CA2676206A1 - Blood processing apparatus with optical reference control - Google Patents

Abstract

A density centrifuge blood processing system with automatic two-dimensional optical control of fluid separation by observing fluid characteristics in observation regions.
The location of the regions is determined by monitoring an optical reference.
Points representing edges of an optical reference are measured and lines are computed through the points. An error measurement is calculated for each line. If the error is too large, the image is abandoned. One of the lines is selected as a referent line. A new line is calculated orthogonal to the referent line. The error function is again computed for the dependant line.
If the error exceeds a selected maximum, the frame is discarded. A
transformation function translates data points from an (r, s) domain derived from measurements of the edges into an (x, y) domain used to identify pixels in the observation areas.

Description

BLOOD PROCESSING APPARATUS W['TH OPTICAL REFERENCE CONTROL
Field of the Invention The present invention relates to an apparatus and method for separating particles or components of a fluid. The invention has particular advantages in connection with separating blood components, such as platelets or white blood cells.

Description of the Related Art In the medical field, it is often necessary to separate blood into components.
Whole blood consists of vai-ious liquid conlponents and particle components. The liquid portion of blood is largely made up of plasma, and the particle components include red blood cells (erythrocytes), white blood cells (leukocytes), and platelets (thrombocytes).
Wliile tllese constituents have similar densities, their average density relationship, in order of decreasing density, is as follows: red blood cells, xhite blood cells, platelets, and plasma. In addition, the particle components are related accoi-ding to size, in order of decreasing size, as follows: white blood cells, red blood cells, and platelets.

Typically, donated platelets are separated or harvested from other blood coniponents using a centrifuge. Wliite cells or other selected components may also be harvested. The centrifuge rotates a blood separation vessel to separate components within the vessel or reservoii- using centrifugal force. In use, blood enters the separation vessel while it is rotating rapidly and centrifugal force stratifies the blood components, so that particular components may be separately removed. Components are removed through ports arranged within stratified layers of blood components.

White blood cells and platelets in plasma fonn a medium-density, stl-atified layer or "buffy coat". Because typical centrifuge collection processes are unable to consistently and satisfactorily separate white blood cells fronl platelets in the buffy coat, other processes have been added to InlprOVe reSU1tS. One Separatlon process is one 1Cnown as centrifugal elutriation. In one common fonn of elutriation, a cell batch is intl-oduced into a flow of liquid elutriation buffer, which carries the cell batch in suspension into a funnel-shaped chamber located on a spinning centrifuge. As additional liquid buffer solution flows through the chamber, the liquid sweeps smaller sized, slower-sedimenting cells toward an elutriation boundary within the chambel-, while larger, faster-sedimenting cells migrate to an area of the chamber having the greatest centi-ifugal fol-ce.

Wl1en the centi-ifugal force and force generated by the fluid flow are balanced, the io fluid flow is increased to force slower-sedimenting cells from an exit port in the chamber, while faster-sedimenting cells are retained in the chanlber. If fluid flow through the clianlber is increased, progressively larger, faster-sedinlenting cells may be removed fi-om the chamber.

The apparatus lias a fluid separation chamber having a first frustro-conical segment adjacent a fluid inlet and radially inwai-d therefi-om, a second frustro-conical seginent immediately adjacent the first fnlstro-conical segment and radially inward therefroni, the second frustro conical segment having a taper suc11 that particles within the second fi=ustro-conical segment are subjected to substantially equal and opposite centripetal and fluid flow 20 forces. The taper of the second frustro-conical segment is selected based on the expected size of particles, such that at least particles of the average size of expected particles will be subjected to substantially equal and opposite centripetal and fluid forces.
The apparatus has at least one punip controlling a rate of fluid flow through the fluid separation chaniber, a camera configured to observe fluid flow witll respect to the fluid separation chanlber, and a controller receiving signals from the camera and controlling the motor and the pump.

For these and other reasons, there is a need to improve control of particle separation and/or separation of components of a fluid.

Additional tecllnology related to this application is disclosed in, for example, U.S.
Patent No. 5,722,926, issued Mar. 3,1998; U.S. Patent No. 5,951,877, issued Sep. 14, 1999;
U.S. patent 6,053,856, issued April 25, 2000; U.S. patent 6,334,842, issued January 1, 2002; U.S. patent application Ser. No. 10/905,353, filed December 29, 2004;
U.S. patent application Ser. No. 111163,969, filed November 4, 2005 and in particular U.S.
patent 7,422,693, filed.luly 1, 2004.

Summary Of The Invention The present invention comprises a blood conlponent separation apparatus having a rotor foi- centT-ifugally separating, blood into phases such as i-ed blood cells, buffy coat, or plasma. A camei-a monitors a separation chaniber and image processing determines the location of boundaries. The apparatus controls the position of the boundaries by adjusting the speed of pumps or the rotor or both.

In the present invention, fluid flow in a blood separation chamber in a centrifugal separation device is selectively controlled by optical sensing of two regions in the separation chamber. Interface position inay be controlled by optical sensing of a two-dimensional view of the interface in the separation chamber in an area adjacent an outflow port or ports. Gross adjustments, that is, relatively large changes in the location of the interface or interfaces are best controlled by this observation of the interface. Tlius in transient states, such as the initial setup of flow conditions, interface position sensing can be effective. Fluid flow may also be controlled in response to the optical intensity (light or dark) of the fluid in the outflow tube. This optical intensity correlates to presence of certain blood components such as red blood cells. Fine adjustiiients, that is, relatively small changes in the location of the interface are best controlled by sensing the optical intensity in the outflow tube. Thus in steady state conditions, such as the extraction of a blood coniponent through the outflow tube, outflow intensity sensing is more effective.

In a high-speed centrifiige for sepai-ating blood components, imaging of the same locations for observation regions from rotation to rotation presents significant problenls, particularly in view of the vibrations with high-speed rotation. The present apparatus controls the interface location by measuring light intensity in the collect port monitoring region in the collect port by detecting the presenc.e or absence of RBC's in the collect port, and by monitoi-ing the interface in the phase boundai-y or interface monitoring region. In order for the apparatus to control the interface a reference position on the disposable blood processing bag, which is car-ried on the rotor, must be rapidly and reliably detenllined. In tllis invention, this is accomplished by a deteetion algorithm, which monitois an L-shaped calibration marker or optical reference. An intersection derived from edges of the optical reference is used as an ol-igin. A series of points representing an edge is measured. A set of data points, preferably about five (5), is collected for each edge, and a line is computed through the points. An error measui-ement is calculated foi- each line. If the error is too large, the image (an "observation" or "frame") for the current rotation is abandoned. The line with the least error is selected as a referent line. A new or dependant line is calculated for the line with the greater error. The error function is again computed for the dependant line. If the ei-ror exceeds a selected maximtun, the observation or frame is discarded.

