evaluation of automatic platelet counting system utilizing whole … · evaluation...

J. clin. Path., 1972, 25, 218-226

Evaluation of an automatic platelet counting systemutilizing whole bloodR. M. ROWAN, W. ALLAN, AND R. J. PRESCOTT

From the Department of Haematology, Royal Infirmary, Edinburgh

SYNOPSIS Technicon's newly developed platelet AutoCounter utilizes an electro-optical systemwhichdetects and counts particles by sensing the scattering of light which occurs when blood cells flowthrough the illuminated sensing chamber of a micro-optical system. The system utilizes whole bloodcollected into EDTA. Blood samples, which can be handled at the rate of 40 per hour, are dilutedwith 2M urea which in addition causes lysis of the erythrocytes. The haemolysate thus obtained isdelivered to the particle counter after a two-minute reaction time, the cell count finally being dis-played on a continuous chart recorder.An evaluation of this machine has been carried out in the Department of Haematology, the Royal

Infirmary, Edinburgh, on hospital specimens from patients suffering from a wide variety of diseases.The results of this evaluation are presented. Reproducibility studies indicate a coefficient ofvariation of 4% at any platelet level. The percentage drift per sample ranges from + 04 to - 0-4%of the total count. Carry-over and departures from linearity are statistically significant; however,the magnitude of these deviations is not sufficient to cause concern in routine use. Acceptableagreement is noted between machine counts and the counts obtained by technicians using phase-contrast microscopy with the exception of two anomalous individuals.

Discrepant counts were noted in patients with elevation of the erythrocyte sedimentation rate andalso in patients with disturbance of immunoglobulins. A modification to the sampling probe whicheliminates the former problem is described.The AutoCounter described in this paper provides a fast, reliable, and accurate service laboratory

platelet counting system.

Morphological and numerical evaluation of formedblood elements constitutes useful guidance in theassessment of normal and disturbed haemopoieticfunction. Because of increasing interest in auto-mation of the haematology laboratory there is needfor a parallel increase in cooperation between thehaematologist and the automated instrumentdesigner in order that each may gain insight intothe other's problems. Already automation proceduresin haematology are well established with respect tored and white cell counting (Nelson, 1969; Barnard,Carter, Crosland-Taylor, and Stewart, 1969) but thissatisfactory state does not yet extend to enumerationof platelets, although the clinical usefulness ofplatelet counts is irrefutable.The accuracy of platelet counting techniques is far

from satisfactory. In 1937 Tocantins evaluated 116different methods of platelet counting. Since thenReceived for publication 8 December 1971.

218

modifications have been made to older methods andadditional newer methods have been introduced sothat the number of techniques now available exceeds160 (Maupin, 1969). This proliferation of methodsproves the inadequacy of available techniques andalso stresses the need for improved systems.

Direct counting methods involve the absoluteenumeration of platelets in a suitable solution andthus the platelet count is not related to a secondknown quantity of formed elements as was thepractice with now discarded indirect methods ofplatelet counting. Examples of direct countingmethods in current use include (1) light or phase-contrast microscopy techniques, (2) electronicparticle counting which detects and counts cells bysensing a change in the electrical conductivity of achannel separating two electrolyte solutions whenblood cells flow through this channel, and (3) theelectro-optical method of counting which detects

copyright. on F

ebruary 22, 2021 by guest. Protected by

http://jcp.bmj.com

/J C

lin Pathol: first published as 10.1136/jcp.25.3.218 on 1 M

arch 1972. Dow

nloaded from

http://jcp.bmj.com/

Evaluation of an automatic platelet counting system utilizing whole blood

and counts cells by sensing the scattering of lightwhich occurs when blood cells flow through theilluminated sensing chamber of a micro-opticalsystem.

