Intellisurg-Net: An AI-Driven Framework for Enhancing Intraoperative Decision-Making and Real-Time Adaptability in Robotic Surgery

Abstract

Robotic-assisted surgical techniques have led to tremendous advancements in terms of the precision and dexterity of the procedures performed minimally invasively. However, existing robotic surgery techniques are highly dependent on the surgeon's input since very little autonomy has been provided in existing systems. Conversely, artificial intelligence (AI) has provided decision-making algorithms for robotic procedures, enabling these surgeries to function autonomously based on specific scenarios. This paper presents a novel approach called IntelliSurg-Net, which incorporates deep learning algorithms and AI in order to enhance the decision-making process during robotic surgery procedures. Convolutional neural network (CNN) algorithms for vision processing, bi-directional LSTM networks for analyzing sequences of events, and reinforcement learning algorithms are implemented in the framework. This framework is evaluated using publicly available datasets such as Cholec80 and EndoVis.