony may be eliminated. If large intersegment differences in activation times are observed, electrical dyssynchrony may be confirmed.

Also, in yet another aspect, the invention provides a method for selecting patients for cardiac resynchronisation therapy (CRT), comprising

- evaluating whether the patient has a primary electrical or a primary mechanical dyssynchrony using the method according to the previous aspect; and

- selecting patients with primary electrical dyssynchrony for CRT.

This method may be applied to patients with a potential dyssynchrony disorder, and such have typically previously been selected based on analysis of a QRS complex from an electrocardiogram.

The precise determination of active force generation in the left ventricle may be applied to improve or optimize settings and electrode placements for Cardiac Resynchronisation Therapy (CRT) device.

Today's methods relies on global measures such as LVOT flow from echocardiography (interventricular delay) and E og A wave in mitral flow (atria- ventricular delay). These measures do not give information of the regional ventricular function. Further, for electrode placement, measures such as OMA does not give a reliable indication of which ventricular segment is activated the last (which is typically where the lateral electrode should be placed).

Hence, in a further aspect, the invention provides a method for using onset of active force in left ventricular muscle segment i, t_0AF,i, according to the previous aspects, as a marker for mechanical activation in left ventricular muscle segment i. This marker may be applied to patients to better adjust and optimise settings of or trig medical equipment such as CRT.

Thus, one aspect of the invention provides a method for adjusting settings of a cardiac Resynchronisation therapy (CRT) device after implementation, the method comprising - obtaining onset of active force (t_0AF) in ventricular muscle segments, using any method described in previous aspects, and determine relative mechanical activation times of these segments;

- adjusting interventricular (v-v) and/or atria-ventricular (a-v) delays on the CRT device using the determined relative mechanical activation times to optimize or improve an activation sequence of ventricular segments.

Another aspect of the invention provides a method for determining electrode placement of a CRT device - obtaining onset of active force in left ventricular muscle segments using the method according to any previous described methods;

- determining the latest activated segment of the left ventricle;

- placing at least one of the LV CRT leads in the determined latest activated segment.

To further optimise and increase the effects of using CRT devices the electrodes have to be placed at the right locations on a segment. Now only one lead is used but this method could be extended and used if multiple leads were to be placed on the same segment.

Other aspects of the present invention relates to a computer program product for preparing data related to onset of active force in left ventricular muscle segments, the product being executed by hardware receiving concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in a left ventricular muscle segment, i, for a patient as a function of time; the product comprising software applications which provides the following when executed by a computer: A. generating a parametric curve C,(t) = (L,(t), P(t)) of received concurrent values of a left ventricular pressure, P(t), and a length, L,(t), in a left ventricular muscle segment, i, as a function of time; B. determining an onset of active force in the left ventricular muscle segment,

Ci(t₀AF,i), as the first point on a diastolic part of the curve where the parametric curve Cι(t) deviates, such as consistently deviates, such as over a period of at least 30 ms, from passive behaviour for the muscle segment, i.e. from the segment's passive elongation during the late diastolic filling phase, and determining the corresponding time, toAF,.; C. identifying t₀AF,i in the parametric curve as the onset of regional LV mechanical activation;

D. identifying global LV mechanical activation

E. assessing and presenting of LV regional and global function.

A further aspect of the present invention provides a cardiac Resynchronisation therapy (CRT) device wherein the interventricular (v-v) and atria-ventricular (a-v) delays have been adjusted using the method described above.

The basic idea of the invention is to utilise a new way of defining the onset of active force of left ventricular segments to prepare data related to onset of active force in left ventricular muscle segments. The new way of defining OAF is based on a pressure-length loop, C(t), as described, and OAF is defined as where this loop deviates from passive behaviour. The prepared data may later be used to assess types of ventricular dyssynchrony (primary electrical or primary mechanical) in patients that are potential candidates for cardiac resynchronisation therapy (CRT).

Brief description of the figures

Figure 1 is an illustration of the passive behavious of muscle segments during the late diastole.

Figure 2 shows a flow diagram illustrating a method and a computer program product according to embodiments of the present invention.

Figure 3 illustrates a medical monitoring apparatus according to an embodiment of the present invention.

Figures 4A, A', B and B' illustrates the differences between OAF as used in the the present invention and onset of shortening.

Figure 5 illustrates ways of determining C,(t₀AF) in an exemplary pressure-length loop. Figure 6 illustrates the determination of OAF according to an embodiment of the invention. For a given left ventricular pressure (LVP) an inactivated myocardial segment will have a given length. A: The passive-elastic curve is derived from repeated end diastolic segment lengths. LVP measurements describe these passive characteristics for a given myocardial segment, i) Pressure segment length loops during caval constriction, ii) High gain LV pressure (LVP) showing end-diastolic points, iii) Exponential fit to end-diastolic points. B: The definition of onset of myocardial activation is deducted from the construction of the passive elastic curve - for a segment to leave the passive elastic curve it must generate active force. Thus, onset of AFG is defined as the first coordinate of the pressure segment length loop that leads to a deviation from the passive curve. C: Timing of onset AFG was extracted from either LVP or segment length traces.

Figure 7. Schematic illustration of placement of myocardial crystals. IM-EMG, intramyocardial electromyogram; LBBB, left bundle branch block.

Figure 8. Representative traces showing (A) segment length, shortening velocity (dL/dt) and intramyocardial electromyogram (IM-EMG) for an anterior and posterior segment during baseline and ischemia and baseline and caval constriction, and (B) corresponding pressure-length loops.

Figure 9 are representative traces showing (A) segment length, shortening velocity (dL/dt) and intramyocardial electromyogram (IM-EMG) for a septal and lateral segment during baseline and left bundle brach block (LBBB), and (B) corresponding pressure-length loops.

Figure 1OA shows pooled data from all experiments illustrating the variability in time from onset R in ECG to timing of different dyssynchrony indices by sonomicrometry during baseline, load alteration and ischemia. 1OB shows peak intersegment time difference for baseline, load alteration and ischemia. Mean±lSD is indicated for each index.

Figure HA shows peak intersegment time difference during baseline and left bundle branch block (LBBB). HB shows ime delay in activation of the lateral vs. septal wall for different dyssynchrony indices during LBBB. Mean±lSD is indicated for each index.

Figure 12. Relationships between time for onset R in intramyocardial electromyogram (IM-EMG) to onset active force generation (AFG) and peak myocardial ejection velocity (S) by sonomicrometry (12A) and echocardiography (12B), measured from onset R in ECG. Data from all interventions are included. Time to onset AFG showed strong correlation with time to onset R in IM-EMG by sonomicrometry as well as echocardiography.

Figure 13. Assessment of onset active force generation (AFG) by LV pressure (LVP) and strain by speckle tracking echocardiography (STE). 13A shows representative traces for anterior (thick line) and posterior (thin line) segments during ischemia. 13B shows representative traces from septal (thick line) and lateral (thin line) segments during left bundle branch block (LBBB). Strain measurements are performed by STE in parasternal short axis view. In addition, electrical activation for the two walls is measured by intramyocardial electromyograms (IM-EMG). Pressure-strain loops are constructed by combining strain by STE and LVP. Aortic valve opening (AVO) and closing (AVC) indicated by arrow.

Figure 14 is a chart showing segment length, shortening velocity (dL/dt) and intramyocardial electromyogram (IM-EMG) for an anterior segment during ischemia, illustrating that a myocardial segment may stretch even after it has been activated.

Figure 15 is a flow-chart over the architecture of a software application for preparing data related to determination of onset of active force in left ventricular muscle segments.

The present invention will now be described in more detail in the following.

Detailed description of the invention

As described previously, the invention can be implemented as a method, as a computer program product (software), as software in a data analysis unit of a medical monitoring apparatus, or as software for updating a medical monitoring apparatus. In the following, embodiments of the different implementations or aspects will be described, and a detailed example of a clinical application of the invention will be described.

It should be noted that embodiments and features described in the context of one of the aspects of the present invention also apply to the other aspects of the invention.

Figure 2 represents a flow chart 19 for illustrating the architecture of an embodiment of a software product in accordance with an aspect of the invention, such as a computer program product for preparing data related to onset of active force in left ventricular muscle segments. In addition, the flow chart 19 illustrates an embodiment of the method for preparing data related to onset of active force in left ventricular muscle segments in accordance with another aspect of the invention. Some of the steps are optional or serve to illustrate the flow of data, and are thus not part of the broadest aspects of the invention as defined by the claims.

In box 1, data 2 representing concurrent L(t) and P(t) are accessed or received. This data may e.g. be received directly from dedicated apparatus for measuring segment length/strain/strain rate/displacement and LV pressure, respectively, from a local memory such as RAM or a hard disk, or from a remote location accessible via a network connection. Different ways of determining L(t) and P(t) will be described later in relation to Table 1. When there are data, L,(t), from more segments, the above and following processes are carried out for each segment, i, either in parallel or successively. Box 1 and data 2 serves to illustrate the flow of data.

In box 3 the parametric Pressure-Length curve, C,(t), is generated, with concurrent values of L,(t) and P(t) as first and second coordinates in an L-P coordinate system. C,(t) is generated at least within the interval [t_QRS ; t_QRs+150 ms], where t_QRS is the time for onset of QRS complex determined from a simultaneously recorded ECG or equivalent. Preferably however, C,(t) is generated for an entire cardiac cycle. Depending on the sampling rate (measurements per time) of L(t) and P(t), it may be advantageous to connect data points in C,(t) using an interpolation algorithm so that a continuous curve is obtained. Thereby, an OAF laying between two data points may be selected.

In box 4, a passive-elastic curve, PE,(t) for the ventricular segment is either accessed, received or generated. PE(L) may be generated specifically for the ventricular segment in question, or may be a generalised expression or standard curves for different types of ventricular segments (axial, lateral, posterior, anterior), which may be fitted to the actual P and or L values in C,(t). Different ways of determining PE(L) will be described later in relation to Table 1, and Box 4 serves to illustrate the flow of data.

As described previously, several alternatives exist for the determination or selection of OAF to be used, and these are illustrated by paths 5, 6 and 7 in flow chart 19, which thereby represent alternative embodiments.

In one embodiment corresponding to path 5, a point representative of OAF is selected in C,(t) in box 8 using an appropriate algorithm or assessment method. The selection may be carried out by a user or a software application, or a combination of these. A multitude of different algorithms and assessment methods are described later in relation to Table 1.