These latter techniques have certain inherentlimitations which, depending on the effectiveness ofinstrument design, affect the ultimate accuracy andreproducibility of the blood cell counting system.Electronic particle counting systems have beenconsidered in a number of recent reviews (Sipe andCronkite, 1962; Eggleton and Sharpe, 1963; Bull,Schneiderman, and Brecher, 1965). Even by suchsophisticated means, however, platelet counts aredifficult to accomplish reliably. The main problemsencountered include unsatisfactory separation ofred cells from platelet-containing plasma; thetendency of platelets to aggregate; and the presencein the diluent of particles which are only littlesmaller than platelets and which are not excluded bythe threshold settings employed. Furthermore,significant errors may arise from the instrumentitself and its calibration.

Fig. 1 The AutoCounter system.

LAMP

Fig. 2 The optical system (by courtesy of the Technicon Corporation).

The purpose of this paper is to present the resultsof an evaluation of the Technicon platelet Auto-Counter carried out in the Department of Haema-tology, the Royai Infirmary, Edinburgh. TheTechnicon system is an electro-optical systemutilizing the light-scattering principle.

The AutoCounter System

The Technicon platelet AutoCounter is fullyautomated to quantitate platelets rapidly from wholeblood without prior manipulation. One operator canhandle the analytical train and the system deliversplatelet counts at the rate of 40 per hour.The five main components of the system are

illustrated in Figure 1. The sampler module consistsof a removable circular tray with spaces for 40sample containers. The blood samples are held inuniform suspension by rotating Teflon paddles. Aprobe automatically dips into each sample cup andaspirates 0 4 ml whole blood. A proportioning pumpcontinuously propels diluting fluids and samples at

FLOWPASSAGE FLOW

PROJECTION \ CELL

219

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/

R. M. Rowan, W. Allan, and R. J. Prescott

precisely determined flow rates through a closedflow system of plastic and glass tubing (manifold), inwhich samples and reagents are brought togetherunder constant physical conditions. The first dilutionin the manifold, as determined by the proportioningpump and the inside diameter of the manifold, is a1:21 dilution and the final dilution eventuallydelivered for counting is 1:1 500. Following aspi-ration of each sample, a small bubble of air isaspirated, followed by wash fluid, followed in turnby another bubble of air, and then the sequencerepeats itself. The air segments provide an anti-diffusion barrier between sample and wash fluid andthe next mixed sample.The cell counter uses an optical system (Fig. 2) to

generate light pulses from platelets passing throughthe flow cell. Light from the lamp passes through acondensing lens, uniformiy illuminating the primaryaperture. The primary aperture has a small rectan-gular opening in its centre. A projection lens picksup the light coming through this aperture and formsa reduced image of the aperture in the centre of theflow cell. This results in a small, brightly illuminatedview film, sharply defined by the reduced apertureimage and the size of the flow passage in the flowcell. The light then emerges from the flow cell and isblocked by a dark field disc which prevents anydirect light from entering the photomultiplier tube.This is known as a reverse dark field illuminationsystem.

If a clear liquid passes through the flow cell, thelight rays are blocked by the dark field disc asdescribed above, and no light energy reaches thephotomultiplier tube; however, if particles, such asblood cells, are dispersed in the liquid stream, theyare illuminated as they pass through the view volume.

SAMPLE POSITION

Flowcell A

Pumpt To Wastef3 0000I

Channel B--- 2 From ManifoDd

Rotor

Wash Reagent

When illuminated, the particles become secondarysources of illumination, scattering light in a forwarddirection towards the objective lens. The objectivelens collects this light and focuses it through a smallaperture in front of the photomultiplier tube. Theaperture blocks out stray light, allowing only thescattered light from the particle to enter the photo-multiplier tube. Thus, as each particle passes throughthe view volume, an electrical pulse is generated.These pulses can be viewed on an oscilloscopescreen located on the front panel. The light pulses,once transformed into electrical pulses, are amplifiedand counted in an electrical system finally beingdisplayed on chart paper.

