A HYBRID YOLO-CNN MODEL FOR DETECTING AND CLASSIFYING LEUKOCYTES IN LEUKEMIA

Abstract

This study introduces a hybrid approach combining You Only Look Once (YOLO) and Convolutional Neural Networks (CNNs) for the effective detection and classification of leukocytes in leukemia patients. Accurate identification of leukocyte types is critical for diagnosing and monitoring leukemia, yet traditional methods can be labor-intensive and prone to error. Our model leverages the real-time object detection capabilities of YOLO to swiftly locate leukocytes in blood smear images, while a CNN is employed for precise classification of the detected cell types. We trained and validated the model using a comprehensive dataset of annotated blood smears, incorporating data augmentation techniques to enhance robustness and generalization. Experimental results reveal that our hybrid model achieves high accuracy, sensitivity, and specificity in both detection and classification tasks, surpassing existing methodologies. This research highlights the potential of integrating advanced deep learning techniques in hematology, offering a pathway toward automated diagnostic tools that can assist clinicians in delivering timely and accurate care for leukemia patients.