A METHOD AND SYSTEM FOR CLASSIFICATION OF RBC MORPHOLOGY THROUGH INTELLIGENT IMAGE PROCESSING

The present invention proposes a novel methodology and system for the classification of red blood cell (RBC) morphology using image processing techniques. The proposed methodology involves the preprocessing of sample blood smear images, morphological approaches for the removal of unwanted blood cells, boundary extraction using the Freeman chain coding approach, validation against ground truth images, and feature extraction using contour-based methods. The proposed system utilizes a hyper-parameter optimizable KNN (HPKNN) classifier for the categorization of RBC morphologies. The proposed methodology and system have been tested on multiple datasets, including the Chula-PIC-Lab and Erythrocytes IDB1, IDB 2 and IDB 3 datasets, and have shown promising results in the accurate classification of RBC morphologies.