Indian Journal of Experimental Biology Vol. 42, October 2004, pp. 955-963

Influence of malaria parasite (P. vivax) on erythrocyte aggregation: A study based on dynamic imaging and analysis

Sanjay Jayavanth & Megha Singh* Biomedical Engineering Division, Indian Institute of Technology, Madras, Chennai 600 036, India

Received 22 January 2004; revised 8 June 2004

Malaria due to P. vivax (PV) is prevalent in many countries. The present work is aimed to determine the cell-cell interaction through formation of aggregates under dynamic conditions. Blood samples are obtained from patients (n=ll) suffering from PV malaria, and the normal subjects (n=lO) in test tubes containing citrate phosphate dextrose (10:1.4), as an anticoagulant. The signature analysis of infected erythrocytes shows significant alterations in their shape and membrane. For aggregation analysis, erythrocyte suspension in plasma at hematocrit 5%, was placed in a glass chamber and mounted vertically on the stage of the video-microscope system. The aggregate images thus acquired show erythrocytes adhering with each other to form mash-like structures. With increase in parasitaemia, the erythrocytes show hyper-aggregation compared to that of normal cells. By processing of the sequence of recorded images during sedimentation, the various aggregation parameters are obtained. These parameters show that the formed aggregates are compact which produce distinct changes in sedimentation pattern with significantly higher sedimentation velocity compared to that in healthy blood samples. These changes in malaria could partly be responsible for alteration in blood flow through microcirculatory system.

Keywords: Malaria, P vivax, Parasitaemia, Erythrocyte aggregation, Video-microscopic images, Shape parameters.

Malaria parasite, Plasmodium Vivax (PV) is neither a bacterium nor a virus but it is a form of unicellular life as sophisticated as the cells of human body. During malaria this parasite invades the erythrocyte and assumes different roles by molecular mimicry of human proteins, thus dodging the human immune system I . Some of the prominent changes due to malaria parasite are also associated with the functional and morphological properties of the erythrocytes2

• This parasite progressively consumes I degrades intracellular protein3

, alters the membrane integrit/, increases the cell stickiness5

, affects the membrane permeabilitl, and reduces regional charge distribution7

, leading to decrease in the RBC deformabilitl. These factors, in combination with adherence of parasitized red blood cells (PRBC) to vascular endothelium and rosetting of PRBC with the non-parasitized RBC (nPRBC)8 could be detrimental to erythrocytes and their vital circulatory function.

Erythrocyte aggregation and deformability contri­bute significantly in maintaining the blood flow

*Correspondent author Phone: + 91-44-22578176 Fax: + 91-44-22570509 E-mail : msingh@iitm.ac.in/msingh_iitm@yahoo.com

through microcirculation. In healthy subjects the aggregation of erythrocytes, a reversible process, contributes in the formation of three-dimensional structure under near stasis conditions in presence of plasma proteins, fibrinogen and globulins9 and is aided by the deformability of erythrocyte lO

• An increase in aggregation due to plasmatic or cellular factors could give rise to sluggish flow and stagnation, and thus contributing to vascular occlusion, cardiac and cerebro-vascular compli­cation9

. Hence precise quantification of aggregation process is of clinical importance both from the point of prognostic value and understanding the disease pathophysiology.

Aggregation of erythrocytes could be quantified by various techniques. The aggregates are visualized by rheoscopic method while measuring the blood viscosity at various shear rates II. Microscopic quantification of this is based on direct observation of the aggregates under stasis conditions and this process is represented in terms of number of aggregates per unit volume l2

. These measurements are carried out either under stasis conditions or by shearing at high shear rate followed by bringing the sample to near stasis conditions. Shiga et.al. 13 obtained the

956 INDIAN J EXP BIOL, OCTOBER 2004

rheoscopic images under stasis or low shear rates conditions and quantified the growth rate of aggregates. Chen et.al. 14 generated the flow of erythrocyte suspension in a glass chamber under low flow conditions and analyzed the variation in size and shape of aggregates. But the formation of three­dimensional aggregate structure requires very low shear conditions, similar to microcirculation. Such a dynamic condition could as well be achieved during the gravitational sedimentation of the aggregates at an appropriate sample hematocrit. This sedimentation approach has been utilized in the present study to overcome the requirements of additional mechanisms to generate flow or shearing. The formation and the movement of the aggregates in a glass micro­chamber, under such conditions, are monitored and recorded by a video-microscopic system. By analysis of the data the parameters, related to erythrocyte aggregation in a sequence of images, are obtained l5

.

