An Image Classification and Retrieval Hybrid Model for Larger Healthcare Datasets using Deep Learning

Abstract

Objectives:The objective of this work is to obtain an efficient medical image retrieval and classification from a larger healthcare datasets using Novel approach. Methods: In this study five different classes of Medical images are taken for input, features are extracted using GLCM (Grey Level Co-occurrence Matrix) by image attributes such as dissimilarity, correlation, homogeneity, contrast, ASM, and energy. The photos are examined at several angles (0, 45, 90, and 135) to extract the characteristics using the layers. The received feature vectors are input into the most often used deep learning models Artificial Neural Networks (ANN) and Convolution Neural Networks (CNN) for image classification. Then CNN model is integrated with a deep learning model based on Long-Short Term Memory (LSTM), which incorporates additional layers into its structure and works on large datasets. Further the retrieval performance is improved by Euclidean Distance Technique. Findings: Performance evaluation is performed by comparing and analyzing the experimental findings of proposed methods, ANN, CNN and CNN-LSTM yields the retrieval accuracy of 97.79%, 98.78% and 99.4%. The Precision, Recall and F1-Score are also compared, and they are more accurate when picture classification is performed on larger healthcare datasets. Novelty: The additional feature extraction using GLCM and the proposed hybrid model can extract better medical image features, and achieve higher classification accuracy compared with earlier image classification models.