Using the parameters of the lines a trausfonnation function is produced, wliich translates data points from an observation from an (r, s) co-ordinate domain derived from measurements of the edges into an (x, y) co-ordinate domain used to identify pixels in the observation areas. To test the transformation, the data points for the two edges are translated from the (r, s) donlain into the (x, y) domain and the error function is coniputed once again. If the error exceeds the maxinnun ei-ror limit, the fi-anle is abandoned.

If the data passes the tests, the pixels falling within the observation regions identified with reference to the origin that has been identified as the intersection of the referent and dependant lines ai-e used deter-iliine the position of phase boundai-ies and out flow characteristics. The process outlined above and described more completely hereafter allows for a frame by frame detennination of the location of an origin in the pixel field of the camera and for a detel-nlination that the image is sufficiently clear for the collection of data. Vibration and relative motion between the rotor and separation cllaniber and the caniei-a causes the image detected by the camera to move in the (x, y) plane and to come in and out of focus. The method described allows the apparatus to discard a frame or observation that is too blurry to provide accLirate data and to locate a consistent origin from frame to frame.

It is an object of the present invention to pi-ovide a density centrifuge blood processing system for sepal-ating fluid components comprising a separation chamber rotating about a centi-al rotation axis, the sepai-ation chamber having an optical reference mounted thereon, and a computational apparatus distinguishing the optical reference and establislling refei-ence co-ordinates for gathering data from at least one observation of the observation region detected by the first detector.

It is also an object of the invention that conlputational apparatus rejects an observation of the separation cllamber if an en-oi- measurenient of the optical reference is not within a pi-e-selected error.

A fiu-ther object of the invention is to translate data from the observation of the observation region fronl a first co-ordinate domain into a second co-ordinate donlain. Another object is to provide an optical reference comprising at least two non-parallel sides and wberein said computational apparatus recognizes a first side represented by a first line and fiirther fits a second line to a second side according to a known angle between said first side and said second side and to reject an observation of the separation chamber if data representing an edge of the optical reference are not within a pre-selected error.

It is also an object of the invention to compute an en-or nleasurenient for each of the first and second sides and to select the side with the least error as a referent line, to compute a dependant llne for the side with the greater error, to re-collipLlte all error nleasurenlent for the dependant line, and to reject an observation of said separation cllanlber if the et-i-or nleasurenlent for the dependant line is not witllin a pre-selected en-or.

These and other objects and features of the present invention will be apparent fronl the following detailed desci-iption. It is to be understood that botll the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the invention as claimed.

Brief Description Of The Drawings FIG. I is a partial perspective, schematie view of a blood processing centrifuge apparatus inetuding a fluid chamber.

FIG. 2 is a partial perspective, scllematic view of the centrifuge apparatus and a control cainera. FIG. 3 is a partial cross-sectional view of blood processing apparatus of Fig. 2, including the fluid chamber of FIG. 1.

FIG. 4 is a partial cross-sectional, schematic view of a portion of a separation vessel and the fluid cllanlber inounted on a centrifuge rotor of FIG. 1.

FIG. 5 is a plan view of a separation chanlber of the separation vessel of Figure 4.
FIG. 6 is a graphic representation of steps for image processing according to the present invention.

FIG. 7 is a diagram sllowing the relationship of an (r, s) co-ordinate system and an (x, y) co-ordinate systenl.

FIG. 8 is a grapll of signals for recognizing data points representing an edge of an optical i-eference.

Detailed Description The present invention preferably comprises a blood processing apparatus having a camera control system, as disclosed in US Patent 7,422,693 and in US
applications 10/905,353 and 11/772,692 and 11/774,073. It may also be practiced wit11 a TRIMA
blood component centrifuge nlanufactured by CaridianBCT, Inc. of Colorado (fonnet-ly Ganibro BCT, Inc.) or, alternatively, with a COBE R SPECTRA single-stage blood component centrifuge also manufactured by CaridianBCT, Inc. Both the TRIMA OO
and the SPECTRA centrifuges incorporate a one-ornega/two-omega sealless tubing connection as disclosed in U.S. Patent No. 4,425,112 to Ito. The SPECTRA centrifuge also uses a single-stage blood component separation cliannel substantially as disclosed in U.S.
Patent No.
4,094,461 to Kellogg et al. and U.S. Patent No. 4,647,279 to Mulzet et al. The invention could also be practiced with a TRIMA R or TRIMA ACCELit centrifugal separation system or other types of centrifugal separator. The method of the invention is described in connection with the aforementioned blood processing apparatus and camera control system for purposes of discussion only, and this is not intended to limit the invention in any sense.
As embodied herein and illustrated in FIG. 1, a centrifuge apparatus 10 has a centrifuge rotor 12 coupled to a motor 14 so that the centrifuge rotor 12 rotates about its axis of rotation A-A. The motor 14 is coupled to the rotor 12 directly or indirectly through a shaft 24 connected to the rotor 12. Altenlately, the shaft 24 may be coupled to the motor 14 through a gearing transmission (not sllown).

The rotor 12 has a retainer 16 including a passageway or annular groove 18 Ilaving an open upper surface adapted to receive a separation vessel 28, shown in pertinent part in FIG. 4. The groove 18 coinpletely surrounds the rotor's axis of rotation A-A
and is bounded by an inne-- wall 20 and an outer wall 22 spaced apart from one another to define the groove ] 8. Although the groove 18 shown in FIG. I completely surrounds the axis of i-otation A-A, the groove could partially suiround the axis A-A if the separation vessel is not anlullar.
Preferably, a substantial portion of the groove 18 has a constant radius of curvature about the axis of rotation A-A and is positioiied at a maxinlum possible radial distance on the rotor 12. This sliape ensui-es that substances separated in the separation vessel 28 undergo relatively constant ceutrifitgal fol-ces as they pass from an inlet portion to an outlet portion of the separation vessel 28.