Figure 3 shows the cell counter wash valve. In thesample position (Fig. 3A) the diluted haemolysateflows through the flow cell from above downwards.A rotating valve permits flow of a sodium hydroxide/tergitol mixture in the reverse direction (Fig. 3B) tocleanse thoroughly the narrow channel of the flowcell between each sample.The diluent, which consists of a two-molar urea

solution, acts in addition as an haemolysing agentwhile preserving platelets and leucocytes. Accord-ingly the total leucocyte count must be subtractedfrom the machine result before issuing a plateletcount.A standard is provided by the Technicon Company

which consists of preserved human platelets sus-pended in plasma. Initially both normal and lowrange standards were recommended but calibrationof the AutoCounter on a single standard of concen-tration approximately 300 000 platelets/c mm hasproved completely satisfactory.

Using materials recommended by the instrumentmanufacturers each platelet count costs 3-6p (July

Wash ReagentFig. 3 The cell counter wash valve (by courtesy of the Technicon Corporation).

220

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/

Evaluation ofan automatic platelet counting system utilizing whole blood

1971). This figure does not take into account thepurchase or depreciation of the machine, nor does itinclude servicing costs.

Materials and Methods

The haematology laboratory of a large teachinghospital was the site of this evaluation, specimenscoming from patients with a wide variety of diseases.

Blood was withdrawn into dry plastic syringesusing a 21 G needle taking care that the venepuncturewas a clean one and that the blood flowed into thesyringe spontaneously without frothing. The needlewas then detached from the syringe and the requisiteamount of blood delivered without frothing into anappropriate vessel containing dipotassium EDTAgiving a final EDTA/blood concentration of 1 mg/ml. Blood and anticoagulant were gently mixedwithout delay. All samples were processed withintwo hours.The visual direct counting technique using the

formal-citrate red-cell diluent described by Dacieand Lewis (1970) was used. Platelets were countedusing phase-contrast microscopy. The number ofplatelets in a minimum of 4 mm2 on a Neubauerchamber was counted. Whenever duplicate ortriplicate counts were made, they were true replicatesand performed by different technicians.The AutoCounter was set up according to the

manufacturer's recommendations. Optical focusingis critical, but once aligned the system proved to bestable.

It is important to wash the AutoCounter manifoldthoroughly twice daily using a sodium hydroxide/tergitol wash solution. The diluent filter (Whatman'sgrade 03 filter candle) proved entirely satisfactoryprovided it was not permitted to dry out when ureawould crystallize. The filter requires changing wheninsufficient diluent is being drawn through thesystem. Such an occurrence should be consideredwhen there is any upset in the manifold bubblepattern.

Results and Discussion

REPRODUCIBILITY AND DRIFTThe effect of platelet levels on the reproducibility ofthe AutoCounter was examined with a series ofduplicate counts on 112 blood samples which wereobtained over a period of six days. On each day theprocedure was to count the complete batch and thento re-count the samples. As no adjustments weremade to the AutoCounter in this time, the differencesbetween pairs of observations on the same samplewere due to the variability of the machine with agiven sample and to drifting of the machine before

resampling. For each day's run the mean propor-tionate change between pairs of observations on thesame blood was determined and a correction basedon this mean change was applied to the secondreading from each pair. The corrected difference isplotted against the corrected average in Figure 4. Alinear regression of the difference on the averagegave a small negative intercept on the y axis whichdid not differ significantly from zero. Accordingly aregression through the origin was performed and isalso shown in Figure 4. Thus, over the rangeencountered in this series, differences of approxi-mately 5 % of the average value would be expected inpairs of observations. Alternatively expressed,the coefficient of variation at any platelet level isapproximately 4 %.

This experiment also enabled the rate of drift to becalculated for each day's set of observations, usingthe mean proportionate change between counts onthe same blood as the basis for the calculation. Therates of drift on the six days, expressed as thepercentage drift per sample, were - 0 3 %,-0-1 %,-0 4%, -0 4%, +0-4%, and 0-0%.The measure of variability in duplicate counts at

different platelet levels, as described above, is usefulin demonstrating the relationship between repro-ducibility and platelet level but as an estimate ofreproducibility error in routine use it will probablygive overestimates, since it is based on long periodsof using the machine without readjustment. A smallerseries of six bloods throughout a range of 30 000 to600 000 platelets which were successively resampled

40

E 50-

xI--

-0 4@

30

c l

0 20

10

0t

v Ii., - *-. .