The objective of the present work is to apply this procedure to determine various changes in aggregation process of erythrocytes during PV malaria. This analysis may further help in determining the disease severity to implement the treatment measures.

Materials and Methods Experimental technique-The block diagram of

the aggregate imaging system is shown in Fig. 1. To record data, microscope in transmission mode (Leitz Dilux22, Germany) was used. A well-collimated beam of light from a 12 V 100 W tungsten halogen lamp was achieved by using a field diaphragm and an interchangeable condenser. The NPL fluotar objective

Micro scope

Objective

(a)

Video . Camera

of magnification 40x (phase contrast) was used. To achieve the gravitational sedimentation conditions the sample chamber was mounted vertically on the microscope stage, which was tilted horizontally by placing it on a stand. The sample chamber (Fig. Ib) was made of optically flat glass slides of dimensions 45x13xO.lmm. The O.lmm gap between vertical plates allowed the free movement of erythrocytes and their formed aggregates . For viewing the images of aggregates the microscope was fitted with color video camera system (VK 4003, -Philips Newvicon, The Netherlands). The images were recorded by a VHS video-cassette recorder (National NV -370, Japan) on a video cassette at 25fps rate. The recorded images were simultaneously displayed on a monitor (Philips Match-line Video RGB Monitor, The Netherlands).

Sample preparation -Fresh blood samples from healthy subjects and malaria patients were collected by venepuncture in test tubes containing citrate phosphate dextrose (10: 1.4), as an anticoagulant. Healthy subjects selected for the study were without any clinical disease, and their plasma and serum biochemical levels were within normal range. The samples from malaria patients were collected after their blood smear examinations were confirmed positive for the PV. Each sample was centrifuged at 3000 g for 20 min. Thereafter the supernatant plasma was separated and the buffy layer on top of the cells was gently removed and discarded. A suspension of 5% hematocrit was prepared in plasma for dynamic imaging of aggregation process of erythrocytes.

Classification of malaria samples-For this purpose digitized blood smear images were scanned and the number of PRBC was counted, which were

VCR

Digitizer PC

.O.01mm

Thick

Fig. I-Schematic diagrame of (a) aggregation imaging system, (b) sample chamber

JA Y A V ANTH & SINGH: INFLUENCE OF MALARIA PARASITE ON RBC AGGREGATION 957

further verified by microscopic observations of thin smears. The PRBC counts between 1-2, 2-5 and above 5 per 50 fields were classified as low (LP), medium (MP) and high (HP) parasitaemia, respectively. A total of 11 PV malarial samples were collected for this study, 5 samples belonged to the LP, 3 to the MP and 3 to the HP categories.

Data acquisition during erythrocytes aggrega­tion -A well-mixed erythrocyte suspension was placed in the glass chamber and was vertically mounted on microscope stage. The field of view was adjusted at the center both with respect to the height and side-walls of the chamber. The sequential images of sedimenting aggregates were recorded on a videocassette for 30 sec duration after every 2 min interval, till the end of process. These images were digitized with Matrox frame capture card (Matrox Marvel G400-TV) and were stored in the computer for further processing.

Data analysis-The image processing of the aggregation process was automated using Matlab software package for a given sequence of images that were recorded under identical conditions. The preprocessing steps involved illumination correction, video de-interlacing, background subtraction and debluring. Usually the images after digitization had poor or non-uniform illumination, which was corrected by enhancing the image using statistical mean and standard deviation of gray levels in the image l5

. These images were further associated with odd fields, which were removed by video de­interlacing by using a rotational symmetry Gaussian filter I 6. This operation was followed by subtraction of the background from this image, which was recorded separately under exactly identical conditions. The minimization of image degradation due to motion blur, using Lucy-Richard blind deblurring algorithm, was carried out l7

. The restored image was found to be satisfactory after 10 iterations. The post-processing of the image involved edge enhancement, thresholding, median filtering and edge detection I8

.19

•

The edge-detected images were further processed to obtain the region of interest (ROI), using morphological reconstruction method20

. The ROI image thus developed matched the aggregate and cellular areas in the original image. The ROI image was further labeled21 and was used for determining the aggregate parameters. The resulting data were stored in a file. The sequence of above operations carried out

for processing of images of aggregates to obtain various parameters is shown in flow chart in Fig. 2.

