REAL-TIME SURGICAL TOOL DETECTION IN MINIMALLY INVASIVE SURGERY USING CONVOLUTIONAL NEURAL NETWORK

Abstract

The sugical tool detection is identifies the surgical tool category and locates the
position using a bounding box for every known tool within an image or video. This
is a significant issue in the use of computer vision to increase laparoscopic surgery
efficacy. This paper presents a real-time surgical tool detection algorithm based on
convolutional neural networks (CNNs). The dataset for this research was derived
from cholecystectomy surgical videos. The experimental results show that the
algorithm can operate in real-time at a frame rate of 25.4 (fps), with a mean
average precision (mAP) of 71.54% over our dataset.