100 200 300 400 500 600Average Count (x103/mm3)

Fig. 4 Reproducibility experiment showing thecorrected difference plotted against the corrected average.

221

I._I

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/


at least 12 times gave standard deviations whichwere approximately half the size of those predictedabove (Table I).

Blood No. Platekt Count x 10J mm'

1 35 115 220 315 460 6602 35 110 225 315 460 6503 35 115 220 320 460 6704 30 100 220 310 470 6505 30 100 225 315 470 6456 30 105 230 310 465 6507 30 105 220 305 465 6658 30 105 230 310 465 6609 30 105 230 305 470 66010 30 105 225 305 460 65011 30 110 235 310 460 65012 30 105 220 320 460 660

Table I Replicate studies on a series of six bloodssuccessively resampled 12 times

LINEARITY AND CARRY-OVERFour hundred ml of whole blood was withdrawnfrom a normal blood donor into sequestrene givinga final concentration of 1 mg/ml. This was subse-quently manipulated by differential centrifugationto give platelet-rich plasma to which red blood cellswere then added in sufficient volume to produce aPCV within normal limits (40%). This manipulatedwhole blood sample, which had a final platelet countof 650 000, was then diluted with normal saline togive 80%, 60%, 40%, 20%, and 10% dilutions.Normal saline was used as a zero control. Each ofthose seven dilutions was subsequently divided intoseven separate aliquots. The 49 samples thus obtainedwere then counted on the AutoCounter. The order ofpresentation was carefully controlled to ensure thatthe seven samples at each dilution were immediatelypreceded by one sample at each of the seven dilutions.When examining the results it was clear that the

machine had drifted, and a correction was appliedassuming a constant proportional drift throughoutthe duration of the experiment. The size of this driftwas estimated in such a way that the error sum ofsquares was minimized in the subsequent analysis,and the magnitude of the drift was found to be 0I %per observation.

A further factor influencing the analysis was theunequal variability at different dilutions. In order togive an appropriate weight to each reading, thevariance of the observations at each dilution was

calculated and a linear regression of variance on

dilution was performed. The weighting factors in thefollowing analysis were then taken to be proportionalto the reciprocal of the fitted variances at eachdilution.The design of the experiment enabled it to be

treated in the first instance as a randomized complete-block design and an analysis of variance determinedthe sums of squares attributable to experimentalerror, carry-over, and differing dilutions. The sums

of squares attributable to dilutions were furthersubdivided into that due to a linear associationbetween the count and dilution and that due todeviations from linearity.

Table II presents the analysis of variance, and it isseen that both carry-over and departures fromlinearity are statistically significant. However, bothof these effects are small and a less detailed experi-ment would have been unlikely to detect them. If weassume no carry-over at the 0% dilution, theobserved carry-over at the 100, 80, 60, 40, 20, and10% dilutions are 2-1, 5 1, 2X4, 4X8, 27, and 2-6 x 103

Blood DilutOn (%'

Fig. 5 Linearity experiment.

Source of Variation df Sums of Squares Mean Square F

Linearity with dilution 1 34754035Deviations from linearity 5 24724 4945 34-6 (p < < 0001)Carry-over 6 2103 305 5 2-5 (p < 0-05)Experimental error' 35 5002 142-9Total' 47 34785864

Table II Analysis of variance

"The degrees of freedom for these entries have been reduced by 1 from their normal values to allow for the drift correction which has beenapplied.

222

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/

Evaluation of an automatic platelet counting system utilizing whole blood

counts/mm3, respectively. The non-linearity is shownin Figure 5. It is evident that agreement is good overthe lower part of the range, with a tendency to countlow at high platelet concentrations. The magnitudeof this deviation would not be sufficient, however, tocause concern in routine use.