In another embodiment corresponding to path 6, a first suggested OAF is determined in box 9 using an appropriate algorithm or assessment method. The parametric Pressure-Length curve, C,(t), is then presented to a user, box 10, with the suggested OAF, C,(t₀AF), marked up as shown in the inserted curves 12. Optionally, the pressure and length traces are also presented with marked up t_0AF- The purpose of this is to allow the user to assess the validity of the suggested OAF in C,(t), e.g. by support in P(t) and L,(t), and to correct C,(t₀AF) if not in agreement, box 11. Such correction if performed via a software application can e.g. be carried out by selection, dragging and dropping the circle indicating the suggested OAF to the a new position on C,(t). Also in box 11, the user can approve the suggested or the corrected OAF. For the functionality of box 11, the software application comprises or can make use of a user interface (UI) such as a mouse and a graphical user interface (GUI) such as cursors and drag/drop functionality provided by most computer operating systems.

In another embodiment corresponding to path 7, the parametric Pressure-Length curve, C,(t), is presented to a user, box 13. Optionally, the pressure and length traces are also presented. This presentation is equivalent to the curves 12 without indication of a suggested OAF. Here, the user can select an OAF in C,(t), possibly with support in P(t) and L,(t), by selection, dragging and dropping a circle indicating OAF to the a position on C,(t). The above comments relating to potential application of UI and GUI applies here as well. Thereby, the user provides his/her selection of OAF, box 14.

Based on the determined or selected OAF from either of paths 5-7, the time for onset of active force for segment i, t₀AF,ι, can be determined, see box 15. This may be by reading the t-values for the L(t) or P(t) sampling points corresponding to the determined OAF, or by using the interpolation algorithm used to connect data points for C,(t).

In box 16, a fixed time marker in the cardiac cycle, t_Cc, is obtained. Such time marker may be received from another apparatus, and different ways of determining a t_Cc will be elaborated later in relation to Table 1. The function of the time marker is to provide a consistent reference point within the cardiac cycle for timing of t₀AF,ι , as t₀AF,ι may otherwise be a floating value.

Now, a comparison between the obtained t_Cc and t₀AF,ι are made in box 17 as the result of the preparation of data according to embodiments of the invention. As an example, the comparison may result in a parameter Δt = t₀AF,ι - tec which represents a time of onset of active force which is fixed in the heart cycle. Box 18 illustrates an optional display of such comparison as a table listing parameters Δt for different left ventricular segments.

In a further optional step, these parameters may be compared to each other as well as to a time for onset of electrical activation to evaluate global and/or regional left ventricular dyssynchrony and assess whether such dyssynchrony is primary electrical or primary mechanical. Figure 3 illustrates another implementation of the invention, here an embodiment of a medical monitoring apparatus 20, with a unit 21 for preparing and presenting data. Typically, the apparatus will have a display 22 for presenting data to a user in a GUI, and a UI 23 such as a mouse and keyboard for receiving user input. Exemplary apparatuses could be echocardiography machines, MRI apparatuses, CT scanners etc., their data processing units and their analysis software (work station).

The apparatus may comprise or be connected to units, apparatus or systems, 24 and 25, for measuring P(t) and L(t) respectively, as described previously in relation to Table 1. In the alternative, the apparatus 20 may access P(t) and L(t) data over a network connection 26. Computer program products for performing data preparation as described in relation to Figure 2 can be stored in memory 27 and executed by processor 28 of the unit 21.

The invention can also be embodied by a computer program product for updating a medical monitoring apparatus to prepare data related to onset of active force in left ventricular muscle segments. Such product can be embodied as a packet managing system or an installation program for downloading and installing the software described in relation to Figure 2 on the apparatus 20 described in relation to Figure 3 over the network connection 26. Such program can be stored and executed by memory 27 and processor 28, or stored and executed by a server (not shown) over network connection 26. Description of measurements, parameters, values etc from Table 1

In the following, more detailed descriptions of the various values, parameters and algorithms that are relied upon in determining OAF from a P-L loop will be given (if not given elsewhere in the application). These are also summarized in Table 1.

Passive-elastic curve, PE(L) Selecting or suggesting an OAF may in some cases require that a passive-elastic curve is determined. The determined PE(L) can be shown as a curve plotted into the P-L loop to facilitate visual selection of OAF or to be used in algorithms for determining OAF. PE(L) can be an algebraic function, e.g. resulting from a regression, fitting or extrapolation of one or more p-L loops Table 1. An estimate of PE(L) can also be determined from measurements from different passive lengths or preload levels (caval constriction, etc.)- A non-linear regression (i.e. PE(t) = be^KL) of end-diastolic length (L_ed) - pressure (P_ecι) coordinates, can be constructed including a measure of variance, see also Figure 6A described later.

An estimate of the passive elastic curve can be made from non linear regression of measured data points (L,P) after onset of QRS in ECG. In a specific example, such regression can be made using an exponential equation : P=be^κv, where constant b always equals 0,43 (determined from experimental studies), and k can be calculated by using the end-diastolic (ed) pressure and length with the formula K= (In Ped - In 0.43)/V_ed. Similar regression equations can be set up using strain, strain rate, or displacement of ventricular muscle segments instead of length.

An estimate of the PE-curve can also be made with the measured length-pressure points during the transition from diastole to systole. Starting from two datapoints (n=-2) before onset of QRS (n=0), a non-linear fit representing the PE-curve including a variance measure can be drawn. For each new datapoint (n = l,2... until about 120 ms (end of IVC)), the following will be evaluated; 1) a) is the new datapoint statistically shifted relative to the estimated PE-curve? b) if yes; is the shift permanent within this early systole? If yes: datapoint n-1 is OAF (end loop). 2) If no in 1); calculate new PE-curve incl. variance measure including datapoint n and loop again with next datapoint.

A number of applicable methods for estimating PE(t) are described in more detail in Glantz et al.

Left ventricular pressure and analogues, P(t)

The left ventricular pressure as a function of time, P(t) can be measured invasively or estimated by a non-invasive measurement technique. As it is the dynamics, i.e. variation as a function of time, and not the absolute numerical value that determines the shape of the pressure/length loop, any time resolved absolute or relative pressure signal may be used for preparing data related to onset of active force. In one example, LVP can be estimated non-invasively utilizing microbubble-based ultrasound contrast agents. Pressure dependant changes in the first, second, and subharmonic amplitudes of the ultrasound contrast agents may yield a dynamic pressure estimate facilitating assessment of OAF.

Alternatively an estimated P can be determined in patients with mitral regurgitation by estimating the velocity profile on the mitral regurgitation jet and a simplified Bernoulli equation. This may be performed on any image modality that allows estimation of the regurgitation jet velocity or similar.

In another alternative, facilitating assessment of OAF without invasive LVP measurement, P can be estimated via linear or non-linear (power, exponential) functions. LVEDP can be estimated depending of clinical condition (non-heart failure 10, heart failure 20 mmHg), by echo measurements (E/e') or neglected (0 mmHg). The measured systolic cuff pressure (or peripheral (radial) arterial pulse wave form) can be utilized to estimate LV pressure at end of IVC (LVEIVCP= lower diastolic estimated aortic pressure) as a given percentage of systolic cuff pressure. The time for EIVC can be measured by measuring aortic flow by Doppler. Depending on the function for estimated time course of LVP during IVC, the estimated start- and stop coordinates, t_OnSetQRs, LVEDP and t_Eivc, LVEIVC, can be utilized to determine the time course for LVP, and used in combination with length or a length analogue to determine OAF. Measurements of aortic pressure can be done by a blood pressure device and aortic valve opening pressure can be estimated. This is then time shifted so that systolic rise coincides with aortic opening by Doppler. (O'Rourke et al.)

Another technique to determinate the pressure could be to use an apex cardiogram. This is a technique of recording pulsations of the chest wall produced by the beating heart. Comparison with measurements of left ventricular pressure made using micromanometer pressure catheters has shown that the upstroke and down stroke of the apex cardiogram is virtually synchronous with the rise and fall of ventricular pressure (Willems et al). When we are assessing onset AFG we are only dependant on early filling and the initial upstroke of ventricular pressure, therefore measurements from an apex cardiogram could be used as a non invasive pressure analogue for assessing onset AFG. Left ventricular segment lengths and analogues, L,(t)

The length of a section in a left ventricular muscle segment can be measured by several image modalities. Also, as described previously, there exist a number of length analogues (e.g. strain, S, strain rate SR, and displacement, D) that display an equivalent variation as a function of time, and which may be used instead.

The length, L(t), strain, S(t), strain rate SR(t), or displacement, D(t) of sections in different left ventricular muscle segments may be determined by sonomicrimetry, conductance measurements (regional volumes), ventriculography (radioopaque markers), tagged- and non-tagged MRI, echocardiography (TDI and STE) and by CT (multi modality tracking). The modalities provide the parameters automatically, or semi automatically with varying degree of user interaction, see Anderson et al. for description of different techniques.

Parametric curve, Ci(t), or pressure-length loop The parametric curve C,(t), or P_L loop, for LV segment i is generated using corresponding P (measured or estimated) and L (or S or D) values for LV segment i. The P_L loop need not be determined for the full cardiac cycle, can e.g. be determined only from onset of QRS complex in ECG and next 150 ms.

Onset of active force generation (OAF) vs onset of shortening/contraction Until muscle segments are activated they exert passive behaviour. The later activated segments must shorten against a greater resistance because LVP has started to rise. This means that such an activated segment must first overcome the LVP before shortening can ensue. Therefore onset of shortening and OMA as determined from strain measurements may not reflect true active force, but rather the, typically later, time in which a segment has generated a greater force than the resistance it is contracting against.

OAF according to the invention takes this into consideration by including both length (or strain analogues) and LVP in the determination. These combined data are used to form a pressure-length loop, and OAF is defined as the point where this loop deviates from passive behaviour.

The deviation from passive behaviour may be either (a) ventricular muscle segment shortening against increased pressure, or (b) a shift in ventricular muscle segment elongation against increased pressure. Whereas (a) is comparable to interpretations of OMA in the prior art, (b) is an entirely new way of looking at activation of ventricular muscle segments. According to (b), an "inertia" in the increasing pressure means that the activation does not momentarily result in a shortening, but initially manifests itself as a deceleration in the continued elongation. First when this deceleration brings the elongation to a halt, the shortening against increasing pressure will set in.

Figures 4A and A' shows representative pressure-segments length loops. The loops in Panel A. shows representative pressure-segments length loops and passive trend curves (dotted lines) with identification of onset active force generation (circle AFG). Figures 4B and B' shows LV pressure with high gain and illustrates how onset AFG was defined in segments with early-systolic lengthening. In these cases onset AFG corresponded to onset of an upward-shift of the pressure-segment length loop relative to the passive curve. Also seen in the loops of 4B and B' are where the onset of shortening (onset of mechanical activation) happens (marked with an X). As can be seen in these examples, the onset for these two activations does not have to take place at the same time.

