Enhancing Acute Lymphoblastic Leukemia Classification and Detection Using Deep Learning
Ajay Kumar, Kumudha Raimond · 2025
Acute Lymphoblastic Leukemia (ALL) is a deadly cancer that can strike at any age and cause uncontrolled growth of immature white blood cells. It affects both children and adults. Leukemic cell classification must be accurate and efficient to improve patient survival. Though generalization and computing efficiency remain an issue, Deep Learning (DL) models have improved medical image categorization. This study compares the performance of five DL classification models in distinguishing abnormal cells from normal cells in the ALL dataset. In addition, this study also compares three Yolo (You Only Look Once) based models for detecting cells in the ALL dataset. The experimental results demonstrate that the Xception model achieves 96.8% accuracy, precision, recall, and F1-score in classifying leukemia white blood cell images. The YOLO11 detection model achieves precision, recall, and mean Average Precision (mAP) scores of 85.3, 94.4, and 96.3 respectively on leukemia cell images proving their localization ability. These optimized DL algorithms in leukemia identification can help doctors make prompt, accurate judgments, increasing the lifetime of patients.