Acute Lymphoblastic Leukemia Classification: Deep Learning Techniques for Blood Diseases Diagnosis

Abstract

The most common types of blood cancer is Acute Lymphoblastic Leukemia. The procedures used to treat it is very costly and time taking. Images from peripheral blood smears serve as an early detection of acute lymphoblastic leukemia (ALL) disease for the blood sample. The manual collection of PBS images for the diagnosis of cancer contains some errors due to certain factors such as interoperability errors and human fatigue. Advanced techniques have surpassed handmade and conventional approaches for the solution of classification of images. In this paper, tuned EfficientNetB3 model used to classify ALL with its subtypes, is considered for the experiments. The model is developed using the dataset that is publicly available on Kaggle. It noticed that the observed performance through the classification on EfficientNetB3 model exceeds expectations, demonstrating an accuracy of 99.84%. One could argue that the proposed approach may assist in differentiating among various classifications of ALL and in establishing the appropriate diagnostic procedures for healthcare professionals in laboratory settings.