Content Based Image Retrieval System using Modified LSTM with Clustering and K-D Tree Indexing Techniques

A Robust Content-Based Medical Image Retrieval System must be designed to support the ever-increasing volume of digital photographs across various medical disciplines in order to successfully return medical photos that are aesthetically comparable to the query image. The CBMIR approach for optimum medical image retrieval is presented in this research work, with a focus on reducing deviation in medical image mining and a full analysis of significant aspects of medical photos. The purpose of this research was to develop a LSTM based deep learning model for classifying CT Liver tumor images from the CT Liver tumor images dataset. The dataset is preprocessed with thresholding, extreme point computation, and bicubic interpolation. Second, the proposed method uses a convolutional neural network based LSTM to extract information from cropped photos. There are four measures used to evaluate the model's efficacy: accuracy, precision, recall. Accuracy at 96.88%, precision at 97.96%, recall at 96.60% are all optimally achieved by the proposed model. According to the findings, the proposed LSTM based RNN Approach with KD Tree model is the most efficient method for identifying CT Liver Tumor images detection and classification.