As sllown in FIG. 1, a bracket 26 is provided on a top surface of the rotor 12. The bracket 26 releasably holds a fluid chaniber 30 on the rotor 12 so that an outlet 32 of the fluid chamber 30 is positioned closer to the axis of rotation A-A than an inlet 34 of the fluid chamber 30. The bi-acket 26 preferably oi-ients the fluid chamber 30 on the rotor 12 Nvith a longitudinal axis of the fluid clianiber 30 in a plane transverse to the rotor's axis of rotation A-A. In addition, the bracket 26 is preferably an-anged to hold the fluid chamber 30 on the rotor 12 with the fluid chamber outlet 32 facing the axis of --otation A-A.
Although the fluid chamber 30 is shown on a top surface of the rotoi- 12, the fluid chamber 30 could also be secured to the rotor 12 at altei-nate locations, such as beneath the top surface of the rotor 12.

Figure 2 schematically illustrates an exemplary embodiment of an optical monitoring system 40 capable of nleasuring a distribution of scattered and/or transmitted light intensities corresponding to patterns of light originating fi-om an observation region on the separation vessel 28. The monitoring system 40 comprises light source 42, light collection elenlent 44, and detector 46. Light source 42 is in optical comniunication with the centrifuge apparatus 10 comprising rotor 12, which rotates about central rotation axis A-A. Rotation about central rotation axis A-A results in separation of a blood sample in the separation vessel 28 into discrete blood coniponents.

Light solu-ce 42 provides incident light beam 54, which stroboscopically illuminates an observation region 58 when the observation region 58 passes under the light collection element 44. Light soLu-ce 42 is capable of generating an incident light beam, a poi-tion of which is ti-ansmitted through at least one blood component undergoing separation in separation vessel 28. At least a portion of scattered and/or transmitted light 56 from the observation region 58 is collected by light collection element 44.
Light eollection element 44 is capable of directing at least a portion of the collected light 56 onto detector 46. The detector 46 detects patterns of scattered and/oi-transmitted light 56 from the observation region. The observation region 58 may also be illuminated by an upper light source 62, which is positioned on the same side of the separation chanlber as the liglit collection element 44 and detector 46. Upper light soiu-ce 62 is positioned such that it generates an incident beam 64, whicli is scattered by the blood sample and/or centrifuge. A portion of the light from upper light source 62 is collected by light collection elenlent 44 and detected by detector 46, thereby measLu-ing a distribution of scattered and/or ti-ansmitted light intensities.

Distributions of scattered and/or transniitted light intensities comprise imaQes coiTesponding to patterns of light originating fi-onl the observation region 58. The images may be monochrome images, which provide a measurement of the briglltness of separated blood components along the separation axis. Alternatively, the images may be color images, which provide a measurement of the colors of separated blood components along the separation axis. Observation region 58 is positioned on a portion of the density centrifi.ige 10, preferably on the separation vessel 28. The fluid chamber 30 may also be an observation region, as explained below. In the exemplary enlbodiment illustrated in Figure 4 and Figure 5, separated blood components and phase boundaries between optically differentiable blood components are viewable in observation region 58.

Detector 46 is also capable of generating output signals corresponding to the measured distributions of scattered and/or transmitted light intensities and/or images. The detector 46 is operationally connected to a device controller 60 capable of receiving the output signals. Device controller 60 displays the measured intensity distributions, stores the measured intensity distributions, processes nleasured intensity distributions in real time, transmits control si(ynals to various optical and mechanical components of the nionitoring system and centrifuge or any conlbination of tliese. Device conti-oller 60 is operationally connected to centrifuge apparatus 10 and is capable of adjusting selected operating conditions of the centrifuge apparatus, such as the flow rates of cellular and non-cellulai- components out of the separation vessel 28 or fluid chamber 30, the position of one or moi-e phase boundai-ies, rotational velocity of the rotor about central rotation axis A-A, the infusion of anticoagulation agents or other blood processing agents to the blood sample, or any combi.nation of these.

Device controller 60 can also be operationally connected to light soLu-ce 42 and/or upper light source 62. Device controller 60 and/or detector 46 are capable of generating output signals for controlling illumination conditions. For example, output signals from the detector 46 can be used to control the timing of illumination pulses, illumination intensities, the distribution of illumination wavelengths and/or position of light source 42 and/or upper light sotirce 62. Device controller 60 and detector 46 are in two-way commtmication, and the device controller sends control signals to detector 46 to selectively adjust detector exposure time, detector gain and to switch between monochrome and color imaging.

Light sources comprise liglit emitting diode (LED) sources capable of generating one or more incident beams for illwninating an observation region on the centrifiige. A
plurality of lamps may be positioned to illuminate a single side or niultiple sides of the centrifuge apparatus 10. Light emitting diodes and arrays of light emitting diode light sources are preferred for some applications because they are capable of generating precisely timed illumination pulses. Prefer-red ligbt soLu-ces generate an incident light beam having a substantially unifonn intensity, and a selected wavelength range.

The optical monitoring systenl comprises a plui-ality of light sources, each capable of generating an incident light beam liaving a different wavelength range, for exanlple, a combination of any of the following: white light source, red light source, green light soui-ce, blue light source and iilfi-a red light sotn-ce. Use of a combination of light sources having different ~vavelength ranges is beneficial foi- discrinlinating and charactei-izing separated blood fi-actions because absoi-ption constants and scattering coefficients of cellular and non-cellular coniponents of blood vary witll wavelength. For example, a conlponent containing red blood cells is easily distinguished from platelet-en--iched plasma by ill1,nnination with light having wavelengths selected over the range of about 500 nm to about 600 nm, because the red blood cell component absorbs light over this wavelengtli significantly mo1-e strongly that the platelet-enriched plasma component. In addition, use of multiple colored light sources provides a means of charactei-izing the white blood cell type in an exti-acted blood component. As different white blood cell types have different absorption and scattering cross sections at different wavelengths, monitoring transmitted and/or scattered light fi-om a white cell-containing blood coniponent provides a means of distinguishing the various white blood cell types in a blood conlponent and quantifying the abundance of each cell-type.

