Computer-assisted Acute Lymphoblastic Leukemia detection and diagnosis

Leukemia is a cancer of white blood cells (WBCs) which damages blood and bone marrow of human body. It can be fatal disease if not diagnose at earlier stage. Generally complete blood count (CBC) or morphological image analysis is used to manually diagnose the leukemia cells. These methods are time consuming and less accurate which needs to be fixed. In this paper we have proposed an automated technique for the detection of acute lymphoblastic leukemia by microscopic blood image analysis. This approach first segment out the different types of cells from the image i-e. white blood cells, red blood cells and platelets. After that Lymphocytes are separated from the white blood cells. Then shape and color features are extracted from these lymphocytes which are given to SVM classifier to classify the cells into normal and blast. We have achieved an overall accuracy of 93.7%, senstivity of 92% and specificity of 91% by testing our proposed algorithm over the ALL-IDB-1 dataset. This automated leukemia detection system found to be more effective, fast and accurate as compare to manual diagnosing methods.