BLOOD CANCER DETECTION USING ENSEMBLE TECHNIQUE

Abstract

In this paper, an ensemble strategy integrating Convolutional Neural Networks (CNN) and Support Vector Machines (SVM) is presented as a novel method of blood cancer detection. The study makes use of a dataset of 3,242 blood cell pictures that have been categorized into four different groups: Benign (512 photos), Malignant_early Pre-B (979 images), Malignant_Pre-B (955 images), and Malignant_Pro-B (796 images). To guarantee robust learning and validation, the ensemble model has been trained using a significant chunk of the dataset using an 80:20 training and testing split. Using CNNs for feature extraction allows deep learning algorithms to identify complex patterns and traits in blood cells that correspond to different stages of cancer. Following the extraction of these characteristics, an SVM is used to accurately classify the pictures. The hierarchical feature learning capabilities of CNNs and the high dimensional classification efficacy of SVMs are leveraged by the ensemble model. The model performs quite well, with 97% training accuracy and 96% testing accuracy. These findings demonstrate how well the model can categorize various blood cancer cell types, which makes it a useful tool for early diagnosis and detection. The results of this study have important ramifications as they indicate that the CNN+SVM ensemble technique combination can be a potent method for medical diagnostics that provides accurate and effective blood cancer diagnosis. The goal of future study is to improve early diagnosis processes and individualized treatment regimens for individuals with blood cancer by honing the model's prediction skills and investigating its use in actual clinical settings.