The light sources provide a continuous incident light beam or a pulsed incident light beam. Pulsed light sources are switclied on and off synchronously with the rotation of the t-otor 12 to illtu71inate an observation region having a substantially fixed position on the rotor 12. Alteniatively, pulsed light sources of the present invention can be configured such that they can be switched on and off at different angular positions, synchronous with the rotation of the rotor 12, illuminating different observation regions for eacli full rotation. This altenlative embodiment provides a method of selectively adjusting the location of the observation region and, thereby, probing different regions of the separation cllaniber 28 or of the fluid chamber 30. Triggering of ilhuiiination pulses may be based on the rotational speed of the centrifuge or on the angular position of the separation chamber or the fluid chamber 30 as detected by optical or electronic methods well known in the art. Triggering may be provided by trigger pulses generated by the device controller 60 and/or detector 46.

Figure 3 is a cutaway view of the optical monitoring systein 40. The illustrated optical monitoring system 40 comprises CCD ("cliarge-coupled device") camei-a (CMOS ("complementai-y metal oxide semiconductor") or otlier camei-as could also be used) eduipped with a fixed focus lens system (corresponding to the light collection element 44 and detector 46), an optical cell 74 (corresponding to the observation region 58), an upper LED light source 76 (cori-esponding to the upper light soui-ce 62), and a bottoin pulsed LED light source 78 (con-esponding to the ligllt source 42). As illustrated in Figure 3, CCD cainera 72 is in optical communication with optical cell 74 and positioned to intersect optical axis 80. Upper LED light source 76 is in optical connnunication with optical cell 74 and is positioned such that it is capable of dit-ecting a plul-ality of colliinated upper light beams 82, propagating along propagation axes that intersect optical axis 80, onto the top side 84 of optical cell 74. Bottom pulsed LED light source 78 is also in optical commL111ication with optical cell 74 and is positioned such that it is capable of directing a plurality of colliniated bottom light beams 86, propagating along optical axis 80, onto the bottom side 88 of optical cell 74.

CCD camera 72 may be positioned such that the focal plane of the fixed focus lens systein is substantially co-planar with selected optical surfaces of optical cell 74, such as optical surfaces con-esponding to an interface monitoring region, calibration inai-kers, one or more extraction ports and one or more inlets. The CCD camera 72 is separated from the center of the fixed focus lens system by a distance along optical axis 80 such that an image corresponding to selected optical surfaces of optical cell 74 is provided on the sensing surface of the CCD camera. This optical configuration allows distributions of light intensities comprising images of rotating optical cell 74 or of fluid chan-ibei- 30 to be measured and analyzed in real tinie.

Referring to Figure 3, first transparent plate 96 is provided between CCD
camera 72 and optical cell 74, and second transparent plate 98 is provided between bottoni LED
light source 78 and optical cell 74. First and second transparent plates 96 and 98 physically isolate CCD camera 72, upper LED light source 76 and bottonl LED
light source 78 froin optical cell 74 so that these components will not contact a sample undergoing processing in the event of sample leakage from the sepai-ation cllamber 28. In addition, first and second transparent plates 96 and 98 mininlize degradation of CCD
camera 72, upper LED light soLU-ce 76 and bottom LED light source 78 due to unwanted deposition of dust and otller contaminants that can be introduced to the systeni upon rotation of the separation chamber and filler. Furtlier, first and second transparent plates 96 and 98 also allow a user to optimize the alignment of the calnera 72, upper LED light source 76 and bottom LED light source 78 witllout exposure to a blood sainple in the separation cllamber 28. Fii-st and second transparent plates 96 and 98 can coinprise any niaterial capable of transmitting at least a portion of upper and bottom illumination light beams 82 and 86. Exemplary materials for first and second transparent plates 96 and 98 include, but ai-e not limited to, glasses such as optical-quality, scratch-resistant glass, transparent polymeric niaterials such as ti-anspal-ent plastics, quartz, or inorganic salts.

FIG. 4 schematically illusti-ates the sepai-ation vessel 28 and fluid chaniber mounted on the rotor 12. The separation vessel 28 has a generally annular flow path 100 and includes an inlet portion 102 and outlet portion 104.

A radial outer wall 108 of the separation vessel 28 is positioned closer to the axis of rotation in the inlet por-tion 102 than in the outlet portion 104. During separation of blood components, this arrangement causes fonnation of a very thin and rapidly advancing red blood cell bed in the separation vessel 28 between the inlet portion 102 and outlet portion 104. The 1-ed blood cell bed substantially limits or prevents platelets from contacting the radial outer wall 108 of the separation vessel 28. This is believed to reduce clumping of platelets caused wlien platelets contact structural components of centrifugal separation devices.

The inlet portion 102 includes an inflow tube 110 for conveying a fluid to be separated, such as whole blood, into the separation vessel 28. During a separation procedure, substances entering the inlet portion 102 follow the flow path 100 and stratify according to differences in density in response to rotation of the rotor 12.
The outlet poi-tion 104 includes first, second, and thii-d outlet lines 112, 114, 116 for removing separated substances fi-om the separation vessel 28. Preferably, each of the components separated in the vessel 28 is collected and removed in only one area of the vessel 28, namely the outlet portion 104. In addition, the sepai-ation vessel 28 preferably includes a substantially constant i-adius except in the region of the outlet portion 104 where the outer wall of the outlet poi-tion 104 is pt-eferably positioned farthei- away from the axis of rotation to allow for outlet poi-ts of the lines 112, 114, and 116 to be positioned at different radial distances and to create a collection pool with greater depth for the high density red blood cells. The outlet port of line 114 is farther from the axis of rotation A-A
than the other ports to i-emove higller density components, such as red blood cells. The port of line 116 is located closer to the axis of rotation than the otlier ports to remove the least dense components separated in the separation vessel 28, such as plasma. The first line 112 collects intennediate density components and, optionally, some of the lower density components. The first line 112 may be coupled to the inlet 34 of the elutriation chamber 30. The outlet 32 of the elutriation chamber 30 is coupled to a line 130. The second and third ]ines 114 and 116 are positioned downstream (not shown) fronl first line 112 to collect the high and low density components.

