826 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 58, NO. 3, MARCH 2011

Shear-Mediated Platelet Adhesion Analysis in LessThan 100 μL of Blood: Toward a POC

Platelet DiagnosticNigel J. Kent*, Sinead O’Brien, Lourdes Basabe-Desmonts, Gerardene R. Meade, Brian D. MacCraith,

Brian G. Corcoran, Dermot Kenny, and Antonio J. Ricco*

Abstract—We report a microfluidic chip-based hydrodynamicfocusing approach that minimizes sample volume for the analy-sis of cell–surface interactions under controlled fluid-shear condi-tions. Assays of statistically meaningful numbers of translocatingplatelets interacting with immobilized von Willebrand factor atarterial shear rates (∼1500 s−1 ) are demonstrated. By controllingspatial disposition and relative flow rates of two contacting fluidstreams, e.g., sample (blood) and aqueous buffer, on-chip hydrody-namic focusing guides the cell-containing stream across the proteinsurface as a thin fluid layer, consuming ∼50 μL of undiluted wholeblood for a 2-min platelet assay. Control of wall shear stress is in-dependent of sample consumption for a given flow time. The devicedesign implements a mass-manufacturable fabrication approach.Fluorescent labeling of cells enables readout using standard mi-croscopy tools. Customized image-analysis software rapidly quan-tifies cellular surface coverage and aggregate size distributions as afunction of time during blood-flow analyses, facilitating assessmentof drug treatment efficacy or diagnosis of disease state.

Index Terms—Cell–surface interactions, hydrodynamic focus-ing, image analysis, lab-on-a-chip device, microfluidics, plateletassay, point-of-care diagnostics.

I. INTRODUCTION

P LATELETS are small, discoid (2–4 μm × 0.5 μm) cellfragments that circulate in the blood stream in states rang-

ing from quiescent to fully activated, the latter associated with

Manuscript received July 22, 2010; revised October 19, 2010; acceptedOctober 19, 2010. Date of publication November 9, 2010; date of current ver-sion February 18, 2011. This work was supported by the Science FoundationIreland under Grant 05CE3B 754. Asterisk indicates corresponding author.

*N. J. Kent is with the Biomedical Diagnostics Institute, Dublin City,Glasnevin, Dublin 9, Ireland, and also with The Biomedical Devices andAssistive Technology Research Group, College of Engineering and BuiltEnvironment, Dublin Institute of Technology, Dublin 1, Ireland (e-mail:nigel.kent@dit.ie).

S. O’Brien, L. Basabe-Desmonts, G. R. Meade, and D. Kenny are withthe Biomedical Diagnostics Institute, Department of Molecular and Cellu-lar Therapeutics, Royal College of Surgeons in Ireland, Dublin 2, Ireland(e-mail: sineadobrien2@rcsi.ie; lourdes.b.desmonts@dcu.ie; gmeade@rcsi.ie;dkenny@rcsi.ie).

B. D. MacCraith is with the Biomedical Diagnostics Institute, Dublin City,Glasnevin, Dublin 9, Ireland (e-mail: brian.maccraith@dcu.ie).

B. G. Corcoran is with the School of Mechanical and Manufacturing En-gineeering, Dublin City University. Glasnevin, Dublin 9, Ireland (e-mail:brian.corcoran@dcu.ie).

*A. J. Ricco is with the Biomedical Diagnostics Institute, Dublin City Uni-versity, Glasnevin, Dublin 9, Ireland (e-mail: ajricco@stanford.edu).

Color versions of one or more of the figures in this paper are available onlineat http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TBME.2010.2090659

afflictions from cardiovascular disease to cancer metastasis todiabetes, suggesting that platelet activation assays could supporta range of diagnostic assessments. In platelets’ critical role inhemostasis, damage to a blood vessel exposes a combination ofextracellular proteins, initiating a complex functional responsein platelets [1], [2], the initial stages of which are dominatedby exposed von Willebrand factor (VWF), a platelet-activatingprotein. VWF–platelet interactions thus offer a means to assessplatelet activation; these protein–cell interactions are highly de-pendent on proximal shear forces [3].