AnaLysis of aggregate parameters-The following parameters of aggregates from the processed images were calculated:

(i) Number of aggregates: The number of labeled objects in the image corresponded to the number of aggregates and single cells in the image. The counting of objects was carried out by a row-wise scan based on neighbourhood connectivity, i.e. an image was scanned from top to bottom searching for white pixels. On encountering white pixel, its neighbors and their subsequent neighbours were looked for to check if these were the members (connected to) of the same object. Thus an object was identified and numbered. This process continued till all the disconnected objects were identified.

(ii) Projected area of aggregate (PAA): The area of each aggregate / individual cell was the number of pixels in that labelled object. The projected aggregate area of a given image was simply the sum of the areas of all the objects. Thus the area (Ai) was equal to the sum of all the pixels that belong to ith object, i.e., Ai = L f(x,y), x,y E I

Start ::::> • Initialization:

Set path. folder & subject names. type of image. liP image. O/P image, and O/P data file names. I =1. N = No.of images

Read input Image

+ Preprocessing:

Illumination correction. deinte~acing and deblurring ...

Post-processlng: Background subtraction. thresholding. median fil tering and edge

detection

... ROI Analysis:

Dilation. erosion, image filling. aggregate labeling

• Aggregate Parameter Analysis:

PM, PPA. Aggregates No·s. SDA and ASV

• Write O/P Image and O/P Data

I = I + 1 ~ "- Update liP. O/P Image

Stop ~

Fig. 2-Flow chart of procedures to process images of aggregates

958 INDIAN J EXP BIOL, OCTOBER 2004

(iii) Projected perimeter of aggregate (PPA): For this purpose, the contour of the ROI - filled image was first extracted and labelled in the same manner as above. The perimeter of each aggregate/individual cell was the number of the pixels on the contour of that labelled object. The PP A was the sum of the perimeters of all the objects. The 'Pi', the perimeter of the ith object, in terms of number of pixel was equal to the sum of all the pixels that belonged to this object, i.e., Pi = L f(x,y), x,y E i

(iv) Size distribution of aggregates (SDA): To obtain the size and spatial distribution of all aggregates, based on above data, the projected aggregate size with respect to aggregates number was plotted.

. (v) Sedimentation completion time (SCT): Time duration between the first frame to the last frame with the PAA of the sedimenting aggregates approaching the minimum size of area (10000 ± 1000) pixels.

(vi) Form factor (FF): Calculated as (PPA)2/4 1t

(PAA). (vii) Aggregate sedimentation velocity (ASV):

This was determined by the number of pixels by which an aggregate traverses per sec in vertically downward direction. For this purpose, the RaI images were analyzed to obtain the centroid of the aggregates in the image. The vertical distance of the centroid point of a given aggregate in successive frames (time duration - 40 ms) was measured. The mean sedimentation velocities of various aggregates, from beginning till the end of the process, were calculated.

Analysis of srrzear images-To show the shape variation of the normal and malaria PRBC, the smear images were first edge-enhanced and thresholded by Otsu's method l9

, by using an interactive gray level threshold followed by median filtering. The edge detection of the filtered image was carried out by Canny's edge detection scheme17

• From these processed images the shape signatures were obtained as a plot of normalized radial distance of contour pixel from the centroid of the cell, versus the angular position of the pixel22. To obtain the signatures a separate program was developed. The edge-detected smear image was the input to this program. The pixel coordinates of the edge and the centroid pixels were

generated and using this information the radial distance of the contour pixel from the centroid was calculated and shape signature plot was obtained. The signature, being a 1-0 line representation of the cell surface, provided the information of the membrane changes due to parasitaemia in erythrocytes.