The task at hand is therefore to develop algorithms for selecting a point in parametric curve C,(t) representative of an OAF in left ventricular muscle segment i, Ci(t₀AF,i), the selection being performed in accordance with the definition that OAF is where C(t) deviates from passive-elastic behaviour.

Selecting onset of active force, C(t_OAF) The following describes how a user can determine C(t₀AF) in the P-L loop

(parametric curve C(t)). As P-L loops can take many different shapes, and since the loop can leave the pressure-elastic curve in many different ways, it is difficult to set up an algorithm that will determine the correct OAF in all cases. However, using just a few generalised rules and little training, most users will be able to correctly asses OAF in almost any P-L loop by visual inspection. As an example, the following set of instructions for a user on how to select OAF in a length- pressure loop can be used. It is presupposed that axes are oriented as in Figure 6A: 1. Locate the length-pressure loop, when moving around the loop, counter clockwise designates forward direction.

2. Locate the passive-elastic curve (PE), or if not indicated extrapolate a trend curve starting from the lower left corner of the loop. 3. Locate lower right corner of the loop, where the loop leaves the PE/trend curve directly and consistently, this corner is the OAF.

4. If the corner is not be a single, sharp turn, then find the point where the loop can be said to leave the PE/trend curve directly and consistently by approaching the corner in both forwards and backwards direction. o Forward direction. Start from the lower left corner of the loop, if the loop leaves the PE/trend curve and then makes small loops or backtracks towards the PE/trend curve then this is not a direct and consistent leaving of the PE/trend curve. Proceed forward till next candidate point. o Backward direction; Start near the upper right corner of the loop and go backwards until you reach the PE/trend curve, or an asymptote or parallel thereto.

Naturally, the selection of OAF, whether performed by a software application or a user, can also be based on one of the algorithms for suggesting an OAF described in the following. Whether an OAF selected by use of such algorithm need to be approved by a user depends on the detailed implementation and factors such as the performance of the algorithm, the complexity of the P-L loops in the relevant class of patients, the requirements to the stability or precision in the selected OAF, and others.

Suggesting an onset of active force, C(t_OAF)

In order to assist the selection of OAF, it may be advantageous for a software application or an apparatus utilising the invention to give a first suggestion of OAF to the user. If in agreement, the user can then approve the suggestion and the suggested OAF will be used in the further analysis. If not in agreement, the user can adjust the suggested OAF, and then approve the adjusted OAF which will then be used in the further analysis.

Algorithms for determining OAF

In the following, a number of suggested algorithms for determining OAF are described. These can be used in a method according to an embodiment of the invention, or in a software application or an apparatus utilising the invention. These algorithms can also be used by a software application or an apparatus to determine a suggested OAF.

5 In one algorithm, Cι(t₀AF,ι) can be determined as the first deflection point that results in C,(t) leaving a region defined by the passive elastic curve PE(L) ± K. By leaving is meant that C,(t) takes values outside this region in the direction of increasing time. Also, a deflection point is a point where the parametric curve has having a clearly identifiable change of gradient or inclination, or where its gradient 10 deviates from the gradient of PE(L), see also next algorithm.

Figure 5 illustrates the region 30 defined by the passive elastic curve PE(L) ± K. The parameter K can be empirically determined, or can be a statistical parameter such as K = σ(PE) or K = 2σ(PE), where σ(PE) is the variance of PE(L). The first 15 point of deflection 31 that results in C,(t) leaving the region 30 is also shown, as is a earlier point of deflection 32 that does not result in C,(t) leaving the region 30.

In another algorithm Cι(t₀AF,ι) can be determined as the first point where the gradient of C,(t) deviates from the gradient of PE(L) over a period of at least 30 20 ms. This can be expressed as a criteria: dP{t) dPE(L)

>β dL£i) dL

The time period of 30 ms can be a longer or shorter period. The time period serves to ensure that the gradient deviation is not simply a small loop as for point 25 32 in Figure 5, but that it results in that C,(t) breaks off for good, β can be adjusted so that small fluctuations in the bottom part of C,(t) are disregarded.

In yet another algorithm C,(t₀AF,ι) can be determined as the first point where:

and for which dP(t)/dt > 0 over the next 30 ms.

UC₁ ( I) d 2*,C, (t)

Here, dt ^t^{1 is tne vector} product between the asymptote to

C(t), dt , and the direction of the rate of change of C(t), j_t ² The expression thereby finds the angle between the asymptote to C,(t) at time t and the direction of the rate of change of C,(t) at time t, and define points where C_ι(t) breaks off from its bottom section with an angle larger than a threshold angle θ_τ. The bottom section, as stated elsewhere, can be seen as a trend curve for PE(L). Applicable values of threshold angle θ_τ can be determined empirically, and suggested values are θ_τ= 20°, θ_τ= 25°, θ_τ= 30°, θ_τ= 35°, or θ_τ= 40°.

In an alternative and less formalistic formulation, this algorithm can be expressed

(IC₁ ( I ) ^> as the first point from where a direction of an asymptote, dt , changes more than θ_τ over a period of 30ms.

The following describes a number of exemplary algorithms that can be used to determine deviation from passive behaviour without using the passive-elastic curve PE(L).

Polynomial fitting from both directions, determine intersection. For data points (L,(t),P(t)) in middle-bottom part of loop (e.g. from onset of QRS) and in the forward direction, fitting a first n^th-order polynomial to C,(t). The fitting i first made using the first M points (e.g. M = IO), resulting in polynomial expression P_M. After the M^th point, the residual/error of each new point (m) in relation to either P_M or P_m-i is evaluated; if above a certain threshold value, the m^th point is not included in the fitting (to not include get rid of small loops and not include a slow bending in the regression). If the residual/error is below the threshold value, the m^th point is included in the fitting resulting in a new polynomial expression P_m. Continue until given time (QRS + ??ms) or until a consistently large residual/error is observed. A similar fitting of the middle-right part of the loop (e.g. QRS + ??ms) and in the backward direction, resulting in a polynomial expression Q_M.

The intersection between P_m and Q_m is determined, and the point on Cι(t) closest to this intersection is selected as C₁Ct₀AF)- Determining to point on a parametric curve that is closest to a give point off the curve is considered basic calculus and within the realms of the person skilled in the art.

Change of gradient

On a pressure-length loop, the gradient is the same as the derivative of the function. The change of gradient can therefore be interpreted as the second derivative of a function. Or in other words: the maximum change of gradient is the point where the function has the largest curvature, or where it bends the most. It is envisioned that OAF will most often be the point in the lower right corner with the largest change in gradient.

The derivative of a function is a mathematical term related to continuous functions. In practice, this is never the case. A loop consists of discrete depth measurements. In order to compute the change of gradient, algorithms can be used, which approximates the second derivative. When examining the change of gradient, for determining OAF, one is often not interested in every local variation in the loop, but rather the change of a tendency over a larger interval. When approximating the derivative using finite differences, the algorithm is similar to the definition of the derivative except that because the function is not continuous, the limit value which tends towards zero is exchanged with a finite difference.

is replaced with

/Hs₁ ) d Δv_; = y« - y,-

This is called the forward difference method. A similar method which involves X₁ and Xi-i is called the backward difference method. The second derivative may be calculated by using the result of one method as input to the other. It can be proved by using Taylor's Theorem that this approximation gives good results if the input points are regularly sampled ( ΔX|-i = Δx,). It is important to note that the derivative at the point x, only involves the value at x, and its two neighbouring points. The algorithm is therefore very sensitive local variations. Two solutions to this problem may be 1) use more data points than the nearest neighbours on each side, and 2) use methods for generating more points between the original data points.

Below is presented : • A method applying the simple differential using only nearest neighbour

• A method applying a simple sliding average for the second derivative.

• A method applying a more complicated average gradient expression

• Methods for increasing the density of points

Change of gradient - simple first and second derivative The first derivative giving the gradient of the loop at point m (time t_m) can be defined as:

The second derivative giving the change of gradient at point m (time t_m) can be defined as: dG,. G - G

G' = m-l

CiL₁ (O LXtJ- LXt_nJ

Calculating G' for each data point in the forward direction will allow a determination of the point with the largest change in gradient, G'_maχ- If there are small loops on the loop, G'_maχ is likely to be on this, and would have to be discriminated manually.

Change of gradient - one-sided gliding average

The gradient of the loop at time t_m can be defined as:

To remove the sensibility to local variations, the change in gradient ΔG_m is defined as the difference from the mean of the gradients at the N previous points (e.g. N = 5), written as:

ΔG_m = \G_m - Mean_N(G_m ) , where

This ΔG_m may be used to select OAF as the first point with a considerable difference to the average of the preceding points. One algorithm could be: IF AG_1n = \G_m - Mean _N (G _m ) >- Threshold THEN t_m = t ¹ _cOAF i

Best values for N and Threshold could be determined empirically. This algorithm would also react on small loops, which should therefore be discriminated manually.

Two-sided change of average gradient

In order to compute the change of gradient at a point, one method uses the average gradient in a given interval before and after the point, and compare the two to find the change of gradient in that point. The algorithm can use two different methods for computing the average derivative at an interval. These two methods are explained in the article ""The effects of using different algorithms for calculating the foot of slope based on the maximum change of gradient" by Jon Mugaas found on www_;ggocap.,.π.o.

All the above algorithms can be applied in a process where the criteria of the algorithm are tried successively for each data point on C,(t) in the direction of increasing time, e.g. starting at onset of QRS complex.

Time for onset of Active force, t_OAF

The time for onset of active force, t₀AF is the look up time, e.g. in P(t) or L(t) measurements, for a selected, determined or suggested C(t₀AF)- If a selected, determined or suggested C(t_0AF) lies between (L(t) , P(t)) data points, a value for toAF can be interpolated or extrapolated using neighbouring data points. Fixed time in cardiac cycle, t_cc, and comparisons between t_cc and t_OAF,\

The fixed time in cardiac cycle, t_Cc is a time marker with a constant and reproducible position from beat to beat and from individual to individual. A t_Cc can be obtained through a reference image-frame in a simultaneously obtained MRI, CT or ultrasound sequence, e.g. image-frame, external making, extra spike on ECG.

A tcc can also be obtained from start of increasing pressure e.g. the first time in cycle where dP(t)/dt>0 after atrial filling (a wave).

In another alternative, a t_cc can be the onset of electrical activation, t_0EA, determined from peaks in QRS complex in ECG e.g. onset of QRS/first deflection of the QRS complex, onset of Q, R or s wave, peak Q, R or S wave or a time that refers to any of these points (eg +- 50ms) or a measurement of onset of systole defined by the first "kick" of the heart measured by micromanonetry over the apex of the heart, see e.g. Malonas et al.