The positions of the interfaces ai-e controlled by the CCD camera 72 monitoring the position of the interface and controlling flow of liquid and/or particles in response to the monitored position. Further details concenling the stn.icture and operation of the separation vessel 28 are described in US Patent 7,422,693 and also in US Patent No.
4,094,461 to Kellogg et al. and US Patent No. 4,647,279 to Mulzet et al.

RefeiTing to Figure 2, the camera 44 is generally focused on the separation vessel 28 and stroboscopic illumination allows an observation region 58 around the first, second, and third lines 112, 114, and 116 to be observed. Using information gatllered tlirough the camera, the controller 60 regulates the position of interfaces between various blood components, such as plasma, buffy coat (containing monocytes and/or white blood cells and platelets) and 1-ed blood cells by controlling the pumps (not sllown) connected to lines 110, 114, 116, and 130. Figure 5 sliows an image of the observation region 58 generated by the nlethods of US patent 7,422,693 corresponding to the separation of a hun-ian blood sample and exti-action of a separated platelet-containing blood component. The observation region 58 shown in Figure 5 includes a phase boLuidary monitoring region 202 and an extraction or collect poi-t monitoring region 204. Visible in phase boundary monitoi-ing region 202 are a red blood cell component 206, a plasnla component 208 and a mixed-phase buffy coat layer 210, which has both white blood cells and platelets.
Calibration nlarkers are also apparent in the image in Figure 5. Near an edge 212 of the optical cell 74 is an L-shaped calibration marker or optical reference 214 for detenllining the absolute position of phase boundaries between optically differentiable blood components. Inner edges 232, 234 of the optical reference 214 are used to indicate the positions and physical dinlensions of the pllase boundary monitoring region 202 and the platelet collect poi-t monitoring regi.on 204. The pllysical dimension may be detennined by adjusting the optics to witllin a selected range and then configtu-ing the software with a paranleter to convert pixels to microns. Alternatively, the tllickness of the optical reference, usually about 1 mni, could be used. Liglit intensities transmitted through the phase boundary monitoring region 202 are acquired as a function of time and analyzed in real time to provide measurenlents of the position of the phase boundary 216 between red blood cell component 206 and buffy coat layer 210 and the phase boundary 218 between the buffy coat layer 210 and plasma component 208. All boundary layer positions are measured relative to the edge of the optical reference 214.

Collect port inonitoring region 204 monitol-s flow in first line 112 of the optical cell for extracting a blood component, for exanzple, buffy coat. The apparatus responds to changes in detected blood coniponent flow to establish a correct phase boundary level and further responds to changes in observed pllase boundaries to maintain a consistent phase boundary level. The system discriminates between a plasina flow condition, a buffy coat flow condition, and a red blood cell flow condition, and can detect pump-induced flow variation in the blood component flow in the collect port measuring area. A
plasma signal limit and a red blood cell signal limit may be set and the flow of fluid adjusted based on said linlits. The system derives a statistical measLu-e of fluid f7ow in the collect port measru=ing area, which may be a moving median of the avei-age value of intensity of pixels in the collect port measuring area.

In tliis exaniple, first line 112 having oi-ifice 224 is configtu-ed to collect buffy coat in the ]nulian blood sample and extends a distance along the separation axis such that it terminates proximate to the buffy coat layer in the rotating separation chamber. The two-diniensional distribution of light intensities of light ti-ansmitted through the collect port in the collect port monitoi-ing region 204 depends oii the concentration, and spatial distribution and cell-type of cellular material exiting the separation charnber. Light intensities transniitted through the collect port nlonitoring region 204 are acquired as a function of time and analyzed to charactei-ize the coinposition and flux of cellular material out of the separation chamber 28. As cellular niaterials, sucli as Nvhite blood cells and red blood cells, absorb and scatter light fi-on1 the light sources, passage of cellular material through the extraction port decreases the observed light intensities. The first collection line 112 is counected to the fluid chamber inlet 34 to pass the intennediate density coniponents into the fluid chamber 30. Components initially separated in the separation vessel 28 are further separated in the fluid chamber 30. For example, white blood cells could be separated from plasma and platelets in the fluid chamber 30.
This further separation prefei-ably takes place by forming a saturated fluidized bed of particles in the fluid clianiber 30. The fluid cllamber 30 may be formed of a transparent or translucent co-polyester plastic, such as PETG, to allow viewing of the contents within the chamber intei-ior with the aid of the caulera during a separation procedure.

The apparatus 10 includes the controller 60 (Fig. 1) connected to the motor 14 to control rotational speed of the rotor 12. The controller 60 is connected to the punips in lines 110, 114, 116, and 130 to control the flow rate of substances flowing to and from the separation vessel 28 and the fluid chamber 30. The controller 60 controls the operation and flow rate of the pumps to permit the tempoi-ary purging of the fluid cllamber 30. The controller 60 may include a computer having prograrnmed instructions provided by a ROM
or RAM as is commonly known in the art. The controller 60 may vary the rotational speed of the cenh-ifuge rotor 12 by regulating frequency, current, or voltage of the electricity applied to the motor 14. Alternatively, the rotational speed can be varied by shifting the at-rangement of a ti-ansmission (not shown), such as by changing gearing to alter a rotational coupling between the motor 14 and rotor 12. The controller 60 may receive input from a rotational speed detector (not sliown) to constantly monitor the i-otation speed of the rotor.