Results The images of normal and PRBC, and their shape

signatures are shown in Fig. 3. The normal cell is more or less similar to that of an elongated circle and its corresponding signature is close to a straight line (unity line in the plots). Due to the indwelling parasite the shape of PRBC is deviated when compared to the normal RBe. The deviation increases with the increase of parasitaemia. As shown, this shape deviation is the maximum in case of PRBC of HP malaria. These cells are enlarged and the enlargement is pronounced in mature trophozoite stage of PV. The shape and size changes are more significant in HP compared to that of normal cells.

The sequence of images after implementing different steps of image processing procedure on a raw digitized image of aggregates under dynamic conditions are shown in Fig. 4. After the implementation of illumination correction and deinterlacing steps the processed original image (a) still shows a constant noisy background. This is effectively removed after background subtraction as seen in background-subtracted image (b). The deblurred image (c) shows edges of the cells and aggregates in better focus . The aggregates can be clearly delineated from the rest of the image after thresholding as shown in image (d) . The edge­detected image (e) shows many discontinuities, which depend on the image quality and the threshold chosen for edge detection. In order to get continuous edge, morphological contour structuring is employed wherein the edges are subjected to a deliberate over­dilation by a known size of line structuring element. Hence in the dilated image (f) the discontinuities in the edge are effectively removed, but the ROI is blown up because of dilation. In order to recover the original ROI, the dilated image is eroded by the same size of the line-structuring element, employed for

. dilation. The eroded image (g) shows that the edge is continuous and the original RaI is restored. The RaI image is further filled with white pixels for determination of the PAA (h). To establish that the region of interest is successfully extracted from the original image. the final ROI image (h) is masked on

JA YAVANTH & SINGH: INFLUENCE OF MALARIA PARASITE ON RBC AGGREGATION 959

the deblurred image (c) to get the masked image (i), which shows that the area of interest is successfully extracted from the original image.

The examples of images of the formed aggregate of normal, LP, MP and HP samples are shown in Fig. 5. The normal aggregate (A) show a neatly arranged stack of cells forming a chain-like structures. The individual cells have deformed from the biconcave shape to facilitate maximum contact area and hence forming stable aggregates. The LP aggregate (B), shows deviation from the chain-like structures seen in the normal and this deviation is more pronounced in MP and HP. In MP (C) the cells form more compact clumps showing a greater degree of adherence. In HP (D) the formed aggregates show that the cell are speculed due to protrusions associated with irregular

E g 1

a> .~ a. S 0 o C 2 o u "-o Q) u c ro ~ "0

~ 0 ro L..2 "0 Q)

.t::! ro E o 1 z

o

membrane similar to those seen in PRBCs of Fig. 3. The cells are not deforming but still adhering with each other to form mash-like structures. With increase in parasitaemia, the erythrocytes exhibit hyper­aggregation and form irregular and compact aggregates compared to that of normal cells. The erythrocytes are seen to be participating in aggregation with little or no deformation. The PRBC in combination with nPRBC form clumps without deforming, attributed to the enhanced intra-cellular adhesion in malaria.

Figure 6 shows the projected areas of aggregates (PAA) in successive frames captured at intervals of 2 min. The PAA progressively decreases due to sedimentation of erythrocytes and their formed aggregates for both the normal and malaria-infested

r---.-----------I MP

HP

'---1-_-'-__ . __ L._--'--_--'-_-'--_--'---. ! I

'----J

o 40 80 120 160 200 240 280 320 360 Angular position of contour pixel (degrees)

Fig. 3--Shape-signatures of the normal. low (LP). medium (MP). and high parasitaemia (HP) erythrocytes

960 INDIAN J EXP mOL, OCTOBER 2004

subjects. The PAA for malarial subjects for all categories (LP, MP and HP) is initially higher and subsequently reduces rapidly when compared to normal. This shows that in malaria aggregates are of compact sizes. In the HP the sedimentation process is faster compared to normal, and LP and MP malaria aggregates. The sedimentation velocity of aggregates in the HP is faster compared to that of normal, and LP and MP malaria.

The projected perimeters of aggregates (PPA) show that the shape of the formed aggregates varies throughout the sedimentation process as observed in successive frames (Fig. 7). The PPA of normal subjects show large variation compared to that of malaria patients of different parasitaemia. The

., , .

f / Jr-. . '" .' ~ .. k;~

~ ..