In another alternative, peaks from global strain, S_G(t), may be used to define onset of shortening as a reference marker, i.e. onset deformation in global strain can be used as a marker for t_cc- S_G(t) is calculated as an average function of typical 6-12 strain values from different segments and can often be automatically generated by imaging devices, such as echocardiography machines, when the individual strain measurements are performed.

The times t_Cc and t₀AF,ι can be compared to determine the time for active force of LV muscle segment i, as there may otherwise be a risk that t₀AF,ι is floating. The comparison can be parameterized by a function of t_cc and t_0AF,ι , such as the difference t_cc - t₀AF,ι- A difference between onset of electrical and active force for LV muscle segment i may be determined as Δt, = t_0EA - t₀AF,ι-

In an alternative embodiment, the time for onset of active force for segment i, toAF,ι, is compared to time for onset of active force for other ventricular segments instead of to a fixed marker. Hence, time for onset of active force is determined for two or more muscle segments in the ventricle, and these are compared to each other in order to determine abnormal delays between these. Such inter-segment comparison could be made by generating differences Δt,_7j = toAF,ι - toAFj, where i is the first activated segment, and j counts over all the following segments. In an alternative approach, segment i denotes the segment that physiologically should be activated first, and j counts over all the other segments. In yet another alternative, Δt,_7j can be a matrix with all permutations of i and j.

The following table summarises some of the various parameters, values and expressions applied in embodiments of the invention.

from the sample before and at onset QRS. Create an estimated PE-curve with these two samples. For each new sample; 1) evaluate if sample deviates from the PE-curve (if yes and permanent- the deviation point is OAF). If no- 2) include sample in PE-curve calculation and evaluate next sample from step 1) e) the first point where: Finds the angle between the asymptote and the

> θ_τ direction of the rate of

change and define points and for which dP(t)/dt > 0 where C(t) consistently over the next 30 seconds. breaks off from its bottom section ~ trend curve for PE(L)) with an angle of more than θ_τ.

toAF: time for OAF a) Look up time for selected

measurements b) Interpolate or extrapolate from (L(t) , P(t)) data points

t_cc: fixed time in Time marker with a constant the cardiac cycle and reproducible position from beat to beat and from individual to individual a) reference image-frame in a

Table 1 Example

In the following, an example involving a clinical application of an embodiment of the invention will be described in detail. The specific apparatus, method steps, parameters, values, algorithms applied in the example represents different embodiments of the invention, but should not be construed as limiting the scope of the invention to these specific embodiments.

Experimental setup Fourteen mongrel dogs of either sex and with body weight ± kg were anesthetized, ventilated, and surgically prepared. In addition pacemaker leads (Medtronic) were attached epicardially on left ventricular lateral free wall, right atrium and endocardially in right ventricular outflow tractus adjacent to the septum. The National Animal Experimentation Board approved the study. The laboratory animals were supplied by Centre for Comparative Medicine, Rikshospitalet University Hospital, Oslo, Norway.

Heamodynamic Measurements

Continuous monitoring of aortic, right and left atrial and ventricular pressures was measured by micromanometers (MPC-500, Millar Instruments Inc) and fluid catheters for calibration purposes.

Sonomicrometry

Sonomicrometry crystals with intramyocardial electromyocardiograms (IM-EMG) were placed in the apex and circumferentially around LV equator and in RV free wall. We also implanted an additional crystal anterio-apical and posterio-apical in LV. Four circumferential and two longitudinal segments were analyzed for group with LAD occlusion, during LBBB two additional longitudinal segments were analyzed. The crystals were connected to a sonomicrometer (Sonometrics Corp), and data were digitized at 200 Hz.

Echocardiography

We used a Vivid 7 digital ultrasound machine (GE Vingmed Ultrasound AS Horten Norway) with a phased-array ultrasound transducer. The frame rate of the recordings was ± per second. We recorded 2-dimensional colour-coded TDI images of 3 consecutive heart cycles. Recordings were performed in apical 4 chamber, 2 chamber and parasternal short axis views.

Calculations of onset of active force (OAF) For a given left ventricular pressure (LVP) an inactivated myocardial segment will have a given length. The passive elastic curve derived from repeated end diastolic segment length - LVP measurements describes these passive characteristics for a given myocardial segment (Figure 6A). We constructed the passive elastic curve by obtaining end diastolic points for a given segment through caval constriction (Figure 6A i and ii). We then drew an exponential line through the end diastolic point generated using these points in graft pad (Figure 6A iii). Further we deduct that for a segment to leave this passive elastic curve it must generate active force. OAF was defined as the time when the myocardial pressure-segment length coordinate deviated from its passive-elastic curve (Fig. 6B). Calculation of time to OAF was done by taking the pressure-segment length coordinate for OAF finding the corresponding time on the ECG trace (Fig. 6C). If a segment is completely passive it will not deviate from the passive elastic curve. We define these segments as inactivated and OAF can not be measured in these circumstances. To quantify a segments deviation from its passive curve we used two confidence intervals as cut off.

Calculations of OAF by echocardiography and LVP Speckle data and LVP from the same beat were identified by their proximity to a spike from an impulse generator used during data sampling. Echocardiography and sonomicrometry data were then synchronised in an excel sheet using spline interpolation. Thereafter, pressure dimension loops were constructed and OAF identified. Because we did not perform measurements by echocardiography during caval constriction passive elastic curves were not constructed. The identification of OAF was therefore based on visual assessment, defined as the first marked deviation of the pressure dimension curve after R in ECG.

Velocity and strain by sonomicrometry

Strain was calculated as percentage of end-diastolic dimension. Regional myocardial velocity was calculated by differentiation of segment length tracings. Time to peak systolic strain (TPS), time to onset peak myocardial ejection velocity (ToS), and peak (TS) and time to OAF (TOAF) were measured relative to onset R in ECG and adhering IM-ECG. These are all common parameters in echocardiography, and are considered known by persons skilled in that field.

Hemodynamic reference

End diastole was defined as onset R in ECG. End systole was defined as dP/dt min.

Experimental Protocol

After a 30-minute stabilization period, baseline recordings were performed. To avoid interference between sonomicrometry and Doppler, we first recorded pressures and ECG and echocardiographic data during 10 seconds and then pressures, and ECG and segment lengths during the subsequent 10 seconds. Data were recorded with the ventilator off. Changes in Loading Condition

Preload was reduced by transient caval constrictions. Hemodynamic variables were allowed to return to baseline values before the start of each intervention.

Left bundle branch block and biventricular pacing Left bundle branch block (LBBB) was induced (n=6) using an ablation catheter (Sulzer Osupta). QRS ± ms. Biventricular pacing was performed; DDD, AV interval 80-100ms VV interval 4ms LV was first activated.

Myocardial Ischemia

Ischemia was induced by placing a suture around the left anterior descending artery and occluding it using a patch. Recordings were performed during baseline and during 15 min of ischemia. Ischemia was assessed by looking at myocardial dysfunction as measured by sonomicrometry. The heart was dissected into 1- 1,5cm slices and was stained with 1% solution of TTC for 15-30min in 37C. Two dogs showed minimal subepicardial infarct in ant-apical and apico-septal region of the LV.

Statistical Analysis

Values are expressed as mean±SD. For multiple comparisons we used 1-way ANOVA (Graph Pad Prism version 4.02 for Windows, Graph Pad Software).

Global changes in load During load alteration OAF remained constant and synchronous. However, onset S, S and PS showed variation in time from onset R in ECG. The greater the load alteration the greater the time variation suggesting that these parameters are load dependant. The mechanisms of load dependent contraction patterns are complex and fall outside the scope of this study. However, afterload dependent asynchronous wall motion has previously been described. One could therefore assume that a global reduction in preload also would affect the LV in a heterogeneous fashion accounting for our observations of increased time difference between segment using echocardiographic indices that are based on wall motion. As, previously described the heart is not a homogenous material nor is it symmetric, so changes in global loading can differentially affect regional loading and thus the timing of local contraction and relaxation as observed by wall motion. This is further strengthened by the fact that we observed great variation in % change of end diastolic length between baseline and caval constriction in the different segments, suggesting that this preload alteration has a heterogeneous effect on regional load.

Regional changes in contractility - ischemia When regional ischemia was induced primary mechanical dyssynchrony to corresponding region occurred. We observed a tendency of delayed local electrical activation in the ischemic region, however, this was not significant (P=O.71). A similar tendency was observed regarding transmural electrical activation (P=O.059). These findings correlate well with previous work assessing subendcardial electrical activation during ischemia. Because the magnitude of electrical delay from baseline to ischemia was < 5ms and our sampling rate during recordings were at 5ms one could argue that the changes observed merely reflect measure error. However, we can conclude that the electrical delay (if present) can not account for the changes in the dyssynchronous contraction pattern observed during ischemia. OAF during this intervention remained constant hence true active force of the segment remained unchanged. It has previously been showed that ischemia induces an inhomogeneous contraction pattern because fibre shortening is affected by ischemia independent of the effects on electrical activation. However, onset S', S' and PS show increased variation reflecting primary mechanical dyssynchrony pattern. We therefore conclude that ischemia may induce primary mechanical dyssynchrony and is most likely not amenable for CRT.

Regional changes in load and contractility - LBBB and BVP with and without ischemia

During LBBB there was dispersion of electrical activation recorded by local IM- ECG^" s as expected. For this intervention OAF also showed great variation in time from onset R in ECG, but correlated well to changes in IM-ECG. However, the time from local IM-ECG to OAF remained constant confirming that electrical dyssynchrony was present but showing no evidence of electro-mechanical (E-M) delay. This is in contrast to previous studies looking at E-M delay on the epicardium. An increased E-M delay in the latest activated segments during RVOT pacing has been showed by others. However, there are several differences in study design that may explain these findings. Firstly, as described in relation to onset of active force (OAF) vs onset of shortening, the latest activated segment during LBBB must shorten against a greater resistance because LVP has started to rise. This means that an activated segment must first overcome the LVP before shortening can ensue. Therefore onset of shortening may not reflect true active force, but rather the time in which a segment has generated a greater force than the resistance it is contracting against. This is further supported by experiments by others, who found that the so called prestrech observed in late activated segments do not display the characteristics of passive stretching are depolarized during the lengthening and therefore presumably able to generate force.

