Computer Assisted Classification Framework for Detection of Acute Myeloid Leukemia in Peripheral Blood Smear Images

Abstract

Acute Myeloid Leukemia (AML) affects myeloma cells in human blood. The manual detection of AML cells in peripheral blood smear image is a difficult task and also needs more time. The computer assisted classification framework for AML cell is proposed in this work. AML causes changes in nu- clues of myeloma cells. The microscopic images of myeloid cell are obtained from online database. Pre-processing and segmentation process are performed to obtain nucleus and cell mask. The shape of nucleus is irregular and its texture is also changed when the human is affected by acute myeloid leukemia. Separation of nucleus and cytoplasm is achieved through k-means clustering algorithm. The morphological feature vectors are used as input for classifiers. Classification framework is done by using different machine learning classifiers and found that Random forest classifier gives an accuracy of 95.89%.