~ . . ~~ ~ -n.>'~" :"'il"'Alo

~71 .. ·

overlapping regions further indicate some commonality of the formed aggregates.

For statistical analysis, the PAA, PPA, FF (based on images at 1 t h min), SeT and ASV of normal and parasitized samples were compared by Student's t test (Table O. These parameters for parasitized samples showed a significant difference when compared with that of normal subjects. Among the parasitized samples, the ASV increased with increase in parasitaemia from low to high and SeT decreased correspondingly. ASV and SeT are complementary to each other; e.g., increase in the aggregate sedimentation velocity lead to faster completion of sedimentation process. These changes were attributed to the formation of larger but compact size aggregates

.-- ,; ~,' . -,. . '':''

'., ~ .w

(c) Deblurred image

(d) Thresholded image (e) Edge image (f) Dilated image

(g) Eroded image (h) ROI image (i) Mask of ROI and deblurred image

Fig. 4--Processed images of the aggregates of normal subject after applying different processing procedures to the digitized images

JA Y A V ANTH & SINGH: INFLUENCE OF MALARIA PARASITE ON RBC AGGREGA nON 961

in malaria. The form factor is normally an indicator of deviation from a circular shape, which shows significant deviation in parasitaemia compared to that of aggregate of normal erythrocytes.

Discussion The present study was aimed at quantifying the

changes in RBC aggregation in healthy and malarial blood samples of PV species. The imaging and image analysis techniques proposed in the study quantify the aggregation process in terms of" various parameters and demonstrated the difference in participation of normal and PRBC in the aggregation process. This methodology is simple and unique, as it does not use additional flow mechanisms, which are essential in the other imaging techniques to generate flow ll

-14

.

The requirement of the low flow conditions essential for this study was achieved by employing the gravitational sedimentation technique.

The present results clearly show the changes in RBC aggregation in malaria of varying degree of parasitaemia. The results indicate that with increasing parasitaemia, the erythrocytes tend to adhere with each other forming clumps with mInImum deformation as seen in aggregate images. This could be attributed to the enhanced cell adhesion of erythrocytes23 and change in the surface charge on RBC membranes. The decrease in RBC surface charge enhances the tendency of aggregation 13, which may be responsible for the formation of larger

B

Fig. 5--Aggregate images for normal (A) and P. vivax malarial sample of low parasitaemia (B). medium parasitaemia (C). and high parasitaemia (D)

aggregates initially as indicated by the PAA of malarial samples.

The erythrocytes participating in aggregation could be seen jagged, due to irregular outer membrane2

.

Normally in P. vivax malaria the shape and size of

........ (/)

Q5 x

c::: 0 0 0 ....... x -........ -

~ -Q) -co -O) Q) .... 0) 0) co -0 co Q) .... co -0 Q) -u Q) .0' .... 0. e co : Q)

~

80

60

40

20

0

80

60

40

20

0

80

60

40

20

0

80

60

Normal

LP

MP

Ii HP

o 4 8 12 16 20 24 28 32 36 40

TIme (min)

Fig. 6--Mean projected area of aggregate (PAA) for normal. low (LP). medium (MP) and high parasitaemia (MP) malaria with standard deviation error bars

962 INDIAN J EXP BIOL, OCTOBER 2004

Table I-Comparison of aggregation parameters, viz. projected area of aggregate (PAA), projected perimete.r of aggregate (PPA), form factor (FF), aggregate sedimentation velocity (ASY) and sedimentation

completion time (SCT) for different categories of malarial subjects with control (Normal). [Yalues are mean ± SO. Number of samples are given in parentheses]

PAA (pixel) PPA (pixel)

after 12 min FF SCT (min) ASY (pixel/s)

Normal (5) 44287.8 ± 3046.8 Low Parasitaemia (5) 37155.3 ± 4201.1 * Medium Parasitaemia (3) 38322.1 ± 1645.8* High Parasitaemia (3) 38774.0 ± 1351.3*

