AUTOMATIC COUNTING OF LEUKOCYTES IN GIEMSA-STAINED IMAGES OF PERIPHERAL BLOOD SMEAR

Submitted by
Seena Sreedhar R
S7 AEI
College Of Engineering, Trivandrum
2007-11 batch

[attachment=6969]
Hi guys, if you like this presentation, say hi to the user at seenu.sr.chinju[at]gmail.com

Overview
Introduction
Segmentation
Histogram Analysis
Selection of Threshold Points
Measuring of Distances
Experimental Results
Conclusion


   Introduction
Examination of Peripheral Blood Smear.
Counting of leukocytes in Giemsa-stained images.
Leukocyte count is used to determine the presence of an infection in the human body.
Here they used histogram of images and intensity of red cells which are major objects in images to select appropriate point for thresholding.
Blood Cells
Red Cells (erythrocyte)
Blood Platelets
White Blood Cells (leukocyte)
1) Neutrophil
2) Eosinophil
3) Basophil
4) Monocyte
5) Lymphocyte
Blood Cells

Methods for Complete Blood Count

Two types

Manual

Automated

Manual (Spectrometry)
Used for determining hemoglobin concentration in whole blood.
The instrument used is spectrophonometer.
This measures monochromatic light transmitted through a solution to determine the concentration of the light absorbing substance in that solution.
Automated

Two types
1. For determining hemoglobin concentration in whole blood.
CELL-DYN 3200

2. Counting different blood cells ( WBC, RBC, Platelets)
Segmentation
Cell segmentation is the process of identifying, then extracting cells from background. Three major categories are:
Boundary based
Region based
Thresholding

Histogram Analysis
An image histogram is a chart that shows the distribution of intensities in an indexed/intensity image.
Used to enhance the contrast between cells and the background.
Choose an appropriate point for thresholding.
For this, images must be stained and Giemsa-stain is used.


Measuring of distance
In neutrophils the nucleus is frequently multilobed.
After thresholding merge these segmented nucleus.
Distances among nuclei have been calculated.
Merge the nuclei which those distances are less than the diameter of one leukocyte.


Operators Used
Erosion



Dilation




Experimental Results
The image data set contains 30 microscopic images of blood smear.
Images are taken by an electronic microscope with digital camera.
The accuracy of this method is nearly 96.7%.
The resolution of images is 600×473 pixels.
Advantages
In labs hematologists analyze blood by microscope, it is tedious to locate and count cells. Thus this process is very helpful and necessary as it is easy and takes less time.
Histogram analysis used in this paper is robust to differences in staining.
Effective and reliable as compared to other conventional methods.
Higher accuracy and better resolution of images.

Conclusion
Proposed a new detection algorithm based on histogram analysis.
Measurement of distance among nuclei.
Can detect almost all WBC in Giemsa-stained images of peripheral blood smear.
Reference
[1] Saif Zahir, Rejaul Chowdhury, and Geoffrey W.Payne, “Automated
Assessment of Erythrocyte Disorders Using Artificial Neural
Network”, IEEE International Symposium on Signal Processing and
Information Technology, 2006.
[2] Silvia Halim, Timo R. Bretschneider, Yikun Li, Peter R. Preiser and
Claudia Kuss, “Estimating Malaria Parasitaemia from Blood Smear
Images, IEEE ICARCV 2006.
[3] Refai, H., Li, L., Teague, T.K., and Naukam, R., “Automatic count of
hepatocytes in microscopic images,” Proceedings of the International
Conference on Image Processing, 2, pp. 1101–1104, September 2003.
[4] FANG Yi, ZHENG Chongxun, PAN Chen and LIU Li, “White Blood
Cell Image Segmentation Using On-line Trained Neural Network”,
Proceedings of the IEEE Engineering in Medicine and Biology 27th
Annual Conference Shanghai, China, September 1-4, 2005.
[5] C.Ruberto, A.Dempster, S.Khan and B.Jarra, "Analysis of blood cell
images using morphological operators," Image and Vision Computing.,
vol.20, pp.133-146, 2002.
[6] Q.Liano, and Y.Deng, "An Accurate Segmentation Method for White
Blood Cell Images," IEEE Conf. Biomedical Imaging, pp.245-248,
2002.
[7] N. Otsu, “A threshold selection method from graylevel histograms,”
IEEE Transactions on Systems, Man, and Cybernetics 9(1), pp. 62–66,
1979.
[8] K. Wu, D. Gauthier, and M. Levine, “Live cell image segmentation,”
IEEE Transactions on Biomedical Engineering 42(1), pp. 1–12, 1995.
[9] T. Markiewicz, S. Osowski, L. Moszczyski, and R. Satat1,
“Myelogenous leukemia cell image preprocessing for feature
generation,” in 5th International Workshop on Computational Methods
in Electrical Engineering, pp. 70–73, 2003.