AGREEMENT WITH PHASE-CONTRAST COUNTSTwo series of blood were counted by experiencedtechnicians using the phase-contrast method andusing the automatic counter. The white blood countwas obtained separately using the Coulter model S.As it was known that AutoCounter results frompatients with myelomatosis may be unreliable(Widmark and Weschler, 1970), such bloods werenot included in the series. However, patients withnon-paraprotein ESR elevation were included to seeif the high ESR as such was instrumental in causingthe discrepant counts. Figures 6 and 7 show the plotof the machine count against the combined plateletand WBC count obtained by technicians. Thosesamples with high ESRs (>20 mm) are shownseparately from those with normal ESRs. In bothseries a statistical comparison of the linearregression lines, obtained separately for low andhigh ESR bloods (by means of a weighted analysis ofcovariance), showed no significant difference in slope,intercept, or scatter about the fitted regression lines(p > 0-05). The lines shown in Figs. 6 and 7 give thebest weighted least squares fit, obtained frompooling all points, irrespective of the ESR.

In both series there is generally reasonableagreement to the regression line though each serieshas one wild point whose distance from the regressionline is greater than that expected by random varia-tion. There is no apparent common factor betweenthem, one blood having a high ESR and the other anormal ESR. As these points differ significantlyfrom the regression lines and would have had anundue effect on the remainder of the statisticalanalysis, they were not included in the analysispreviously described.

COMPARISON OF LINEARITY WITH AUTO-COUNTER AND PHASE-CONTRAST METHODSThree bloods were prepared at differing dilutionsand each sample was counted on the AutoCounterand by two technicians using the phase-contrastmethod. Figure 8 shows the results of the Auto-Counter and the average of the two technicians. Theindividual counts of the two technicians weregenerally in extremely close agreement, and neverdiffered by more than 38 x 103 counts/mm3. Thelines in Fig. 8 show the best fitting line through theorigin for the autocounts. For each blood thelinearity is seen to be good. The manual counts also

600

500'

E

E400cx

2 3001c

0

00-

K)LI

70(

604

CEE

-,Xt,c

40(0

0

30C

20C

lO1

Xx .xx x

Axx.. .../x xXLOW ESR

x HIGH E.SR

2 300 400 500 600Manual Count (Phase-Contrast)+ WBC Count

(x10/mm )

Fig. 6 Machine count plotted against combinedplatelet and white cell count obtained by technicians.

Ann-

V~~~~~~~~~~~~~~~~~~

0~~~~~~~~~~~~~

I~ ~~~~~~~~~) Xx%. *x

x/x: x

X.K~~~~~~~~X x

./ . ~~~~~LOWESR

X 'XES

100 200 300 400 500 f600Manual Count (Phase-Contrast) + WBC CQut

(x103/mM3)

Fig. 7 Machine count plotted against combinedplatelet and white cell count obtained by technicians.

223

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/


0A'

CL

* Machine Count

x Technician Count

25 50 75 00 25 50 75 100 25 so 75Blood A Blood B Blood C

Dilution %

Fig. 8 Experiments comparing linearity with the AutoCounter and the phase-contrast microscopy method.

show reasonable linearity for bloods A and B.However, for blood C, which was manipulated togive a particularly high platelet count at 100%dilution, there is a noticeable non-linearity, takingthe form of a low count at the higher plateletconcentrations.

Discussion

Evaluation of the platelet AutoCounter system hasbrought to light certain conditions which sub-stantially alter the automatic counts. Bloods with ahigh ESR produce a peaking phenomenon on thetracing (Fig. 9) which, although not causing seriouslydiscrepant results, may in fact lead to difficulty inreading the result. The only consistent abnormalityin these bloods was marked elevation of fibrinogenlevels. Protein electrophoresis on cellulose acetatefailed to demonstrate any consistent or significantanomaly. Substitution of patient's plasma witheither physiological saline or ABO compatiblenormal plasma removed this 'peaking' phenomenon.It was concluded that 'peaking' was produced bysedimentary tailing akin to erythrocyte sedimenta-tion in the sample as it passed through the dilutingsystem. This was in fact visible on examination of thered cells in the sample at various points in themanifold. In the counting chamber a greaterproportion of particles presented in the last third ofthe sample but the increase was not linear and theproblem could not be cured by taking the meancount. Accordingly a modified sampling probe wasconstructed which permitted air segmented sampling

Fig. 9 Peaking phenomenon produced by bloods with a

high ESR.

instead of continuous blood sampling. This removedthe problem of peaking (Figs. 10 and 11).

