Surgical Tools Detection and Localization using YOLO Models for Minimization of Retained Surgical Items

Abstract

Accurate identification and localization of surgical instruments during procedures are essential for minimizing the risk of retained surgical items. Such retained items result in serious and preventable health complications for patients. This paper investigates the application of YOLO (You Only Look Once) deep learning models for the real-time detection and localization of surgical instruments, aiming to minimize the occurrence of retained surgical items (RSIs) during surgical procedures. With the increasing complexity of surgical environments, the need for robust and efficient object detection systems has become paramount. The methodology involves a comparative analysis of YOLO versions, including YOLOv5, YOLOv8, and YOLOv10, to evaluate their performance in detecting and classifying surgical tools. The models were trained on a specialized dataset that included diverse surgical scenarios, ensuring a comprehensive assessment of their capabilities. Key performance metrics such as precision, recall, and F1 score were utilized to quantify the effectiveness of each model, with results indicating high performance across the board, particularly in distinguishing between various surgical instruments. The findings demonstrate that YOLO models, particularly the latest versions, possess significant potential for Integration into surgical environments, enhancing the tracking and monitoring of surgical tools. The study concludes that implementing these models could improve patient safety by reducing the risk of RSIs, thereby contributing to the advancement of automated systems in healthcare.