Abstract—Ovarian cancer is the 7th most common malignant tumor and the 8th leading cause of death in women. Therefore, ovarian cancer detection early and on the image data as ultrasound images is an issue that needs to be studied. YOLO is a highly accurate CNN, especially with very low processing time, and can calculate on the CPU for object detection problems in computer vision. In this paper, we perform a comparative study on the latest version of the YOLO family, YOLOv8 (YOLOv8n, YOLOv8s, YOLOv8m, YOLOv8l, and YOLOv8x) for the detection and classification of ovarian tumors on the OTU 2D-OS. This study performed a fine-tuned model to detect and classify the OTU 2D-OS: (1) 1 label (with or without ovarian tumor); (2) 8 labels; (3) 2 labels (benign ovarian tumor and malignant). The precision of YOLOv8X is the best and higher than YOLOv7 is 19% for detecting and classifying 8 ovarian tumor classes on the OTU 2D-OS subset. The calculation time of YOLOv8 is also shown, and the processing time of YOLOv8x is slower than YOLOv7 (YOLOv8x is 186fps on GPU, 1.84fps on CPU). However, these results are still low compared to the requirements for the actual diagnosis and detection of ovarian tumors. It can be accepted that “it is better to catch a mistake than to miss it”. Therefore, the problem of detecting ovarian tumors on ultrasound images brings many challenges and needs further research.
 
Keywords—ovarian tumor detection, ovarian tumor classification, benign and malignant, YOLOv8, ultrasound images, Convolutional Neural Networks (CNNs)


Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License (CC BY-NC-ND 4.0), which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.