Accumulated buffy coat components, comprising platelets, some white blood cells, and plasma, are 1-emoved via the first collection line 112. As the platelets, plasma, white blood cells, and possibly a small nwliber of red blood cells pass through the first collection line 112, these coniponents flow iilto the fluid chamber 30, filled with the priming fluid, so that a saturated fluidized particle bed may be fonned. The platelets flow toward the first collection line 112. The priming fluid along the inner walls of the separation vessel 28 reduces the effective passageway volunie and area in the separation vessel 28 and thereby decreases the anlount of blood initially required to prime the system in a separation process.
The reduced volume and area also induces higher plasnia and platelet velocities next to the stratified layer of red blood cells, in particular, to "scrub" platelets toward the first collection line 112. The rapid conveyance of platelets increases the efficiency of collection.
The fluid chamber 30 is configured to allow cyclic collection of selected particles, such as platelets, followed by efficient evacuation of the cells into a collection bag. In contrast to otller chamber designs for forming saturated fluidized beds, the fluid chamber described herein has particular application for the autoniated collection of blood components in that a bolus of cells having selected characteristics can be collected in the fluid chamber 30 and then flushed with low density fluid into a collection bag and these steps can be repeated multiple times, allowing a larger quantity of the selected cells to be collected from the donor or patient while reducing the amotuit of time necessary for the donatioil process. Collection of cells in the fluid chamber can be monitored by the caniera 72 and the device controller 60. When a selected quantity of cells have been collected in the fluid chamber 30, the flow of plasma through the chamber can be inereased and the collected cells can be washed out of the chamber and directed into a collection bag.
Fi-ame-by-Frame OriQin Calibration In a high-speed centrifuge for sepai-ating blood components, control of the interface between blood components presents significant control problems_ The present apparatus controls the interface location by measuring light intensity in the collect port monitoring region 204 in the collect port by detecting the presence or absence of RBC's in the collect port, and by nlonitoring the interface 216 or 218 in the pllase boLuldary or interface monitoring region 202. The ligllt intensity in the collect port can be nieasured by both an average value over a relatively brief period of time or by a median value over a longer period of time or by a combination of both measui-ements. The location of the interface is detected by a series of image processing steps, which allow the apparatus to recognize a boundary or interface despite limitations such as the high speed of the centrifuge rotor, the char-acteristics of sti-oboscopie light used for observation, or the limits of data processing time. Monitoring the interface in the interface monitoring region 202 allows the apparatus to determine and control the location of the interface reliably. In order for the apparatus to control the interface, a t-eference position on the disposable blood processing bag, which is can-ied on the rotor, must be rapidly and reliably determined. In this invention, this is accomplished by a detection algorithm 230, which monitors the L-shaped calibration marker or optical reference 214. The detection algorithm 230 is shown in Figures 6a through 6f.

An optically controlled centrifuge for blood separation, as described herein, presents certain problems for the control of the apparatus. With a camera mounted on the frame and observing indistinct phenomena on a rotor spinning in excess of 3000 rpm, vibration is a persistent problem. As viewed tllrough the camera, the image of the separation cllamber shakes. Moreover, distances between the camera and observation areas cannot be controlled to the i-equisite tolerances for observation purposes. This is particularly true l8 where, as herein, the blood is to be pi-ocessed in a disposable blood separation chamber.
Consequently, the location of observation regions niust be determined dynamically. In the present invention, an optical i-eference having at least two non-parallel edges is mounted iiear the separation chamber. An intersection dei-ived from edges is determined for use as an origin. Observation regions are established with respect to the o1-igin and pixels falling within the observation regions are used to detect phase boundaries and outflow conditions for controlling the blood processing apparatus, as moi-e fully described in US
application 11/772,692 and US application 11/774,073. Because of vibi-ation and relative motion between the camera and the separation chamber, it is unlikely that the same pixels in the camera will image the observation regions from one rotation to the next.
Nevertheless, by selecting pixels witll reference to a dynamically detennined origin, a stable view of the observation regions can be obtained.

In the present invention, a series of points representing an edge is measured by collecting raw intensity data fi-om pixels in the cainei-a imaging a region that crosses the edge, filtering the intensity data to reduce variation, and differences between adjacent pixels. There is an abi-upt change fi-on1 light to dark at the edge, wllich is detected by a difference minima. A set of data points, preferably about five (5), is collected for each edge, and a line is coinputed thi-ough the points. An error measurement, for example, the root mean square error, is calculated for each line. If the ei-ror is too large, the image (or "frame") for the current rotation is abandoned. The data is deenied too imprecise or noisy.
A new frame would be available in abont 40 microseconds, and the process can begin again. The line with the least error is selected as a referent line. A new or dependant line is calculated for the line with the greater en-or. The dependant line is a mathematical constr-uct created at the known angle between the two edges. This angle may be any acute or obtuse angle less tlian 180 degrees. The preferred angle is 90 degrees.

The intersection of the two edges is usually pliysically chamfered. To provide a precise intersection of the referent and dependant lines, they are preferably calculated at an offset froni their constituent data points. The intersection of the calculated lines will not fall in the chamfer area and data from the chamfered area will be excluded f--oni calculations. The erroi- fLuiction is again computed for the dependant line.
If the ei-roi-exceeds a selected maximum, the fi-ame is discai-ded, as described above.

Using the parameters of the lines a transformation fiuietion is produced, which translates data points from an (r, s) co-ordinate doniain derived from measurements of the edges into an (x, y) co-ordinate domain used to identify pixels in the observation areas. To test the transformation, the data points for the two edges are translated from the (r, s) domain into the (x, y) domain and the ei-i-or fi.inction is coinputed once again. If the error 1o exceeds and maximum en-or liinit, the frame is abandoned.

If the data passes the tests, the pixels falling within the observation regions identified with reference to the origin that has been identified as the intersection of the referent and dependant lines are used deter-mine the position of phase boLmdaries and out flow characteristics. The process outlined above and described more completely hereafter allows for a frame by frame determination of the location of an origin in the pixel field of the cainera and for a determination that the image is sufficiently clear for the collection of data. Vibration and relative nlotion between the rotor and separation cllamber and the camera causes the iinage detected by the camera to inove in the (x, y) plane and to come in 20 and out of focus. The method described allows the apparatus to discard a frame that is too blurry to provide accurate data and to locate a consistent origin from frame to frame.

As shown conceptually in FIG. 7, the (r, s) co-ordinate doniain is an orthogonal planar reference system in two c.o-ordinates. Herein, the (r, s) domain is associated with the machined, rigllt-angle reference block 214 shown in FIG. 5. The (x, y) co-ordinate domain is also an oi-thogonal planar reference system in two co-ordinates. The (x, y) is associated with the phase boundary nlonitoring region 202 or the collect port monitoring region 204 or both. The origin of the (x, y) domain is usually offset from the origin of the (r, s) donlain and the two axis of the two domains are not necessarily parallel to each otller. The present 30 iiivention develops a transformation of infonnation fronl one doulain to the other. The transformation takes the followiug fornl, which specifies the convei-sion between S,R
(pixel) coordinates and X,Y (engineering units) coordinates. The matrix is defined as:
R1=T[1 1 S
R =T-' Y

P, cosrl PR sina -P,Ds cosu-PkDR sina T= - Ps sin a P,z cos u PS. D.,;sin a- P,z Dk cos a COS Cl Sln a -DS
Ps 1's T sinu cOs(, D
n Pk Pk Where Ps is the pixel size (in nlicrons) along the S axis, PR is tlle pixel size (in inicrons) along the R axis, a is the angular rotation between the S,R and X,Y coordinate systetns, D. is the S position of the reference conler, and DR is the R position of the refei-ence corner.

