An intelligent surgical video retrieval for computer vision enhancement in medical diagnosis using deep learning techniques

This paper addresses the challenge of efficiently retrieving surgical videos from large databases for computer vision enhancement in medical diagnosis. The exponential growth of surgical video archives has made manual searching time-consuming and difficult. Therefore, there is a need for intelligent methods that can retrieve relevant videos from such vast databases. In this work, we propose an intelligent surgical video retrieval system that leverages feature optimization and adaptive semantic checking. To achieve accurate similarity checks and address data dimensionality issues, we introduce a modified golden eagle optimization (MGEO) algorithm for feature extraction and optimal feature selection. This step aims to improve the efficiency of the retrieval process by reducing the complexity of the data. Furthermore, we employ a hybrid nonlinear belief neural network (NLBNN) that computes the correlation between medical semantics in both training and testing medical videos. This approach ensures that data with minimal similarity to the query is retrieved, enhancing the accuracy and relevance of the retrieved videos. To evaluate the proposed MGEO-NLBNN framework, we conduct experiments using a benchmark dataset consisting of 20 Meniscus surgery videos. These videos are collected from reputable open-source datasets such as AAOS, MedlinePlus, vjortho, and Northgate. The results obtained from the evaluation demonstrate the effectiveness of the MGEO-NLBNN framework in retrieving relevant surgical videos for computer vision enhancement in medical diagnosis. The proposed approach shows promise in improving the efficiency and accuracy of video retrieval from large surgical video databases, thus facilitating enhanced computer vision applications in the field of medical diagnosis.