Secondly, onset of both electrical and active force arises in the endocardium and subsequently spread to the epicardium. Others have previously described a slower propagation velocity of myofibre shortening (0.25m/s) compared to electrical conduction velocity (0.49 m/s) from endo to epicardium which could explain differences in E-M delay depending on left ventricular wall thickness. During subsequent BVP time from R in ECG to OAF was resynchronized compared to baseline. These findings further strengthen OAF^' s ability to differentiate between electrical and mechanical dyssynchrony.

To predict responders for CRT it is important to remember what kind of dyssynchrony that is amenable for this treatment. Dyssynchrony as observed by echo parameters gives us information about events during systole. These parameters reflect onset of shortening but because they are measured a considerable time after true onset of active force they are subject to other factors that arise during systole. When selecting patients for CRT one needs to be able discriminate between primary electrical dyssynchrony (amenable for CRT) and primary mechanical (probably not amenable for CRT) which current echocardiographic indices fail to do. With this study we have highlighted some of the parameters that may influence regional contraction in the left ventricle and using OAF we can show that these are independent of electrical activation (this may help to explain the non-responders to CRT and also findings of dyssynchrony by echocardiography with narrow QRS). We therefore propose that OAF may be used to identify LV dyssynchrony caused by disparity in active force. Scientific study

In the following, a scientific study relating to assessment of onset of active myocardial force generation are presented

Background: Better tools for measuring left ventricular (LV) electrical dyssynchrony are needed when evaluating patients for cardiac ^synchronization therapy. This study investigates if onset of active myocardial force generation (AFG) may serve as a surrogate for timing of electrical activation.

Methods: In anesthetized dogs (n=14) with LV micromanometers we measured segment length by sonomicrometry and strain by speckle tracking echocardiography (STE). Onset R in intramyocardial electromyograms (IM-EMG) was used as reference method for timing of regional electrical activation. Onset AFG was calculated from myocardial pressure-segment length and pressure-strain loops and was defined at time of upward-shift from the passive curve. Dyssynchrony was quantified as peak intersegment time difference and as standard deviation (SD) of timing for 6-8 segments.

Results: IM-EMG indicated synchronous electrical activation of all segments during baseline, reduced preload and myocardial ischemia. After induction of left bundle branch block increments in peak intersegment time difference and in SD of timing were similar for onset R in IM-EMG and onset AFG. Timing of onset AFG correlated well with timing of onset R in IM-EMG (r=0.90 and 0.81 for sonomicrometry and STE, respectively). Time from onset R in IM-EMG to onset AFG remained unchanged during all interventions, indicating constant electromechanical activation-time. Myocardial shortening velocity and strain had limited ability to reflect timing of regional electrical activation.

Dyssynchrony is defined as uncoordinated regional myocardial contractions and may in principle have the following etiologies; 1) Electrical conduction delay which causes non-uniform timing of myocyte depolarization, 2) abnormalities in excitation-contraction coupling, and 3) abnormal myocardial contractility or load which cause regional delay in fiber shortening. This implies that mechanical dyssynchrony may have electrical as well as non-electrical etiologies, and these need to be differentiated since delay in electrical conduction is most likely the only etiology amenable to CRT. In the present study we will refer to the different etiologies as primary electrical dyssynchrony, excitation-contraction related dyssynchrony and primary mechanical dyssynchrony, respectively. The latter two we also refer to as non-electrical etiologies of dyssynchrony. We believe that clear differentiation between etiologies is essential for the understanding and appropriate clinical interpretation of dyssynchrony indices.

The general objective of this study was to establish a method which can differentiate between electrical and non-electrical etiologies of left ventricular (LV) dyssynchrony. As reference method for timing of regional electrical activation we used onset R in intramyocardial electromyograms (IM-EMG). As reference method for mechanical activation we introduce onset of active myocardial force generation (AFG) calculated from regional myocardial pressure-segment length and pressure- strain loops. Electromechanical activation time was used as an index of excitation- contraction coupling, and was measured as time from onset R in IM-EMG to onset AFG. Mechanical dyssynchrony was measured as regional differences in timing of myocardial shortening velocity and strain. By exclusion, dyssynchrony was categorized as primary mechanical when it could not be attributed to delay in electrical activation or prolongation of electromechanical activation time.

In the present study electromechanical activation time remained constant during a wide range of interventions. This implies that onset AFG had a constant time delay relative to local electrical activation, and we therefore propose onset AFG as a surrogate for timing of electrical activation. The specific objectives of the study were to test the hypotheses that onset AFG represents a means to quantify LV primary electrical dyssynchrony and to differentiate between electrical and nonelectrical etiologies of LV intraventricular dyssynchrony. In addition, we evaluated the ability of myocardial shortening velocity and strain to serve as markers of primary electrical dyssynchrony. The study was carried out in a dog model during different loading conditions, during myocardial ischemia and after the induction of left bundle branch block (LBBB). Methods

Animal preparation

Fourteen mongrel dogs of either sex and body weight 32.5±3.2 kg were anesthetized, ventilated, and surgically prepared. This included partial splitting of the pericardium from apex to base and loose resuturation of the pericardial edges after completion of instrumentation. Pacemaker leads were attached epicardially on the left ventricular lateral wall and right atrium and endocardially in the right ventricular outflow tract close to the septum. In 8 dogs a snare for performing occlusion was applied around the left anterior descending coronary artery (LAD). In the remaining group of 6 dogs, LBBB was induced as described below. The National Animal Experimentation Board approved the study. The laboratory animals were supplied by Center for Comparative Medicine, Rikshospitalet University Hospital, Oslo, Norway.

Induction of LBBB A 7F ablation catheter (Celsius, Biosense Webster, Inc., CA) was introduced via a carotid artery, advanced to the LV apex and then pulled back to the basal septum where the left bundle potential was identified. Radiofrequency energy was delivered at a location with a large left bundle potential. This position was at a relative distance 2/3 from the atrial and 1/3 from the ventricular signal and was delivered in the temperature mode with a set temperature of 50 degrees Celsius and 30 watts. Energy was delivered 30 seconds after LBBB had been induced.

Hemodynamic measurements

Aortic, left atrial and LV pressures (LVP) were measured by micromanometers (MPC-500, Millar Instruments Inc, Houston, Tex). A fluid-filled catheter placed in the left atrium served as an absolute pressure reference for the LV micromanometer.

Sonomicrometry and regional electromyograms

In each dog 2 mm sonomicrometry crystals (Sonometrics Corp., London, Ontario, Canada) with bipolar electrodes for measuring IM-EMG were implanted endocardially and epicardially as illustrated in Figure 7. All endocardial crystals were implanted in the inner third of the LV wall. In three dogs there were technical problems with one of the mid-apical dimension crystals and we used the distal apical crystal instead. Data were digitized at 200 Hz. The epicardial crystals were attached to the epicardial surface.

Echocardiography

A Vivid 7 ultrasound scanner (GE Vingmed Ultrasound AS, Horten, Norway) was used to record color-coded TDI images in apical 4 and 2 chamber views. In addition, conventional 2-D grayscale images (frame rate 63±13 s^"1) of the LV equatorial short-axis were acquired for speckle tracking echocardiography (STE).

Data analysis

Timing of regional shortening velocity and strain Regional myocardial shortening velocity was calculated by differentiation of segment length tracings and segment length by sonomicrometry was used as an analog for strain. Timing of onset- and peak myocardial shortening velocity during ejection (S) and peak systolic strain were determined by sonomicrometry. In addition, timing of onset and peak S were determined by TDI and peak systolic strain by STE. Aortic valve opening defined as start of upstroke of aortic pressure, was used to define time of onset ejection. Aortic valve closure, defined as peak negative dp/dt was used to define end of systole.

Timing of regional electrical activation

Timing of regional electrical activation was measured as onset R in IM-EMG, defined as the first deflection of more than 20% of total QRS amplitude.

Timing of onset AFG by sonomicrometry and LVP

The time of onset AFG was determined by analyzing myocardial pressure-segment length loops, and was defined as the time when the pressure-segment length coordinate was shifted upwards relative to the passive curve for the same segment. The calculation of onset AFG is illustrated in Figure 6. The passive curve was constructed by an exponential fit to a series of end-diastolic pressure- segments length coordinates obtained during caval constriction. Since the pressure-segment length relationship provides no timing information, onset AFG was extracted from a corresponding time point in either the pressure or the segment length curve (Figure 6c). In segments with no sharp deflection from the passive curve after onset of R in ECG (12 of 266 segments) we used 2 confidence intervals of the fitted passive curve as cut off to define a shift from a passive to active state.

Timing of onset AFG by speckle tracking echocardiography and LVP Onset AFG was also assessed by combining LVP with strain by STE. Strain traces extracted from equatorial short-axis and 2 chamber views were used as substitutes for the segment length traces in the AFG analysis. Because strain represents a relative value, this analysis does not provide a range of end-diastolic pressure-dimension relations, and the diastolic portion of each loop was used to define the passive state. Identification of onset AFG was based on subjective, visual assessment, defined as the first marked upward deviation of the pressure- strain loop that resulted in a continued upward shift after onset of R in ECG.

Electrical conduction time and electromechanical activation time Left ventricular electrical conduction time was calculated as time from onset R in ECG to onset R in IM-EMG. Electromechanical activation time was calculated as time from onset R in IM-EMG to onset AFG.

Quantification of LV dyssynchrony:

Left ventricular dyssynchrony was quantified by two different approaches; 1) as peak intersegment time difference, measured as time difference between the earliest and the latest activated segments, and 2) as standard deviation for 6-8 segments of time from onset R in ECG to timing of each of the indices, and will be referred to as SD of timing.

Experimental protocol

Baseline recordings were performed after a 30-minute stabilization period following completion of instrumentation. To avoid interference between sonomicrometry and echocardiography, recordings were performed. Data were recorded with the ventilator off. In all animals caval constriction was performed to enable construction of passive curves.

The experimental protocol included measurements during baseline (n=8), during reduced preload by transient caval constrictions (n=7), and during myocardial ischemia by LAD occlusion for 15 minutes (n=8). Measurements during caval constrictions were performed after a 20% reduction in peak LVP. In the remaining 6 animals LBBB was induced. Recordings were done during baseline, LBBB and in 4 of the animals during biventricular pacing (BVP), with AV interval 80-100 ms, VV interval 4 ms, and with first activation of the LV lead. Statistical analysis

Values are expressed as mean±SD. Variables were compared using least-squares linear regression and Bland-Altman methods. For multiple comparisons we used repeated measurements ANOVA with LSD post test (SPSS 15.0, SPSS Inc, Chicago, IL). A value of P<0.05 was considered significant. Baselines for each group of interventions were compared to load alteration, ischemia, LBBB and BVP, respectively.