Shear-mediated responses in platelets are often studied us-ing parallel-plate flow chambers (PPFCs) [4], which provide alow-aspect-ratio sample-flow path for near-constant wall shearrate over an area suitable for protein deposition and imaging ofsignificant cell numbers. Due to the low aspect ratio, however,most of the blood sample passes through the device without in-teracting with the protein surface or entering the recorded fieldof view: it is effectively wasted. Coupled with the use of tub-ing and syringes to conduct the sample, a single hemodynamicstudy can consume over 5 mL of blood in a 2-min “flow run” [5],precluding, for example, mouse studies and neonatal sample as-says. Repetitive diagnostic blood sampling of low-birth-weightinfants can cause anemia, requiring transfusion [6], and repet-itive blood sampling from adults, especially those in intensivecare units, can have the same consequence [7]. Critically illpatients are often at risk for bleeding, so diagnostic assays ofplatelet function should use small blood sample volumes.

Despite the large blood volumes required, PPFCs can provideuseful data, typically a sequence of fluorescent images, fromeach flow run. These data have, however, proved challengingto analyze: image-based platelet-function analysis typically in-cludes significant sample manipulation and/or laborious manualimage analysis [8], [9].

Hydrodynamic focusing, well known for flow cytometry,transports cells single-file in a cylindrically sheathed streamfor individual analysis. To study cell–surface interactions usingminimum sample volume (50 μL), we have harnessed microflu-idic technologies to implement 2-D surface-contacting sheathedflow, wherein the sample—here, whole blood—is shaped intoa thin layer of cell-containing fluid that traverses a protein-functionalized surface with well-defined shear stress in the in-terfacial region.

Likely clinical applications for a low-blood-volume plateletactivation assay include repetitive monitoring during high-riskpercutaneous coronary intervention (angioplasty) to guide the

0018-9294/$26.00 © 2011 IEEE

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 58, NO. 3, MARCH 2011 827

Fig. 1. Schematic of “hydrodynamic shaping” device to investigate shear-mediated cell–surface interactions. (a) Exploded view of device constituentcomponents: PMMA top plate, laser-patterned PSA layers, and microscopecover slip base plate. (b) Layers from (a) shown fully assembled: (i) detail viewof sample entry area to the device and (ii) detail schematic of blood flow throughthe main channel of the device. (c) Photograph of an assembled, fully functionaldevice.

use of new reversible antiplatelet agents such as ticagrelor [10],and similar monitoring during cardiopulmonary bypass. Com-pelling rationale to monitor the well-documented variable re-sponse to the platelet inhibitor clopidogrel is provided by theadverse events that occur in patients with implanted stents whenplatelet function is insufficiently inhibited [11]. Looking ahead,a platelet activation assay could provide one important part ofan overall prognostic look at a person’s risk of future cardiovas-cular disease [12].

In this letter, we report the design, construction, and testing ofa novel microfluidic device that implements surface-contactingsheathed flow to characterize and study platelet/protein–surfaceinteractions using approximately 100 times less volume for thesame flow run time as commercial PPFCs.

II. EXPERIMENTAL METHODS AND MATERIALS

A. Device Design and Fabrication

Relative to PPFCs, our device: 1) reduces sample volumesand 2) efficiently utilizes the sample by ensuring that most of itpasses directly over the detection area by exploiting microscaleflow and utilizing a surface-adjacent variant of hydrodynamicfocusing. Multiple layers of pressure-sensitive adhesive (PSA)form a 3-D structured flow path [see Fig. 1(a)]. Ease of stack-ing multiple patterned PSA layers makes this an attractive andpowerful basis for microfluidic device design and manufacture.

This 3-D configuration focuses the sample stream at the cen-tre of the device while shaping and directing it into contactwith the protein-coated surface. “Hydrodynamic shaping” isimplemented via a buffer solution that flows through the deviceparallel to the sample stream [see Fig. 1(b)]. Since flow in mi-crodevices is primarily laminar, the added buffer impacts thephysiological aspects of the sample stream minimally: buffer-sample mixing is primarily diffusive and, with principal ioniccomponents matched in sample and buffer, only diffusion ofblood-borne proteins and cells is of potential concern. The dif-fusion coefficients for platelets and for the protein VWF (presentnot only on the device surface, but also in the blood sample) arevery similar, ∼10−7 cm2 /s [13]–[15]. For the flow velocities

used in these experiments, the device transit time of ∼1 s re-sults in a relatively short diffusion distance of ∼6 μm (abouttwo platelet diameters) between the initial buffer–sample con-tact point and the area of interrogation 20 mm down the channel(details below).