The optical reference 214 llas a llorizontal edge 232 and a vei-tical edge 234.
Pi-eferably, these edges 232, 234 intersect at a right angle, although the angle might also be acute or obtuse. The detection algorithnl begins 236 by identifying an r Axis Line representative of the horizontal edge 232. An index i is initialized 238. Data points i-epresenting detected positions along the edge are added 240 and the index is incremented 242 until a selected nunlber of data points are collected 244. Data points repi-esent transitions from dark to light along the horizontal edge, as detetnlined by the canlera. Tlle pi-ocess of selecting data points is represented graphically in FIG. 8. Raw intensity data 252 acquired froni pixels in the field of view of the camera along a line generally pai-allel to the s axis (that is, radially) for detecting points on the horizontal edge 232 nlay be filtered 254 to i-educe noise i-elated variations. Taking the difference 256 between adjacent pixels, inoving from liglit 258 to da--k 260, a minimum (absolute value maximLun) difference 262 is found at the pixels closest to the edge 232. A horizontal sweep would be used to locate points on the vel-tical edge 234. If a satisfactory set of data points is collected, a representative line is fitted 264 through the data points. If the data points are sufficiently close to the line, for example if the niean square fit of the data to the line is less tlian a predeteniiined limit, the r Axis Line fit critei-ia is satisfied 266, and the program can continue. Otherwise, the program is interrupted 268. An interruption in the prog--an1 implies that the apparatus will attempt to acquire a satisfactoi-y in-iage on the subsequent rotation of the centrifuge.

The program then calculates selected cllai-acter-istics of the r line. The slope of the line is computed 270. An offset 272 is added to the data. This essentially moves the line representing the llorizontal edge 232 upward so that the line does not intersect a chamfer at the junction of the two edges 232, 234. Data points for the vertical edge will be selected in a region above the offset line. Finally, an en-or co-effici.ent for the data with respect to the r Axis line is computed. This co-efficient may be, for example, the root mean squared error of the data with respect to the line.

The detection algorithm next identifies an s Axis Line representative of the vertical edge 234. An index i is initialized 276. Data points representing detected positions along the edge are added 278 and the index is increnlented 280 until a selected number of data points are collected 282. Data points represent transitions from dai-k to light along the vertical edge, as described above. If a satisfactory set of data points is collected, a representative line is fitted 284 through the data points. If the data points are sufficiently close to the line, for example if the niean square fit of the data to the line is less than a predetermined limit, the s Axis Line fit criteria is satisfied 2S6, and the program can continue. Otherwise, the program is inte--rupted 288.

The program then calculates the selected charactei-istics of the s Axis line.
The slope of the line is computed 290. An offset 292 is added to the data. As before, this moves the line representing the vei-tical edge 234 sideways so that the line does not intersect the chamfer at the junction of the two edges 232, 234. Finally, an e--ror co-efficient for the data with respect to the s Axis line is computed. This co-efficient may be, for exainple, the i-oot mean squai-ed ei-i-or of the data with respect to the s Axis line.

The next nlajor feature of the pi-ogram selects the line having the nlost consistent data, that is data that is most linear, and then re-calculates the other line at a right angle (or other angle depending on the optical reference 214) to the selected line. An offset for the calculated line is selected to miniinize ei-i-or of the data points with respect to the calculated line. The error is computed, and, if the erroi- exceeds a pi-e-determined limit, the program is interrupted in favor of the next fi-anie, as explained above. Tl1is feature comprises steps 296 through 360.

In step 296, the program compares the error for the s Axis line to the ei-ror for the r Axis line. If the er-ror for the s Axis line is smallest, a line will be calculated for the horizontal or r Axis, representing the liorizontal edge 232. The program will calculate an offset for the point wllere the calculated r Axis will cross the s Axis according to the following formula:

~is;-~1r;m a=

where a is the offset, N is the number of data points in the selected line, s;
and i; are data points along the vertical and horizontal edges 234, 232 respectively, and ni is the slope of the horizontal or r Axis line, set, in this example, to -90 degrees fronl the s Axis line. To calculate the offset a, an index i is set 298 to zero. Position sruns (Es; and Yjr;) for the s Axis 300 and for the r Axis 302 ai-e accumulated for the number of data point along the s Axis and the index i is incremented 304 until the iuimbei- of points along the s Axis has been reached 306. The i-espective position sums are multiplied by 2 and divided by the nuuiiber of data of data points along the r Axis 308 and the s Axis 310, respectively. The r Slope is set 312 to -90 degrees from the s Axis line. This slope is forced to match the known included angle in the optical reference 214, in this case, 90 degrees.
Of course, other angles could be used. The offset for the calculated r or horizontal line is calculated 314 as lialf of the accumulated s Axis values minus the sLun of the r Axis values times the r Slope. The sum for the r Axis and an index are cleared 316, 318. The error each data point io along the r Axis, that is, the deviation of the point from the calculated and offset r Line, is calculated 320 and the square of each deviation is accumulated 322. The index is increniented 324 tnltil all points have been accLunulated 326. The average en-or for the r Axis points is calculated 328.

On the other hand, if the error for the r Axis line is smallest, a line will be calculated for the vertical or s Axis, representing the vertical edge 234. The program will calculate an offset for the point wllere the calculated s Axis will cross the r Axis. To calculate the offset, an index i is again set 330 to zero. Position sums (Esi and Er) for the s Axis 332 and for the r Axis 334 are accwnulated for the number of data point along the r Axis and 20 the index i is incremented 336 until the nuniber of points along the r Axis has been i-eached 338. The respective position sums are multiplied by 2 and divided by the number of data of data points along the r Axis 340 and the s Axis 342, respectively. The s Slope is set 344 to -90 degrees fronl the r Axis line. The offset for the calculated s or vertical line is calculated 346 as half of the accumulated r Axis values minus the swn of the s Axis values times the s Slope. The sum for the s Axis and an index are cleared 348, 350. The error each data point along the s Axis, that is, the deviation of the point from the calculated and offset s Line, is calculated 352 and the square of eacll deviation is accumulated 354. The index is increnlented 356 until all points llave been accumulated 358. The average error for the s Axis points is calculated 360.

