Deep Learning in Content-Based Medical Image Retrieval
Harpal Singh, Priyanka Kaushal, Meenakshi Garg, Gaurav Dhiman · 2021
This chapter proposes a deep learning system for medical image retrieval based on content by training a profound convolutionary neural network for the purpose of classification. The first step is to get a prediction of the query class of the query picture by the qualified network and then find related pictures in that particular class. The second approach is to search for all applicable images in the entire database without adding details about the query image type. Deep learning-based technique Deep Neural Network (DNN) classifier is used to classify medical images for categorizing the images. For tuning the parameters of deep learning models, Spotted Hyena Optimizer (SHO) is used. By learning discriminatory characteristics from pictures directly, the proposed solution reduces semantic distance. The network has succeeded in training medical images with 99.77 per cent average rating accuracy. In order to remove the recovery functionality, the last three fully connected network layers were used. Widespread measurements such as accuracy and reminders have been used to assess the efficiency of the proposed system for recovery of medical images, which helps the doctors to analyse the disease better.