5678 ± 152.5 3987 ± 347.0** 4268±417.2** 4499 ± 508.7**

58.2 ± 5.1 35.l± 10.3** 37.9 ± 6.2** 43.2 ± 8.5*

39.4 ± 2.5 18.8 ± 3.9 32.6 ± 1.5* 24.5 ± 1.4*

26.3 ± 1.5** 27.6 ± 2.2* 19.5 ± 0.7** 30.4 ± 0.4**

P values: * S 0.0 I ; ** < 0.005;

8,----------------------------,

7 YJ 6 a> x '0. 5 o 8 4 ..... ~ 3

~ 2 a... 1

~Normal

~LP

~MP

O+O-.~._~~_r~~~._~~_rl

o 4 8 12 16 20 24 28 32 36 40 Time (min)

Fig. 7-Mean projected perimeter of aggregates of normal, and low (LP), medium (MP) and high parasitaemia (HP) samples

PRBC's, due to indwelling parasite and its growth into multi-nucleated trophozoite, are changed24

• The signatures of these erythrocytes further support these changes. Such shape changes could contribute in the pathology of malaria.

For assessment of the disease severity, the malarial subjects were classified into three types of parasitaemia. The present results show that the aggregation parameters, particularly ASV and SCT are significantly altered in the HP when compared to that of normal, and LP and MP malaria, thus further supporting the enhanced aggregation in malaria. As the changes in hemorheological parameters for a given disease are considered to be indicators of the disease severity25, these may be chosen for description of the acuteness of the disease to supplement the parasitaemia based classification. The hyper­aggregation tendency leads to formation of larger and stronger aggregates that may not dis aggregate, particularly in diseased conditions. Recently such a

tendency has been identified as an important factor for disturbing the rheological behaviour of blood in microcirculation26

•27

, . which may initiate other cardiovascular complications9

•

Erythrocytes infested with P. vivax are characterized by Schuffner's dots-small brick-red dots leading to formation of caveola-vesicle complexes and clefts, which traffic the parasite antigen to the erythrocyte surface membrane4 leading to changes In their signatures. The present observations further show that the Increase in adhesiveness due to this antigen, even at normal levels of plasma proteins, is a contributing factor in the aggregation process, which increases with the increase of the disease severity. The enhanced aggregation, associated with clumping may be responsible for life-threatening complications2

• Acute infection of PV is generally associated with anemia, low birth-weight babies and respiratory pathogenesis I. As per this study hyper-aggregation tendencies are seen in PV malaria, which may also be a contributing factor for such anemic conditions, thus highlighting the need to look at aggregation fof prevention strategies to tackle the acute phases of the disease.

In conclusion, despite the decrease in the RBC deformability on account of cytoplasm and membrane factors28

, the erythrocyte aggregation in malaria was increased. The presence of relatively strongly bounded aggregates associated with enhanced molecule expression on the vascular endothelium in malaria may lead to vascular obstruction, complicating the blood flow in microcirculation. The morphological changes in formed aggregates and associated parameters, along with erythrocyte deformability, could be used to .assess the clinical condition of the patients and for their therapeutic measures.

JAYAVANTH & SINGH: INFLUENCE OF MALARIA PARASITE ON RBC AGGREGATION 963

Acknowledgement Authors gratefully acknowledge the valuable

guidance and support provided by Dr. K. Jagadeesan and Ms C. Abraham, K. J. Hospital, Chennai, for clinical analysis, and Dr. G.S. Rajarathinam and Mr. B. Dhanraj, Health Department (Malaria), and Corporation of Chennai, Chennai, India, for collection of malaria blood samples.

References 1 Chaudhury R R, A closer look at Malaria in India, Nature

Med, 28 (1966) 833. 2 Aikawa M, Morphological changes in erythrocytes induced

by malarial parasite, Biology Cell, 64 (1988) 173. 3 Mellville D, Paul F & Routh S, Direct magnetic separation of

red cells from whole blood, IEEE TrailS. Magnetics, 11 (1975) 1701.

4 Yuthawong y, Alterations of erythrocyte membrane in malarial infection, 1. Sci. Soc. Thailand, 11 (1985) 53.

5 Howard R J & Gilladoga A D, Molecular study related to pathogenesis of cerebral malaria. Blood, 74 (1989) 2603.

6 Nakoranchai S, Sathithudsahakorn C, Chonggchirasiri S I, Yuthwong, Mechanism of enhanced fusion capacity of mouse red cells infested with plasmodium berghei. 1. Cell. Sci., 63 (1883) 147.