Depending on wllich Axis line was calculated, the average en-or for eitllei-the r Axis or the s Axis is compared 362 to a pre-selected maximum. If successfu], the calculated line will be at right angles to the other line and will pass through the data points in such a way that the error is comparatively low. If the error is too large, however, the program inten-upts 364 processing and allows the device to try to capture another image on the next rotation of the centrifilge.

If the lines representing the edges 232, 234 have been successfully established, the program is prepared to call a "get Tt-ansfonn" subroutine 366 to 1-elate the (r, s) co-ordinates to the (x, y) coordinates of the observation regions. An acceptable average ei-ror and an acceptable maximum error are selected 368. Parameters Ps , P,- , Dõ D, and the offset "a"
are selected 370. The transfoi-in preferably takes the fonn set fortli above.

The program then implements the transfonn against the data collected for the horizontal and vertical edges 232, 234 of the mai-ker 214. An error measw-ement is generated, (preferably a root-mean square measurement), to check the efficacy of the transformation. First, for the hoi-izontal edge, an index p is set 372 to zero. For each data point along the x axis, the data point is transfonned 374 into the (x, y) co-ordinate system.
The error in the data is computed and sunulied together 376. If the error is greater 378 than the preselected limit, the program goes 380 to "BREAK", and once again waits for the next frame to appear. The P-counter is incremented 382 until P equals or exceeds 384 the number of data points on the x axis.

Next, for the vertical edge, the index p is set 386 to zero. For each data point along the y axis, the data point is transformed 388 into the (x, y) co-ordinate system. The en-or in the data is conlputed and summed together 390. If the error is greater 392 than the preselected limit, the pi-ograin goes 394 to "BREAK", and once again waits for the next frame to appear. The P-cotuiter is incremented 396 until P equals or exceeds 398 the number of data points on the y axis.

lf the signal pi-ocessing algoritlnn described above has successfiilly located an origin, that is, the intei-section of the edges 232, 234, and the data llas successfully each of the error tests described, and the algorithm has produced a robust transformation for tiie franie, the apparatus may begin processing 398 the data in observation regions. A preferred apparatus and method for such signal processing has been described in US
Patent application 11/772,692 and US Patent application 11/774,073, the disclosw-e of which is incorporated lierein. Such processing would allow the apparatus to distinguish phase boundaries in the observation region 202 and to distinguish cell types in the outflow region 204. In response to observed changes in the i-egions 202, 204, various operating parameters, such as the speed of peristaltic pumps, may be adjusted to control the operating characteristics of the blood processing apparatus.

The optical reference control described herein allows for frame by frame recognition of the location of ilnportant control features within a vibrating two-dimensional optical field. Moi-eover, throughout the process of recognizing the reference point, the quality of the visual image being detected is checked. If the error in the quality of the image exceeds cei-tain limits (for example, when the data points are not sufficiently linear), the fraine is abandoned without further processing in favor of the next fi'ame or visual in7age. Finally, the transfonnation itself is checked against data associated with optical reference 214 before any attempt is made to process data derived from the observation regions. This assures consistent recognition of the sanie physical observation regions, despite imaging of those regions on different pixels in the cainera.
Therefore, sensitive optical recognition techniques can be used with respect to the observation regions, in spite of the vibrations and otlier optical noise assoeiated with a high speed centrifuge.

It will be apparent to those skilled in the art that various modifications and variations can be nlade to the structure and methodology of the present invention without departing from the scope or spirit of the invention. Rather, the invention is intended to cover inodifications and variations provided they conie witllin the scope of the following clainls and their equivalents.

Claims (13)

1. A centrifuge blood processing system for separating fluid components comprising:
a separation chamber rotating about a central rotation axis, said separation chamber having an optical reference mounted thereon, a first detector in optical communication with said separation chamber to detect a plurality of observations from said observation region, a computational apparatus distinguishing said optical reference and establishing reference co-ordinates for gathering data from said observations of said observation region as detected by said first detector, characterized in that said optical reference comprises at least two non-parallel sides and wherein said computational apparatus recognizes a first side represented by a first line and further fits a second line to a second side according to a known angle between said first side and said second side.

2. The centrifuge blood processing system of claim 1, wherein the computational apparatus computes an error measurement for each of said first and second sides and selects the side with the least error as a referent line.

3. The centrifuge blood processing system of claim 2, wherein the computational apparatus computes a dependant line for the side with the greater error.

4. The centrifuge blood processing system of any one of claims 1 to 3, wherein the computational apparatus translates said data from said observation region from a first co-ordinate domain into a second co-ordinate domain.

5. The centrifuge blood processing system of any one of claims 1 to 4, wherein said computational apparatus rejects an observation of said separation chamber if translated data representing said optical reference is not within a pre-selected error.

6. The centrifuge blood processing system of any one of claims 1 to 5, wherein the first and second sides are orthogonal to each other.

7. A method for controlling a centrifuge blood processing system for separating fluid components comprising:
rotating a separation chamber about a central rotation axis, said separation chamber having an optical reference mounted thereon, detecting a plurality of observations from said observation region;
distinguishing said optical reference, establishing reference co-ordinates, and gathering data from said observations of said observation region as detected by said first detector characterized in that said optical reference comprises at least two non-parallel sides and said method further comprising recognizing a first side represented by a first line and fitting a second line to a second side according to a known angle between said first side and said second side.

8. The method of claim 7, further comprising computing an error measurement for each of said first and second sides and selecting the side with the least error as a referent line.

9. The method of claim 7, further comprising computing a dependant line for the side with the greater error.

10. The method of claim 9, further comprising computing a dependant line for the side with the greater error.

11. The method of any one of claims 7 to 10, further comprising translating said data from said observation region from a first co-ordinate domain into a second co-ordinate domain.

12. The method of any one of claims 7 to 11, further comprising rejecting an observation of said separation chamber if translated data representing said optical reference is not within a pre-selected error.

13. The method of any one of claims 7 to 12, wherein the first and second sides are orthogonal to each other.