Automatic count of hepatocytes in microscopic images

This paper describes a part of current research work on counting dead and live hepatocytes (liver cells) in cultures from microscopic images. The requirement of the work is to develop an automatic cell counting process that is simple, fast, and achieves high level of count accuracy. Cells in the acquired images are difficult to identify due to low contrast, uneven illumination, gray intensity variations within a cell, irregular cell shapes. For automatic counting, our cell images undergo three-stage image processing: conditioning, segmentation, and mathematical morphology operations. Local adaptive thresholding technique is employed in the segmentation stage. At the end of the morphological process, cells are identified and counted based on size. Compared to a manual cell count, the automatic count has achieved an on average accuracy of 95% for single cell counting and 85% for total cell counting.