Detection and Sub-Classification of Acute Lymphoblastic Leukemia Cell Types from the Microscopic Images Based on The Object Detection Model YOLOV5

Acute lymphoblastic leukemia (ALL), a disease that is quite common, necessitates invasive, costly, and time consuming diagnostic procedures for a final diagnosis. The early classification of cancer cases from noncancerous samples plays a crucial role in ALL detection utilizing peripheral blood smear (PBS) images. This research highlights the successful application of the YOLOv5 object detection model in detecting ALL cells and classify it into four distinct types, achieving 97.8%,96%,96.7%, accuracy, precision, and recall respectively using an ALL image dataset after adding the bounding boxes manually. Moreover, a comparison is conducted among three popular convolution neural network (CNN) models, namely GoogleNet, ResNet, and AlexNet, alongside YOLOv5 using the ALL dataset. The results clearly demonstrate that YOLOv5 model outperforms the other CNN models, further confirming its superiority in ALL detection.