Blood corpuscles classification schemes for automated diagnosis of hepatitis

The paper proposes three methods for automated diagnosis of hepatitis based on shape recognition and classification of blood corpuscles. The processed data are binary images of blood samples taken from positive diagnostic and negative diagnostic patients. The idea followed in the paper is to identify T-lymphocyte response (Danne corpuscles) using approximations of shape coefficients as perimeter, area and circularity. The automatic diagnosis is accomplished through an analysis process on the class of the most circular corpuscles identified in the image. A series of experimentally derived conclusions are supplied in the final part of the paper. 2000 Mathematics Subject Classification: Medical applications (general) ( 92C50), Pattern Recognition (68T10), Image processing (68U10).