An IoMT-Enabled Surgical Monitoring System Utilizing Robotics and AI With E2ARiA-RESNET-50 and MI-KMEANS

Abstract

Robotic automated Surgery uses robots to assist with surgeries, making procedures more precise and recovery faster. It is popular in healthcare because it enables surgeries with smaller incisions, leading to quicker healing and shorter hospital stays. However, existing research often neglects the implementation of strong safety measures and fail-safes in robotic surgical systems. Therefore, this paper presents a robotic-based AI framework for monitoring the surgical phase, utilizing E2ARiA-RESNET-50 AND MI-KMEANS. Initially, the input video is preprocessed, including frame conversion, key frame extraction, blur and distortion removal using AKRDF with sharpening. Next, data are balanced using SMOTE. Super-resolution is then performed using PWLC-SRGAN, followed by variability analysis in tissue appearance using MI-KMEANS and patch extraction. In the meantime, from super-resolution, segmentation is done by ROI-WA, followed by masking. Then, features are extracted from both patch-extracted and masked images. Finally, these extracted features are classified using E2ARiA-RESNET-50 for monitoring. The experimental results revealed that the proposed model reached a high accuracy of 98.625%, outperforming traditional methods.