To assess interobserver variability 20 segments from 6 experiments including all interventions were randomly selected and analyzed by 2 independent observers, using the interclass correlation coefficient (α value) and Bland-Altman method.

The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agreed to the manuscript as written.

Results Figures 8 and 9 are representative traces showing (A) segment length, shortening velocity (dL/dt) and intramyocardial electromyogram (IM-EMG) for a septal and lateral segment during baseline and left bundle brach block (LBBB), and (B) corresponding pressure-segments loops with passive curves. In 8A and 9A, dotted lines define timing of first onset R in IM-EMG (R), onset active force generation (AFG) (O), onset of ejection velocity (Δ) and peak systolic shortening (D) . The actual onset for these parameters for each trace is also shown by the same symbols in each trace. Aortic valve opening (AVO) and closing (AVC) indicated by arrow. In 8B and 9B, onset AFG is identified with a circle.

Timing of regional electrical activation by IM-EMG Figures 8 and 9 show IM-EMG traces with representative, distinct R waves which were used as markers of regional electrical activation. The recordings demonstrate that onset R was essentially simultaneous in all segments except during LBBB. The SD of timing of onset R in IM-EMG was 4±lms during baseline, reduced preload and ischemia, and peak intersegment time differences were 10±3, 10±4 and 9±2ms, respectively, indicating synchronous electrical activation of all segments (Table 2, Figure 10, lower panel). Figure 10, upper panel displays pooled data from all animals and shows very limited variability in timing of onset R in IM-EMG during baseline, reduced preload and ischemia. Furthermore, when comparing onset R in subepicardium and subendocardium from all interventions, we found no significant changes during any of the interventions

Induction of LBBB caused an increase in QRS duration from 68±5 to 116±7 ms (P<0.05). This was associated with a marked delay in onset R in IM-EMG in the lateral wall relative to septum, indicating primary electrical dyssynchrony (Figure 9). Standard deviation of timing for onset R in IM-EMG increased from 6±6 to 22±2 ms (P<0.05) after induction of LBBB, and peak intersegment time difference from 15±7 to 53±20 ms (P<0.05). With biventricular pacing QRS duration decreased to 71±12 ms (P=ns), SD of timing decreased to 7±2 ms and peak intersegment time difference to 19±7 ms (P=ns), indicating electrical resynchronization.

Timing of regional mechanical activation by onset AFG

Figures 8A and B and 9A and B illustrate how onset AFG was identified by combining LV pressure-segment length loops and passive curves. In non-ischemic segments with normal intraventricular conduction, onset AFG was represented by a sharp deflection in the lower right corner of the LV pressure-segment length loop, and in most cases coincided with onset of shortening. In ischemic segments, however, there was early-systolic lengthening, and onset of shortening was markedly delayed and did not coincide with onset AFG. This is illustrated in Figure 8, which shows that onset AFG during ischemia was represented by a sharp bend in the pressure-segment length loop near end-diastole, with continuing lengthening rather than shortening. There was simultaneously an upward-shift of the pressure-segment length loop relative to the passive curve, which confirmed that the segment generated active force while lengthening. Similarly, during LBBB the late-activated lateral wall demonstrated early-systolic lengthening, and onset AFG was represented by a bend in the pressure-segment length loop, and an upward-shift relative to the passive curve. Similar to electrical activation by onset R in IM-EMG, SD of timing and peak intersegment time difference for onset AFG demonstrated very limited variability except during LBBB (Table 2, Figure 10). Induction of LBBB caused delay in onset AFG in the lateral wall relative to the septum (Figure 9). Peak intersegment time difference for onset AFG increased from 13±6 to 55±8 ms (P<0.05), comparable to an increase for onset R in IM-EMG from 15±7 to 53±20 ms (P<0.05) (Table 2). Furthermore, the increase in SD of timing was similar for onset AFG and onset R in IM-EMG. Therefore, primary electrical dyssynchrony during LBBB was mirrored quite accurately by the delay in onset AFG (Figure HA). BVP caused partial reversal of these changes, consistent with better synchronization (Table 2). As shown in Figure 12A (left panel) there was a strong correlation between time to onset AFG and time to onset of R in IM-EMG.

Similar findings as those found using sonomicrometry were achieved when onset AFG was obtained from LV pressure-strain loop assessment using STE (Table 3). This is illustrated by representative traces in Figure 13. Onset AFG by STE also correlated well with time to onset R in IM-EMG (r= 0.81) (Figure 12). Good agreement between onset R in IM-EMG and onset AFG was observed by mean difference 14.5±7.4 and 17.9±10.2 ms for onset AFG by sonomicrometry and STE, respectively.

The electromechanical activation time, measured as time from onset R in IM-EMG to onset AFG, was essentially similar during baseline, reduced preload, ischemia and LBBB, with mean values ranging from 12 to 16 ms for the different interventions (Table 2).

Timing of regional mechanical activation by shortening indices Figures 8 and 9 show representative examples of shortening indices measured by sonomicrometry, and their relationship to regional electrical activation by onset R in IM-EMG, and to mechanical activation by onset AFG. During baseline, reduced filling and myocardial ischemia, the variability in peak intersegment time difference for the shortening indices was substantial, and far exceeded the variability in onset R in IM-EMG and in onset AFG (Tables 2, 3 and 4, Figure 10). Similarly, the SD of timing for shortening indices exceeded the values for onset R in IM-EMG and onset AFG. Induction of LBBB markedly increased SD of timing and peak intersegment time difference for all shortening indices (Table 2). However, the ability of shortening indices to accurately measure electrical delay for a given segment was limited. This is illustrated in Figure HB which shows that onset S and peak S in most cases indicated that the lateral wall was activated prior to the septum during LBBB, whereas electrical activation by onset R in IM-EMG demonstrated the opposite. As demonstrated in Figure 9, IM-EMG indicated that the septum was electrically activated prior to the lateral wall and this was associated with marked septal shortening during isovolumic contraction and simultaneous lengthening of the late activated lateral wall. However, as illustrated by this experiment, the delay in electrical activation of the lateral wall was not reflected by similar delays in peak S and peak strain.

Inter observer variability Measurements of onset AFG by STE strain and LVP analyzed by 2 independent observers showed a mean difference between the 2 analyses of -1.1±3.4 ms. The intraclass correlation coefficient between the 2 observers was 0.99.

Discussion

The present study introduces assessment of onset AFG as a method to quantify LV electrical dyssynchrony and to differentiate between dyssynchrony with primary electrical and primary mechanical etiologies. In a clinical setting this differentiation is essential since only dyssynchrony with primary electrical etiology can be treated by CRT. The principle behind this novel method is that onset of active force generation is the first mechanical sign of actine-myosin interaction, and in contrast to indices based on myocardial velocity and strain, it is independent of loading conditions and contractility.

Onset AFG was defined as the time when the LV pressure-segment length coordinate leaves the passive diastolic curve, and was compared to onset of electrical activation defined as onset R in IM-EMG from the same myocardial segment. During changes in loading, myocardial ischemia and LBBB, the electromechanical activation time was essentially constant. Accordingly, onset AFG, which directly measures mechanical activation, was an accurate measure of electrical activation as well. This method was feasible not only with sonomicrometry, but also when strain by STE was used as an analog for segment length, suggesting a potential for measuring onset AFG clinically during left heart catheterization. Using onset R in IM-EMG as reference method for electrical activation we demonstrated that onset AFG was superior to conventional timing indices based on myocardial velocity and deformation. There was a trend that onset of ejection velocity performed better than peak values of ejection velocity and systolic strain.

Relationship between regional electrical and mechanical activation

In the present animal model we measured each of the three components that may contribute to LV dyssynchrony; i.e. delay in electrical conduction, delay in electromechanical activation and delay due to mechanical factors such as reduction or delay in regional force development. Electrical conduction time was measured from onset R in ECG to onset R in regional IM-EMG, and electromechanical activation time from onset R in IM-EMG to onset AFG. By exclusion, delay was attributed to mechanical factors when onset AFG was synchronous throughout the ventricle. During changes in load and myocardial ischemia there was significant LV mechanical dyssynchrony, but there were no regional differences in intraventricular conduction time measured in the subendocardium or in endo-to-epicardial conduction time. Measurements from multiple IM-EMG electrodes showed that all segments were activated within 9±3 ms, indicating almost instantaneous electrical activation of the entire ventricle during these interventions. Furthermore, the electromechanical activation time remained unchanged during all interventions. Therefore, delay in electrical conduction via the Purkinje system and changes in electromechanical activation time did not contribute to LV dyssynchrony during ischemia and changes in load. These findings imply that the observed dyssynchrony during changes in load and ischemia has "non-electrical" causes and therefore represents primary mechanical dyssynchrony, most likely caused by non-uniformities in regional contractility or wall stress.

As predicted, during LBBB there was LV dyssynchrony by velocity and strain indices, and there was delay in intraventricular conduction by IM-EMG, confirming the presence of primary electrical dyssynchrony. Therefore, in the present study mechanisms of LV dyssynchrony were either primary electrical as in LBBB, or primary mechanical as during changes in load and myocardial ischemia. The observation that electromechanical activation time remained unchanged during all interventions implies that onset AFG has potential to become a means to measure electrical conduction delay and to identify mechanisms of LV dyssynchrony as being either primary electrical or primary mechanical, or a combination of the two.

The finding in the present study that electromechanical delay remained unchanged after induction of LBBB is in apparent contrast to previous studies which have used epicardial sensors to measure electromechanical delay. Prinzen et al showed increased electromechanical delay in the latest activated segments during pacing in the right ventricular outflow tract. However, the different findings in our study may be explained by the different definition of mechanical activation : i.e. onset AFG vs onset of shortening. The latest activated segment during LBBB must shorten against a greater force than the early activated segments since LVP has started to rise. This means that the latest activated segment must overcome the LVP generated by early activated segments before shortening can ensue. Therefore onset of shortening may not reflect true onset of mechanical activation. Such a mechanism was proposed by Prinzen et al⁷ to explain discrepancy between timing of electrical activation and onset of shortening.