It has been previously reported by Widmark andWeschler (1970) that patients with abnormally highserum proteins, such as paraprotein disturbancesand red cell autoantibody disorders, may cause 'off-scale' results. This has been our experience in that inaddition to 'peaking', serious discrepancies have

100

224

0

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/

Evaluation ofan automatic platelet counting system utilizing whole blood

LI4; e

Fig. 10 Two series of counts. Those counts on the right-hand side of the illustration show peaking produced bybloods with a high ESR. The tracings on the left-handside show correction of this phenomenon following theincorporation of air segmented sampling.

!SOO

1400.

a300*/

200

tOO 200 300 400 SO 600 00Modified Sampling Technique (Counts x103/mn3)

Fig. 11 Counts using the original sampling techniqueplotted against those obtained with the modified samplingtechnique.

been demonstrated between automatic counts andvisual counts in patients with paraprotein disorders.In a series of 10 patients with paraprotein dis-turbances (mainly IgG) six gave an acceptablecorrelation between automatic and visual counts,the remainder giving totally unacceptable results,the machine results being two to three times higherthan the visual count. One patient with myelo-matosis and an IgG level of 8 g/100 ml gave counts of412 000 on the AutoCounter and 208 000 by phase-

contrast methods. On the other hand, a secondpatient, also suffering from myelomatosis, with anIgG content of 9-6 g, gave counts of 62 500 on themachine and 47 000 by phase-contrast microscopy.Obviously the degree of variation between counts isnot related to the quantity of immunoglobulinpresent. Although the exact cause of this discrepancyremains conjectural, it is possible that coating ofsubthreshold fragments, possibly red cell stroma, bythe abnormal protein may increase the number ofthreshold particles presented to the counting cell.This abnormality is not corrected by the samplingmodification applied to the AutoCounter. Peaksproduced in this situation are often 'off-scale'.Saline dilution experiments give platelet countswhich are not linear, in contradistinction to truethrombocytoses. Instances in which such problemsare encountered include any condition with ab-normally high serum proteins and/or erythrocyteautoantibodies. Non-linearity of diluted samplesshould serve as an indication to check the level ofplasma proteins and to perform a direct antiglobulintest.

Further problems may arise in situations wherethere is a high total white count, numerous nucleatedred cells, Howell-Jolly bodies, or malarial parasites,all of which provide particles countable by thesystem. It should also be remembered that plateletautoantibodies may produce a low platelet count asrecorded by the counter due to platelet agglutination.Unfortunately such cases did not arise during theperiod of the survey, but it is clear that such con-ditions can be detected by careful scrutiny ofperipheral blood films.

In conclusion, this fully automated plateletcounting system, although not fully satisfyingstringent conditions of linearity and carry-over,gives results which are totally acceptable in terms ofservice laboratory platelet counting. An importantadvantage is that the system processes whole bloodsamples without prior manipulation at the rate of 40samples per hour.

We wish to thank the Technicon Corporation forproviding the AutoCounter on which the evaluationwas carried out. We are grateful to Dr S. H. Davies,Department of Haematology, the Royal Infirmary,Edinburgh, for encouragement and advice.

References

Barnard, D. F., Carter, A. B., Crosland-Taylor, P. J., and Stewart,J. W. (1969). An evaluation of the Coulter Model S. J. clin.Path., 22, Suppl. (Coll. Path.), 3, 26-33.

Bull, B. S., Schneiderman, M. A., and Brecher, G. (1965). Plateletcounts with the Coulter counter. Amer. J. clin. Path., 44, 678-688.

