Detection and Classification of Leukemia And Myeloma Using Soft Computing Techniques

Abstract

Abstract: Blood cancer is a type of cancer that affects the blood cells. Medical image processing technology is essential in both early disease identification and cancer cell analysis. Blood cancer can impact the lymph nodes, bone marrow, blood cells, lymph nodes, and other lymphatic system components. A primary cause of blood cancer is an unusual and excessive amount of white blood cellular proliferation. Traditional cancer cell detection is time-consuming and inaccurate to a large extent, hence an automated approach based on soft computing techniques is presented to predict cancer cell presence and identify two types of blood cancer which are leukemia and myeloma. The dataset for Myeloma is acquired from TCIA (The Cancer Imaging Archive) repository and the Leukemia dataset comes from Kaggle-Blood Cell Images, both of which are available to the public. The datasets are already pre-processed. In our study, we have compared different hybrid models like DenseNet with XGBoost, InceptionResNet with SVM, etc from which the combination of VGG-19 for feature selection and SVM for classification gives the best performance. We have achieved Classification Accuracy of 96.4%(0.964), Precision(0.964), F1 Score(0.964) and Recall(0.964) for SVM.