Deep Learning-Based Automated Detection and Classification of B-cell Acute Lymphoblastic Leukemia
Aahash Kamble, Utkarsha Sumedh Pacharaney, Nusrat Parveen, Shamim Akhtar, Gouri Morankar · 2024
This study uses cutting-edge deep learning approaches to accurately classify peripheral blood smear (PBS) images of acute lymphoblastic leukemia (ALL). We present a convolutional neural network (CNN) model that uses the Blood Cells Cancer (ALL) dataset, which includes of 3242 PBS images from 89 patients, based on the EfficientNetB0 architecture. The program aims to distinguish between cases of benign and malignant leukemia, including specific subtypes, by processing and analyzing high-resolution blood smear images. Our approach involves a thorough data preparation process that includes image scaling, normalization, and augmentation. The CNN model achieved an impressive 99.54% test accuracy, demonstrating its effectiveness in correctly identifying leukemia. The findings demonstrate how deep learning models may increase the accuracy of leukemia diagnoses, which is crucial for timely and effective patient care.