Nearly all the sample flowing through the device contactsthe interaction surface in a fluid layer ∼15–30 μm thick. The“Sample Gasket” layer [see Fig. 1(a)] directs buffer around thesample on two sides; the “Buffer Gasket” layer ensures that thesample stream is bounded on “top” as well [see Fig. 1(b-ii)] bybuffer, ensuring sample contact with immobilized protein on thedevice base. The sample port needs no plumbing to an externalactuator [see Fig. 1(c)], so the sample can be dropped into thesample reservoir for “finger-stick” blood sampling.

The three fluid connections and sample reservoir are inte-grated in the 6-mm-thick polymethylmethacrylate (PMMA) topplate, formed by conventional machining (for volume manufac-ture, it would be injection molded). The flow path is defined asstated earlier using CO2-laser-cut double-sided PSA (ArCARE8890, Adhesives Research; a 12.5-μm polyester film is coatedon both sides with medical-grade acrylic adhesive for an overallthickness of 50 μm). A standard glass cover slip is the baseplate. The assembled device [see Fig. 1(c)] is 75 mm × 25 mm(microscope slide size) and 6.3-mm thick.

The main flow channel in the “Buffer Gasket” is 4-mm widefor ease of visualization; each of the bifurcated channels is2-mm wide, minimizing pressure drop where they combine toform the main channel. The entrance area, a 750-μm-wide ×10-mm-long channel, delivers sample to the device. The mainchannel length, 35 mm, ensures that platelet–surface interac-tions are observed well downstream of any entrance effects.A 5-mm-diameter hole in the “Buffer Gasket” directs sampleto the “Sample Gasket” layer, which defines the main channeldimensions (4 mm× 35 mm × 50 μm thick) and contains a sim-ilar bifurcation, again minimizing buffer-fluid entrance effects.Entrance length, commonly defined as 0.06 times the Reynoldsnumber multiplied by the hydraulic diameter [16], is the regionwhere flow velocity deviates from its fully developed value bymore than 5%; this length, about 3.6 mm for our device and flowparameters, is much less than the 20-mm downstream location ofthe device’s 250 μm × 300 μm interaction-and-imaging region,which is also located ∼2 mm away from the channel side walls,ensuring unidirectional shear forces acting on the platelets.

The device base plate, a standard glass microscope coverslip,was chosen because: 1) protein deposition and characterizationtechniques are well established on glass with its predictable,reliable surface chemistry; 2) the slides’ high optical qualityand optical properties are compatible with a range of off-the-shelf optical systems to detect platelet–surface interactions atappropriate resolution; 3) cover slips are a widely availablecommodity with tightly controlled dimensional, optical, andsurface-finish tolerances. Cover slip base and machined top plateare washed with absolute ethanol and dried by evaporation priorto device assembly.

For assembly, the glass cover slip is placed in a customalignment jig and one release liner removed from the PSA“Sample Gasket” layer, which is then aligned and adhered to

828 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 58, NO. 3, MARCH 2011

the cover slip; a similar process adds the “Buffer Gasket” layer.The remaining release liner is removed and the PMMA topplate aligned and adhered to the exposed PSA surface. Theentire assembly is removed from the jig and pressed togetherthrough two rollers. Inlet and outlet fittings are 1.6-mm threadedpolypropylene connectors for connection to silastic tubing(Glycotech). Completed devices are rinsed by perfusion withaqueous phosphate-buffered saline (PBS).

B. Experimental Procedure

After device assembly and prior to experimental runs, theglass cover slip surface within the flow device was coated insitu with purified VWF by static incubation (2 h, 100 μg/mLVWF solution in PBS) [17]. The VWF-coated cover slip wasthen “blocked” against adsorption onto any remaining “active”(adsorption-prone) surface regions by static incubation with 1%BSA in PBS for 1 h, and finally washed by gentle perfusionwith >3 mL of PBS through the assembled device. Deviceswere typically prepared on the morning of the experiments andstored at room temperature prior to use.

Whole blood was drawn from the antecubital vein of healthydonors not known to have taken any medication within the previ-ous ten days, using a 19-gauge needle. After discarding the first5 mL to prevent platelet activation, blood was drawn into a cleanpolypropylene syringe containing trisodium citrate as anticoag-ulant (final concentration, 0.32%: 1:9 ratio of citrate solutionto blood). Some samples were treated with ReoPro (20 μg/mLfinal concentration; Eli Lilly and Company; added 20 min priorto experimental analysis). For fluorescence imaging, plateletsin blood were labeled with DiOC6(3) (1 μM; Invitrogen) byaddition and gentle mixing by inversion followed by incubationfor 10 min prior to analysis at 37 ◦C.