Although myocardial ischemia caused no significant delay in electrical activation in ischemic myocardium, there was a slight, statistically non-significant increase in time from onset R in ECG to onset R in IM-EMG (Table 4). Possibly, more sensitive methods with higher temporal resolution could have detected minor changes in electrical conduction. However, because the magnitude of electrical delay from baseline to ischemia was <5ms (< temporal resolution of measurements), we can conclude that electrical delay (if present) cannot account for the observed dyssynchrony during ischemia. Our findings are consistent with the work of Ruffy et al which showed no significant delay in subendocardial electrical activation during ischemia. They are also in agreement with the findings of Delhaas et al who showed that ischemia induces inhomogeneous contraction patterns because fiber shortening is affected by ischemia, independent of effects on electrical activation. Therefore, as shown in the present study ischemia induces primary mechanical dyssynchrony and is not likely to respond to CRT. Relationship between regional electrical activation and shortening indices During load alterations and ischemia there were marked regional differences in timing of myocardial velocity and strain indices, whereas onset R in IM-EMG and onset AFG indicated synchronous electrical and mechanical activation in the different LV myocardial regions. Therefore, the segmental differences in time to onset S, peak S and peak strain could not be accounted for by differences in onset of electrical activation or onset AFG and were attributed to primary mechanical dyssynchrony.

The mechanisms of load dependent contraction patterns are complex and the investigation of these fall outside the scope of this study. However, Miura et al described afterload dependent asynchronous wall motion. One could therefore assume that a global reduction in preload also would affect the LV in a heterogeneous fashion and may account for our observations of an increased time difference between segmental wall motion by echocardiographic indices. As stated by Kass the heart is not a homogenous material nor is it symmetric, thus changes in global load can differentially affect regional load and thus timing of local contraction and relaxation as observed by wall motion. This is further strengthened by the fact that we observed great variation in change of end- diastolic length between baseline and caval constriction in the different segments in each of the animals. On average the segment that exhibited the least change of end-diastolic length altered length by 1±1% while the segment with the most change altered end-diastolic length by 11±2%, suggesting that global preload alteration had a heterogeneous effect on regional load.

Limitations

The present study utilized a heavily instrumented animal model and this preparation may not always represent normal physiology. Although the open- chest condition and instrumentation may have induced some degree of LV dysfunction during baseline, this should not modify the main conclusions from this study.

The temporal resolution of measurements of timing of onset R in IM-EMG and onset AFG were 5 ms, and smaller regional differences in timing may not have been detected. However, the magnitude of electrical and mechanical dyssynchrony was in the order of 40-100 ms and therefore the temporal resolution was sufficient for the purpose of exploring mechanisms of dyssynchrony.

The use of onset R in ECG as time reference for calculation of LV intraventricular conduction would tend to underestimate the time intervals. Ideally one should use IM-EMG in the earliest activated segment or a His bundle signal. However, for measurement of regional differences this does not make a difference, and therefore should not change our conclusions regarding the role of intraventricular conduction.

Potential for clinical application

The present experimental study suggests that not only is presence of dyssynchrony and its magnitude important, but that the etiology of dyssynchrony should be defined when evaluating candidates for CRT. Echocardiographic indices provide quantitative information about the magnitude of dyssynchrony, but do not provide conclusive information regarding etiology. Because regional differences in timing of myocardial ejection velocities and strain represent the sum of all mechanisms that may contribute to dyssynchrony, these echocardiographic indices have limited ability to identify electrical conduction delay. This may help to explain why some patients with dyssynchrony are non-responders to CRT and why echocardiography in patients with narrow QRS may fail to identify patients who may benefit from CRT. One should search for those candidates who have primary electrical dyssynchrony as opposed to primary mechanical dyssynchrony since the latter is most likely not amenable to CRT. As demonstrated in this study, conventional echocardiographic indices fail to differentiate between these two etiologies, whereas onset AFG represents a means to identify patients with primary electrical dyssynchrony. Since most patients who are evaluated for CRT undergo left heart catheterization, invasive pressure is available and pressure- strain loops can be constructed. We therefore propose that onset AFG may be used to identify LV dyssynchrony caused by electrical conduction delay and may be used as a reference method for future search for markers of primary electrical dyssynchrony. Conclusions

The present study demonstrates that onset AFG is an accurate marker of timing of regional electrical activation, allowing for differentiation between primary electrical and primary mechanical dyssynchrony, independent of regional differences in load and contractility. Furthermore, it shows that current indices based on myocardial shortening velocity and strain have significant limitations, and although they measure dyssynchrony, they are unable to establish underlying etiology. Further studies should be performed to investigate if onset AFG can be used clinically for identifying patients who may benefit from CRT. Table 2. Hemodynamic and timing variables

Load alteration and Ischemia Left Bundle Branch Block and (n=8) Biventricular Pacing (n=6)

Caval LAD-

Baseline constriction occlusion Baseline LBBB BVP

(n=8) (n=7) (n=8) (n=6) (n=6) (n=4)

Hemodynamic variables

Heart rate, 129±21 133+19 132+22 121± 17 122± 13 120±28 bpm

QRS width, ms 67+3 66+5 63+3 68±5 116±7* 72± 11

Systole, ms 268+41 254+29 263+49 277± 16 331± 10* 306±33^;

LV dP/dt_max,

1488+237 991+295* 1371+331 1202±224 969± 107* 1154±210 mmHg/s

LV EDP, mmHg 9+1 5+2* 11+3 10±3 10±3 9±4

Electromechanical delay Time from onset R in IM- EMG to 14+7 15+6 16+9 12±7 13±7 18± 10 onset AFG, ms

Dyssynchrony variables

Time to onset R in IM-EMG, ms

4+1 4+1 4+1 6+3 22+2* 7+2

SD of timing 10+3 10+4 9+2 15+7 53+20* 19+7

Peak intersegment difference

Time to onset AFG, ms

4+2 3+2 5+3 5+2 22+4* 10+3

SD of timing 9+5 8+5 13+8 13+6 55+8* 27+6*

Peak intersegment difference

Time to onset S, ms

14+4 18+6 20+5* 15+6 38+7* 19+7

SD of timing 36+11 43+12 50+13* 41+15 103+19* 53+20

Peak intersegment difference Time to Peak S, ms

17+7 28+9* 24+9* 15+6 38+11* 14+4

SD of timing

41+18 76+23* 65+21* 43+18 110+33* 38+8

Peak intersegment difference

Time to peak systolic strain, ms

SD of timing 11+4 18+9 21+9* 14+1 23+11* 19+5

Peak 27+10 45+22* 53+27* 41+20 69+38* 57+23 intersegment difference

Values are mean+SD. LV dP/dt_max, maximal time derivative of LV pressure; LV EDP, LV end-diastolic pressure; IM-EMG, intramyocardial electromyography; AFG, active force generation; S, myocardial shortening velocity during ejection.

*P<0.05 vs. baseline.

Table 3. Dyssynchrony as assessed by echocardiography

Baseline Ischemia LBBB

(n=8) (n=8) (n = 6)

Peak intersegment time difference,

9±7 42±15* ms, by:

8±10 54±23* 29±10

Onset AFG by STE 62±35* 40±30

26±13 71±32 62±26

Onset S by TDI 29±13

Peak S by TDI 48±30

Peak systolic strain by STE

Values are mean+SD. LBBB indicated left bundle branch block; AFG, active force generation; STE, Speckle Tracking Echocardiography; S, myocardial shortening velocity during ejection; TDI, Tissue Doppler Imaging. * P<0.05 vs. baseline.

Table 4. Electrical and mechanical timing in ischemic vs. non-ischemic segment during left anterior descending artery (LAD) occlusion

Tool for device optimization and lead placement for CRT

In further aspects, the invention provides combination of imaging modalities that display changes in dimension with left ventricular (LV) pressure or estimates of LV pressure to construct pressure-dimension or pressure segment loops for optimizing CRT device settings and lead placement site by a) identification of onset of regional and global LV mechanical activation and b) assessment of LV regional and global function. These aspects, the invention provides methods according to claims 14, 15 or 16, a computer program product according to claim 17 and a CRT device according to claim 18.

Background Cardiac resynchronisation therapy (CRT) has proven to be a promising treatment option for patients with heart failure and ventricular electrical conduction delay. However, in about 30 % of patients who are selected for CRT on basis of QRS criteria, there is no improvement in symptoms (non-responders) after implantation of the CRT device. Therefore a major challenge has been to implement new tools for evaluation of dyssynchrony, optimizing device settings and lead placements.

Currently LVOT flow by echocardiography is used to optimize interventricular delay (v-v delay) and assessment of E and A wave patterns are used for intraventricular delay (a-v delay) settings. These parameters assess global function and are unable to give information about regional function. The main objective of CRT is to recreate a physiological activation sequence to prevent remodeling and improve pump function. The global parameters mentioned above only gives us information about instantaneous response in pump function but leaves out important information about regional activation sequence that is essential to prevent/reverse remodeling and improve long term outcome.

Optimal lead placement may also be of critical importance to improve outcome of CRT. The aim when assessing lead placement site is to establish the LV region that is activated last. Currently many echocardiography indices exist to aid in this assessment, however, they have proved to give little added value.

In a further embodiment of the invention, pressure-dimension loops can be used for optimizing CRT device settings and lead placement site by a) identification of onset of regional and global LV mechanical activation using onset of active force, and b) assessment of LV regional and global function.

Applications

1. Marker for regional mechanical activation and function 2. Assessment of global and regional left ventricular function after CRT implantation to guide optimization of device settings.

3. Assessment of latest activated region for optimal lead placement.

Marker for regional mechanical activation As previously mentioned, it has been observed that current definition of onset of mechanical activation based on onset of shortening for a given myocardial segment often does not reflect true mechanical activation. A myocardial segment may demonstrate stretch even after it has been activated, as indicated in Figure 14 by a dashed circle in the strain trace of an experiment during ischemia.

We therefore propose that onset of active force generation (OAF) defined as the coordinate were a segment leaves its passive curve, reflects electro-mechanical coupling and thus mechanical activation. This is previously described in relation to Figure 5. Assessment of ventricular regional activation to guide device settings and lead placement.

By using onset OAF on several walls of the left ventricle one can define onset of mechanical activation in the different ventricular walls, this is previously described and shown in Figure 9. Echocardiographic parameters alone have not been able to differentiate between mechanical and electrical dyssynchrony within the ventricle. However, by combining strain/dimension measurements by novel imaging techniques with left ventricular pressure or pressure analogues we are able to differentiate between the two. In Figure 9A, dotted lines define timing of the first onset of R in EMG (R), onset AFG (O), onset of ejection velocity (Δ) and peak systolic shortening (D) in the segment length and velocity (dL/dt) traces. The actual onset for these parameters for each trace is also shown by the same symbols in each trace. It can be seen that during LBBB, OAF (O) in the lateral segment is delayed in relation to the first OAF in the septal segment (dotted line marked O). As OAF accurately reflects regional electrical activation, this allows for identification of latest activated region or segment.

Disparity in onset of activation will allow us to assess synchronicity of activation within the ventricle and will be of great value when optimizing the CRT device. This will allow the device operator to program the device in a fashion that most accurately reflects normal physiology by adjusting v-v and a-v delays on the CRT device. This will also allow us to select the latest activated region of the LV for placement of the LV CRT lead.

