Improved Breast Cancer Detection from Ultrasound Images Using YOLOv8 Model

The number of cases of breast cancer has steadily risen during the past few decades. It is one of the major causes of death in women. There is a good chance of recovery if it is identified early. In order to discover and classify breast cancers, a modified version of the YOLOv8 network is employed, which overcomes the drawbacks of earlier models. Publicly accessible data sets of ultrasound breast cancer images (BUSI) are used for the detection. Preprocessing is also done in the initial phase, which comprises image enhancement methods as well as the removal of labels and pectoral muscles. The data set is annotated, augmented and divided into three parts for training (70%), validation (15%), and testing (15%). The simulation is done with the parameters of batch size of 10, learning rate 0.01 and epoch value of 300. This proposed model is compared to YOLOv7 and YOLOv6 and other models for comparative performance analysis. The outcomes demonstrate that YOLOv8 model achieved mAP (mean average precision) 99.5%, recall 98.40%, and accuracy 96.50%. The experimental results show that the proposed approach outperforms state-of-the-art methods on breast ultrasound cancer detection.