The prepared device was placed on a thermally controlledmicroscope stage and the outlet plumbed to a neMESYSpumping system (cetoni GmbH) via a syringe of PBS using1.6-mm inner-diameter tubing. Similarly, the buffer inlet portwas plumbed to the pump via a syringe containing Histopaque,a biocompatible buffer (viscosity = 0.004 Pa·s) chosen to min-imize diffusion effects and viscosity mismatch between wholeblood and sheathing buffer solution. Viscometric measurementsin our laboratory yield a typical range of viscosity of wholehuman blood of 0.0032–0.0040 Pa·s. At arterial shear rates(>1000 s−1), platelet adhesion depends primarily on plateletinteractions with VWF [9], which are largely independent ofsmall variations in shear rate. For our experiments, blood vis-cosity of 0.0032 Pa·s results in a shear rate at the platelet/proteininterface of 1615 s−1 , while a 0.004-Pa·s sample is subject toa shear rate of 1500 s−1 . This 8% shear rate difference shouldnot significantly influence platelet activation state nor affectplatelet/protein–surface interactions.

For imaging sequences, a custom-designed miniaturizedepifluorescence optical system, described in detail elsewhere[18], [19], with a 20×, non-oil-immersion objective, was uti-lized. Whole blood samples were fluorescently excited using a16-mW, 473-nm laser diode (Photometrics) in combination witha filter set (ALPHA Vivid: XF100-3) from Omega Optics.

Blood samples were assayed at a physiologically relevantwall shear force (stress) of 6 N/m2 , which, assuming a bloodviscosity of 0.004 Pa·s, occurs at an outlet volumetric flow rateof 600 μL/min. To this end, the outlet port of the device wasplumbed to a syringe and negative pressure applied to generatethe appropriate flow rate through the main channel of the device.Sample volume consumption was defined by a positive buffervolume flow rate of 575 μL/min: the difference between bufferflow rate and total outlet flow resulted in a blood flow ratethrough the device of just 25 μL/min.

In total, ten experimental runs were performed over two daysusing blood from multiple donors. Each donated blood samplewas divided and assayed under both normal and ReoPro-treatedconditions (see Section III).

C. Image Analysis

For each experiment, 600 images were recorded at∼4.9 frames/s, for an overall run time of ∼120 s, requiringa blood volume of 50 μL. Acquired images passed througha custom-designed algorithm to determine coverage areas ofadhered platelets. First, individual objects (platelets) in eachimage (frame) were detected using custom software developedin LabVIEW [20], based primarily on fluorescent intensitiesof platelets relative to background. To increase throughput andeliminate operator bias, an automated algorithm assigns pixelswith intensities above a predetermined threshold a numeric valueof one and those below it a value of zero. Frame-by-frame calcu-lation of the threshold limit optimizes it over a single flow run,particularly useful toward the end of a series of images, whereincreasing numbers of platelets may contribute significantly tobackground noise, making a static threshold limit ineffective. Todetermine platelet surface coverage for each frame, the softwarecalculates the total area covered by those pixels with intensityvalue one as a percentage of the total area of pixels in the entireframe area.

A particle-analysis suite of tools within LabVIEW storesgroups of contiguous nonzero particle clusters and records in-formation related to particular attributes, such as size, shape, andlocation. For platelet-function studies, the software stores over-all areas of individual platelets and platelet aggregates adheredto the surface at any given time point.

III. RESULTS AND DISCUSSION

To characterize the performance and diagnostic potential ofthe hydrodynamic-shaping blood-flow device, platelet–surfaceinteractions were dynamically imaged in flowing whole humanblood and analyzed. Citrated blood with and without addedc7E3-Fab fragment, ReoPro, were compared. ReoPro blocksthe αIIbβ3 receptor on the platelet surface, inhibiting stableadhesion to the VWF surface via its interaction with the C1domain of VWF [21]. ReoPro also inhibits fibrinogen-mediatedplatelet–platelet aggregation via the αIIbβ3 receptor, suiting itto multiparameter characterization of our device.

IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 58, NO. 3, MARCH 2011 829

Fig.2. Dynamic adhesion results for fluorescently labeled platelets adheringto the VWF-coated surface in our custom parallel-plate device. Blue curveshows development of percentage coverage of detected platelets over 120 s foran average of five non-ReoPro treated blood samples. Red curve shows thedevelopment of percentage coverage of detected platelets from the same donorsover the same time period with the addition of ReoPro.