Dacie, J. V., and Lewis, S M. (1970). In Practical Haematology, 4th.ed., p. 70. Churchill, London.

225

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/

226

Eggleton, M. J., and Sharpe, A. A. (1963). Platelet counting using theCoulter electronic counter. J. clin. Path., 16, 164-167.

Maupin, B. (1969). In Blood Platelets in Man and Animals, vol. 1,p. 37. Pergamon, Oxford.

Nelson, M. G. (1969). Multichannel continuous flow analysis on theSMA-4/-7A. J. clin. Path., 22, Suppl. (Coll. Path.), 3, 20-25.

Sipe, C. R., and Cronkite, E. P. (1962). Studies on the application of


the Coulter electronic counter in enumeration of platelets.Ann. N. Y. Acad. Sc!., 99, 262-270.

Tocantins, L. M. (1937). Technical methods for the study of bloodplatelets. Arch. Path., 23, 850-879.

Widmark, R. M., and Weschler, W. A. (1970). Evaluation of a fullyautomatic platelet counting technique. Technicon Quarterly,2, 23-28.

The February 1972 IssueTHE FEBRUARY 1972 ISSUE CONTAINS THE FOLLOWING PAPERS

Ischaemic lesions of the alimentary tract G. S. A.MCDONALD AND D. O'B. HOURIHANE

Hilar cell tumour of the ovary M. C. ANDERSON

The diagnosis of trophoblastic tumours fromuterine curettings c. W. ELSTON AND K. D. BAGSHAWE

The structure of the breast in mucoviscidosisA. MILFORD WARD

The value of a study of the mucosubstances in rectalbiopsies from patients with carcinoma of the rectumand lower sigmoid in the diagnosis of premalignantmucosa M. ISABEL FILIPE

Evaluation of the Lundh test in the diagnosis ofpancreatic disease G. E. LEVIN, G. R. YOUNGS, ANDI. A. D. BOUCHIER

Measurements of concentrations of human serumimmunoglobulins

Immunoglobulin levels in infantile pneumocystosisELFRIEDE KOHOUT, CORNELIUS POST, BAHRAM AZADEH,WERNER DUTZ, BASHI BANDARIZADEH, AND DARIUSKADIVAR

Serum hydroxybutyrate dehydrogenase (HBD)assays in the clinical laboratory K. MONTAZEMI AND1. G. LINES

The nature of the alkaline phosphatases of bileC. P. PRICE, P. G. HILL, AND H. G. SAMMONS

Clinical application and mechanization of a simpleand inexpensive method for estimating T3 uptake inserum W. W. WALTHER

Haem biosynthesis studied in patients with rheuma-toid arthritis s. V. JOHANSSON AND P. O. STRANDBERG

A satisfactory quantitative test of lymphocyteresponse to phytohaemagglutinin for the definitionof normal control values and recognition ofimmuno-logical defects MARGARET G. FITZGERALD

Target cells in haemoglobinopathies L. R. DAVIS

Microbiological assay of amino acids in serum:valine, leucine, and methionine J. A. BLACKMOREAND T. E. PARRY

Changes in antibiotic sensitivity patterns of Gram-negative bacilli in bums ELIZABETH ROE ANDE. J. L. LOWBURY

The isolation of an X-dependent strain of Haemo-philus from otitis media identified as H. haemo-globinophilus (canis) JOYCE FRAZER AND K. B. ROGERS

Technical method

The effect of autoclaving characteristics on therecovery of serum vitamin B12 determined by aradioisotope dilution method D. H. ORRELL ANDA. D. CASWELL

Letter to the Editor

Book reviews

Notices

Copies are still available and may be obtained from the PUBLISHING MANAGER,BRITISH MEDICAL ASSOCIATION, TAVISTOCK SQUARE, LONDON, WCIH 9JR, price, £1-05

copyright. on F


http://jcp.bmj.com

/J C


arch 1972. Dow

nloaded from

http://jcp.bmj.com/

evaluation of automatic platelet counting system utilizing whole … · evaluation...

Documents