An Efficient Deep Learning Model for Content Based Medical Image Retrieval

This research introduces a novel scheme for Content-Based Medical Image Retrieval (CBMIR). By integrating ResNet50 and Euclidean models, this scheme implements a strategic training procedure aimed at refining feature extraction and enhancing the performance of previous methods. The embedded training process mitigates misinformation during feature extraction, enhancing the system's pattern recognition capability. Through the incorporation of a distance mechanism within ResNet50, this model enhances and adapts its feature extraction capabilities, resulting in improved performance. Evaluation results indicate notable average accuracy, precision, and recall scores of 91.25%, 93.75%, and 91.82% respectively. These findings underscore the model’s capability in retrieving medical images accurately, highlighting its potential for real-world applications. The proposed system distinguishes itself by achieving superior performance compared to prevailing schemes, showcasing the efficiency of integrated ResNet50 and Euclidean models in multimodal CBMIR.