Intelligent System for Detecting Medical Instruments Used in Surgery

The system proposed in this paper aims to assist new nurses with the accurate and timely identification of medical instruments used in the operating room or in the central sterile supply department. An important aspect of surgical intervention flow is the efficient communication between the operators and the rest of the medical staff to ensure the proper procedure times. That also means the positioning in order of the medical instruments on the Mayo trays, which is often a problem for new nurses. In those situations, quick identification is needed; otherwise, stressful situations within the medical team can be induced. Our approach implements a new method for rapid detection of the instruments in real-time using a low-cost video camera and development motherboard, which embeds two machine learning algorithms, MobileNetv2 and YOLOv5, by Transfer Learning. The results show that high identification accuracies can be obtained. We also emphasize that many operating rooms that are equipped with adjustable lights and operating cameras with such solutions can also be implemented as an automatic step for surgery preparation. Moreover, this can also avoid possible contamination between sterile and non-sterile environments.