Additional fields of interest All present and future imaging modalities that give dynamic dimension/strain/volume measures (MR, CT; etc) can be combined with LVP (and existing and future LVP-analogues) and has the potential to give similar results.

Clinical application:

OAF is a potentially powerful tool for selecting patients for CRT, optimizing device settings and guiding lead placement.. Using pressure- volume-loops or pressure- global/regional strain/dimension-loops we are able to assess left ventricular function in a more precise manner that also is virtually operator independent.

Implementation example

A software application for preparing data related to determination of onset of active force in left ventricular muscle segments has been written in MatLab®. The software application is adapted to use the different algorithms for the determination of t₀AF presented previously.

On the following pages a brief presentation of the developed software will be presented.

Figure 15 is a flow-chart over the software architecture. The first part (1) is to start the software and load the necessary data;

1. A strain file containing 6 strain curves the ECG plot (low sample rate). 2. A file comprising the pressure curve and also high sample rate ECG data. 3. A picture file comprising e.g. ultrasound images of the heart.

The next step (2) is to, if needed, adjust the two ECG plots so that they are synchronized on the same time axis. At this stage a time interval for the heart beat is defined since the strain and the pressure files could comprise data from more than one heart beat. The last part at this stage is to define to (or onset of QRS) from which the AFG point can be defined. The used could then choose to either set the AFG point manually or automatically (3).

If needed the display properties could be changed (4) e.g. • the current time interval (in sec),

• the current cycle duration (in ms),

• the pulse (number of cycles per minutes),

• the Onset of QRS value (in ms),

• the currently selected OAF times (in ms). • the difference with the minimum OAF times (in ms).

It lets you also decide to show or hide the low-sample rate ECG plot, and whether or not a marker should be displayed for each samples of the "Strain/Pressure" curves.

The last part (5) is to either print a report or save that could later be re-opened.

Here follows some screen dumps as examples of how the software looks and works:

Figures 16A and B show an example from a patient with synchronous onset AFG. Figure 16A shows strain traces and P-L loops for late diastole for various myocardial segments. 16B shows a report displaying the time of OAF for each segment and the difference between early and later activated segments. Max delay between earliest and latest activated wall is 17ms (circle) indicating synchronous activation.

Figures 17A and B show the same patient and recordings as in Figure 16A, however now onset of shorting is used as activation point. Note that the lateral segment now is delayed by 35ms (circle) compared to early activated anterior- septal wall indicating dyssynchrony of mechanical shorting also seen on standard echocardiography. This is a patient that therefore has synchronous OAF but dyssychronous onset of shorting and will therefore not respond to pacing treatment, because the underlying mechanism of dyssynchrony is mechanical not electrical. References

Glantz S. A. Ventricular pressure-volume curve indices change with end-diastolic pressure. Circ Res 1976 December 1; 39(6) : 772-8

PRINZEN FW, AUGUSTIJN CH, ALLESSIE MA, ARTS T, DELHASS T, RENEMAN RS. The time sequence of electrical and active force during spontaneous beating and ectopic stimulation. Eur Heart J 1992 April 2; 13(4) :535-43.

Chung ES, Leon AR, Tavazzi L et al. Results of the Predictors of Response to CRT(PROSPECT) Trial. Circulation 2008 May 20; 117(20) :2608-16.

WO 06/104869

WO 07/022505

WO 03/037428

US 2008/0195167

Jon Mugaas, www.geocap.no

Anderson, LJ., Miyazaki, Sutherland, Oh. Circulation 2008; 117; 2009-23

Willems JL, De Geest H Kestelot H. On the Value of Apex Cardiography for Timing Intracardiac Event. The American Journal of CARDIOLOGY.1971 July 20:59-66

Ruffy R, Lovelace DE, Mueller TM, Knoebel SB, Zipes DP. Relationship between changes in left ventricular bipolar electrograms and regional myocardial blood flow during acute coronary artery occlusion in the dog. Circ Res. 1979;45:764-770.

Delhaas T, Arts T, Prinzen FW, Reneman RS. Regional electrical activation and mechanical function in the partially ischemic left ventricle of dogs. Am J Physiol. 1996;271 :H2411-H2420. Miura T, Bhargava V, Guth BD, Sunnerhagen KS, Miyazaki S, Indolfi C, Peterson KL. Increased afterload intensifies asynchronous wall motion and impairs ventricular relaxation. J Appl Physiol. 1993;75:389-396.

Claims

1. A method for preparing data related to onset of active force in left ventricular muscle segments, the method comprising

A. generating, by means of a computer, a parametric curve C,(t) = (L,(t), P(t)) from concurrent values of a left ventricular pressure, P(t) and a segment length, L,(t), in a left ventricular muscle segment, i, as a function of time;

B. determining an onset of active force in the left ventricular muscle segment, Ci(t₀AF,i), as the first point on a diastolic part of the curve where the parametric curve C_ι(t) deviates consistently from the segment's passive elongation during the late diastolic filling phase, and determining the corresponding time, t₀AF,ι;

2. A method according to claim 1 wherein the determined onset of active force is a suggested C,(t₀AF,ι), and wherein determining onset of active force comprises

- presenting the parametric curve C,(t) with a marked up suggested C(t₀AF,ι) to a user

- receiving user input related to an optional adjustment of the suggested C,(t₀AF,ι) and an approval of the suggested or adjusted C,(t₀AF,ι);

3 A method according to claim 2, wherein presenting the parametric curve C,(t) to a user comprises also presenting P(t) and L,(t) with a marked up t_0AF,ι corresponding to the suggested C,(t₀AF,ι)-

4. A method according to claim 1 wherein determining an onset of active force comprises determining Cι(t₀AF,ι) as the first deflection point that results in C,(t) leaving a region defined by the passive elastic curve PE(L) ± K.

5. A method according to claim 1 wherein determining an onset of active force dP{t) dPE(L) comprises determining Cι(t₀AF,ι) as the first point where >β over a dL£i) dL period of at least 30 ms.

6. A method according to claim 1 wherein determining an onset of active force comprises determining C,(t₀AF,ι) as the first point where:

≥ θ_τ and for which dP(t)/dt > 0 over the next 30 ms.

7. A method according to claim 1 wherein t_Cc is obtained as onset of QRS complex from ECG.

8. A computer program product for preparing data related to onset of active force in left ventricular muscle segments, the product comprising software applications which provides the following when executed by a computer: A. generating a parametric curve C,(t) = (L,(t), P(t)) of concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in a left ventricular muscle segment, i, as a function of time;

B. determining an onset of active force in the left ventricular muscle segment, Ci(t₀AF,i), as the first point on a diastolic part of the curve where the parametric curve C,(t) deviates consistently from the segment's passive elongation during the late diastolic filling phase , and determining the corresponding time, t₀AF,ι;

C. presenting a comparison between t₀AF,ι and a fixed time marker in the cardiac cycle, tec, and/or a time for onset of active force for another left ventricular muscle segment j, toAFj-

9. A computer program product for updating a medical monitoring apparatus to prepare data related to onset of active force in left ventricular muscle segments, the product comprising means for installing software applications which provides the following when executed by a computer: A. generating a parametric curve C,(t) = (L,(t), P(t)) of concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in a left ventricular muscle segment, i, as a function of time;

10. A medical monitoring apparatus comprising a unit for preparing and presenting data, the apparatus further comprising software means for:

B. determining an onset of active force in the left ventricular muscle segment, C,(t₀AF,ι), as the first point on a diastolic part of the curve where the parametric curve C_ι(t) deviates consistently from the segment's passive elongation during the late diastolic filling phase, and determining the corresponding time, t₀AF,ι;

11. A method for determining whether a patient has a primary electrical dyssynchrony or a primary mechanical dyssynchrony, comprising : - generating, by means of a computer, a parametric curve C,(t) = (L,(t), P(t)) of concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in two or more left ventricular muscle segments, i, as a function of time; - determining an onset of active force in the left ventricular muscle segments, Ci(t₀AF,i), as the first points on a diastolic part of the curve where each the parametric curve C,(t) deviates consistently from the segment's passive elongation during the late diastolic filling phase, and determining the corresponding times, toAF,.; - determining intersegment differences in activation times using the determined times for onset of active force, t₀AF,ι, and evaluating whether the patient has a primary electrical or a primary mechanical dyssynchrony.

12. A method for selecting patients for cardiac resynchronisation therapy (CRT), comprising

- evaluating whether the patient has a primary electrical or a primary mechanical dyssynchrony using the method according to claim 11;

- selecting patients with primary electrical dyssynchrony for CRT.

13. A method according to claim 12, wherein patients have previously been selected based on analysis of a QRS complex from an electrocardiogram.

14. The use of onset of active force in left ventricular muscle segment i, t₀AF,i, as determined from the method according to any of claims 1-7 as a marker for mechanical activation in left ventricular muscle segment i.

15. A method for adjusting settings of a cardiac Resynchronisation therapy (CRT) device after implementation, the method comprising - obtaining onset of active force (t₀AF) in ventricular muscle segments using the method according to any of claims 1-7 and determine relative mechanical activation times of these segments;

- adjusting interventricular (v-v) and/or atria-ventricular (a-v) delays on the CRT device using the determined relative mechanical activation times to optimize an activation sequence of ventricular segments.

16. A method for determining electrode placement of a CRT device

- obtaining onset of active force in left ventricular muscle segments using the method according to any of claims 1-7; - determining the latest activated segment of the left ventricle;

17. A computer program product for preparing data related to onset of active force in left ventricular muscle segments, the product being executed by hardware receiving concurrent values of a left ventricular pressure, P(t) and a length, L,(t), in a left ventricular muscle segment, i, for a patient as a function of time; the product comprising software applications which provides the following when executed by a computer: 1. generating a parametric curve C,(t) = (L,(t), P(t)) of received concurrent values of a left ventricular pressure, P(t), and a length, L,(t), in a left ventricular muscle segment, i, as a function of time;

2. determining an onset of active force in the left ventricular muscle segment, C,(t₀AF,ι), as the first point on a diastolic part of the curve where the parametric curve C,(t) consistently deviates from the segment's passive elongation during the late diastolic filling phase and determining the corresponding time, t₀AF,ι;

3. identifying t₀AF,i in the parametric curve as the onset of regional LV mechanical activation;

4. identifying global LV mechanical activation 5. assessing and presenting of LV regional and global function.

18. A cardiac Resynchronisation therapy (CRT) device wherein the interventricular (v-v) and atria-ventricular (a-v) delays have been adjusted using the method according to claim 15.