7 Sherman I W & Greenan R T, Plasmodium falciparum: Regional differences in the lectin and cationized ferritin binding to the surface of the malarial-infected human erythrocytes, Parasitology,. 93 (1986) 17.

8 Nash G B, Brien E 0, Gordon-Smith E C & Dormandy J A, Abnormalities in the mechanical properties of red blood cells caused by plasmodium falciparum, Blood, 74 (1989) 855.

9 Stoltz J F, Singh M & Riha P, Hemorheology ill practice, (lOS Press, Amsterdam) 1999.

10 Skalak R, Aggregation and disaggregation of red blood cells, Biorheology, 23 (1984) 463.

11 Schmid-Schoenbein H, Wells R & Schild kraut R,. Microscopic and viscometry of blood flowing under uniform shear rate (rheoscope). 1. Appl. Physiol,. 26 (1969) 674.

12 Chien S, Usami S, Dellenback R J, Gregerson M I, Nanninga L B & Guest M M, Blood viscosity: Influence of erythrocyte aggregation. Science, 157 (1967) 829.

13 Shiga T & Maeda N, Kinetic study on aggregation of red blood cells by successive analysis of moving video images, in Hemorheologie e/ aggregation erythrocytaire, Vol. 3

edited by J F, Stoltz M Donner & A L Copley (E MInter. Paris) (1991) 38.

14 Chen S, Barshtein G, Gavish B, Mahler Y & Yedgar S, Monitoring of red blood cell aggregability in a flow chamber by computerized image analysis. Clin. Hemorheol., 14 (1994) 497.

15 Sanjay J, Singh M & Jagadeesan K, Sequelllial imaging and analysis oj erythrocyte aggregation, in Proc. 2nd Annual Intern. IEEE-EMBS Conference on Micro Technologies in Medicine and Biology, Madison ( 2002) 458.

16 Gonzalez R C & Wood R E, Digital image processillg, (Addison- Wesley, New York) 1999.

17 Lucy L B, An iterative approach for the rectification of observed image, Astronom. 1, 79 (1974) 8.

18 Canny J A, A computational approach to edge detection. IEEE TrailS Pattern Allaly. Machine In/eli., 8 (1986) 679. .

19 Otsu N A, Threshold selection method from gray-level histograms, IEEE Trans. System Man Cybernet. 9 (1979) 62.

20 Soil Ie P, Morphological image analysis: Principles alld applications (Springer-Verlag, Berlin) 1999.

21 Haralick R M & Shapiro L G, Computer and robot visioll. Vol. I, (Addison-Wesley, New York) 1992.

22 Sanjay J & Singh M, Computerized shape descriptors Jor identificatioll oj changes in normal and hyper­cholesterolemic erythrocytes in Proc. Internl. Conf. Biomed. Eng., Chennai, India (2001) 290.

23 Ray A P, Narasimham M V V L & Kalra N L, A brochure Jor malaria medical officers, (Indian Society for Malaria and Other Communicable Diseases, New Delhi) 1992.

24 Nash G B, Brian M, Cook, Marsh K, Rerendt A, Newbold C & Stuart J, Rheological analysis of the adhesive interactions of red blood cells parasitized by Plasmodium Jalciparum, 1 Blood, 79 (1992) 798.

25 Puniyani R, Clinical hemorheology: New horizons, (New Age International, New Delhi) 1996.

26 Abugo 0 0, Pedada R R, Kelly J F, Roth G S & Rifkind J M, Effect of cholesterol content in diet on capillary flow of rat erythrocytes. Part I. Geometric and flow characteristics. Clin Hemorheol. Micircul, 17 (1997) 437.

27 Alexandratou E, Yova D & Cokkinos 0 V, Morphometric characteristics of red blood cells as diagnostic factors for coronary artery disease., Clill. Hemorheol. Micircul . 21 (1999) 383.

28 Glenister F K, Coppel R L, Cowman A F, Mohandas N & Cooke B M, Contribution of parasite proteins to altered mechanical properties of malaria-infected red blood cells., Blood, 99 (2002) 1060.