A. Surface Coverage

Fig. 2 shows average increases in surface coverage over timefor both ReoPro-treated and untreated citrated blood samples.Untreated blood samples (see the blue curve in Fig. 2) show asteady increase in surface coverage with blood flow time, froman initial coverage of ∼6% to a final coverage of ∼25% over120 s. For ReoPro-treated blood, however (see the red curve inFig. 2), an increase similar to that of untreated blood is observedduring the initial 60 s of the run, but, subsequently, the surfacecoverage decreases to ∼14% after 120 s.

We attribute similar initial increases in surface coverage withtime to platelets initially interacting with the A1 domain ofVWF via the receptor GpIb/IX/V on the platelet surface, whichReoPro does not inhibit. Subsequent differences (surface cover-age decreases only for ReoPro-treated samples) are attributed toReoPro’s inhibition of the αIIbβ3 receptor, impairing plateletadhesion to VWF over longer times.

B. Aggregate Size Distribution

For platelet aggregation analysis, ten images (frames 575–585) were taken as a representative snapshot of surface coverageat that time (∼118 s), providing 2 s of data and averaging outsome of the noise. The area of each detected platelet or aggregatein each of the ten images was recorded and a mean particle sizeand standard deviation were determined.

Mean particle sizes for each flow run (see Fig. 3) compare re-sults from ReoPro-treated (green) and untreated (red) blood flowruns. ReoPro treatment results in a mean aggregate size of 12.5± 0.5 μm2 (1 standard deviation) while, in contrast, untreatedsamples have mean aggregates of 19 ± 2 μm2 . Because Re-oPro inhibits fibrinogen-mediated platelet–platelet binding viaαIIbβ3, platelet aggregation on the VWF surface is inhibited.Untreated samples have larger aggregates with larger standarddeviations: ReoPro appears to obscure some donor-to-donor bi-ological variability.

Fig. 3. Aggregation results for fluorescently labeled platelets adhering to theVWF-coated surface in our custom parallel-plate device. Bars represent meansize of detected platelet and platelet aggregates taken over frames 575–585 ofeach flow run. In the case of the ReoPro treated samples (shown in green) themean aggregate size is approximately 12.5 μm2 while in the case of the non-ReoPro treated samples the mean size of platelets and platelet aggregates variesfrom approximately 17.5 μm2 to approximately 20 μm2 .

IV. SUMMARY AND CONCLUSION

We have developed a new flow device for shear-mediatedplatelet function analysis with several advantages over exist-ing commercial systems. A unique combination of microflu-idic techniques minimizes sample consumption independent ofoverall flow rate for a given shear force. The “finger-stick”compatibility of sample delivery to the device has the potentialto increase throughput and minimize patient discomfort, pro-vided a non-platelet-activating finger-stick technology is devel-oped. The design strategy and materials of device manufacturelend themselves to both rapid prototyping and mass production,pointing the way toward a cost-effective alternative to currentcommercial systems. In combination with the customized im-age analysis algorithm we have developed, this device couldenable a compact desktop system to bring platelet assays toroutine point-of-care use. Near-term applications to monitorangioplasty, cardiopulmonary bypass, and clopidogrel efficacyare anticipated. A patient’s platelet surface adhesion parametersmay some day be as routinely measured as total blood count orglucose levels.

REFERENCES

[1] A. D. Michelson, Platelets, 2nd ed. San Diego, CA: Academic, 2006,pp. 145–148.

[2] P. Harrison, “Platelet function analysis,” Blood Rev., vol. 19, pp. 111–123,2005.

[3] M. H. Kroll, J. D. Hellums, L. V. McIntire, A. I. Schafer, and J. L. Moake,“Platelets and shear stress,” Blood, vol. 88, pp. 1525–1541, Sep. 1996.

[4] E. Gutierrez, B. Petrich, S. Shattil, M. Ginsberg, A. Groisman, andA. Kasirer-Friede, “Microfluidic devices for studies of shear dependentplatelet adhesion,” Lab Chip., vol. 8, no. 9, pp. 1486–1495, Sep. 2008.

[5] D. C. Brown and R. S. Larson, “Improvements to parallel plate flowchambers to reduce reagent and cellular requirements,” BMC Immunol.,vol. 2, pp. 1–7, Sep. 2001.

[6] M. Obladen, M. Sachsenweger, and M. Stahnke,, “Blood sampling in verylow birth weight infants receiving different levels of intensive care,” Eur.J. Pediatr., vol. 147, no. 4, pp. 399–404, May 1988.

