The Classification of Abnormal Red Blood Cell on The Minor Thalassemia Case Using Artificial Neural Network and Convolutional Neural Network

Abstract

The morphological disorder of the red blood cell is one of the indications of a certain type of diseases. On the minor thalassemia, such cases like the erythrocyte having a nucleus, a few number of the fragment cell and the target cell will be seen. This research study aimed at classifying four types of abnormal blood based on the shape, texture, and colour which was obtained from the image of the peripheral blood smear. The preprocessing stage using histogram equalization, segmentation stage using morphological operation, until feature extraction had been done. On the classification stage, the best accuracy to classify the blood into five types using algorithm of momentum backpropagation neural network was 93.22%, while the result of classification using the convolutional neural network (CNN) was 92.55%.