830 IEEE TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 58, NO. 3, MARCH 2011

[7] B. R. Smoller and M. S. Kruskall, “Phlebotomy for diagnostic laboratorytests in adultsPattern of use and effect on transfusion requirements,” N.Engl. J. Med., vol. 314, no. 19, pp. 1233–1235, May 1986.

[8] A. S. Kantak, B. K. Gale, Y. Lvov, and S.A. Jones, “Platelet functionanalyzer: Shear activation of platelets in microchannels,” Biomed. Mi-crodevices, vol. 5, no. 3, pp. 207–215, 2003.

[9] N. A. Mody, O. Lomakin, T. A. Doggett, T. G. Diacovo, and M. R. King,“Mechanics of transient platelet adhesion to von willebrand factor underflow,” Biophys. J., vol. 88, no. 2, pp. 1432–1443, Feb. 2005.

[10] L. Wallentin, R. C. Becker, A. Budaj, C. P. Cannon, H. Emanuelsson,C. Held, J. Horrow, S. Husted, S. James, H. Katus, K. W. Mahaffey,B. M. Scirica, A. Skene, P. G. Steg, R. F. Storey, R. A. Harrington, A. Freij,M. Thorsen, and PLATO Investigators, “Ticagrelor versus clopidogrel inpatients with acute coronary syndromes,” N. Engl. J. Med., vol. 361,no. 11, pp. 1045–1057, Sep. 2009.

[11] L. Bonello, U. S. Tantry, R. Marcucci, R. Blindt, D. J. Angiolillo, R.Becker, D. L. Bhatt, M. Cattaneo, J. P. Collet, T. Cuisset, C. Gachet, G.Montalescot, L. K. Jennings, D. Kereiakes, D. Sibbing, D. Trenk, J. W.Van Werkum, F. Paganelli, M. J. Price, R. Waksman, and P. A. Gurbel,“Consensus and future directions on the definition of high on-treatmentplatelet reactivity to adenosine diphosphate,” J. Am. Coll. Cardiol., vol.56, no. 12, pp. 919–33, Sep. 2010.

[12] L. Basabe-Desmonts, G. Meade, and D. Kenny, “New trends in bioany-litical micro devices to assess platelet function,” Expert Rev. Mol. Diag.,vol. 10, pp. 869–874, Oct. 2010.

[13] E. C. Eckstein and F. Belgacem, “Model of platelet transport in flowingblood with drift and diffusion terms,” Biophys. J., vol. 60, pp. 53–69, Jul.1991.

[14] V. T. Turitto, A. M. Benis, and E. F. Leonard, Ind. Eng. Chem. Fundam.,vol. 11, pp. 216–223, 1972.

[15] J. Loscalzo, M. Fisch, and R. I. Handin, “Solution studies of the quaternarystructure and assembly of human von Willebrand factor,” Biochemistry,vol. 24, pp. 4468–4475, 1985.

[16] B. R. Munson, D. F. Young, and T. H. Okiishi, Fundamentals of FluidMechanics. New York: Wiley, 1998, p. 464.

[17] N. J. Kent, L. Basabe-Desmonts, G. Meade, B. D. MacCraith, B. G.Corcoran, D. Kenny, and A. J. Ricco, “Microfluidic device to study arte-rial shear-mediated platelet–surface interactions in whole blood: Reducedsample volumes and well-characterised protein surfaces,” Biomed. Mi-crodevices, vol. 12, no. 6, pp. 987–1000, 2010.

[18] N. J. Kent, “Development of a microfluidic platform and detection systemfor platelet function analysis,” Ph.D. dissertation, Dept. Mech. Manf. Eng.,Dublin City Univ., Dublin, Ireland, 2009.

[19] N. J. Kent, S. O’Brien, Lourdes B.-Desmonts, G. R. Meade, B. D.MacCraith, B. G. Corcoran, D. Kenny, and A. J. Ricco, “Miniature nearpatient readout system for platelet diagnostics,” in preparation.

[20] NI Vision Concepts Manual, NI Corporation, Austin, TX, 2000–2005[Online]. Available: http://www.ni.com/pdf/manuals/372916e.pdf

[21] S. H. Tam, P. M. Sassoli, R. E. Jordan, and M. T. Nakada, “Abciximab(ReoPro, chimeric 7E3 Fab) demonstrates equivalent affinity and func-tional blockade of glycoprotein IIb/IIIa and alpha(v)beta(3) integrins,”Circulation, vol. 98, pp. 1085–1091, 1998.