Content-Based Medical Image Retrieval Using CNN Feature Extraction and Hashing
Nepoleon Keisham, Arambam Neelima · 2024
Content-Based Medical Image Retrieval (CBMIR) plays a vital role in enabling medical practitioners to efficiently access extensive image databases. This study introduces a novel approach to CBMIR that harnesses Convolutional Neural Networks (CNN) for feature extraction and utilizes Hashing technique for Data Integrity Verification and fast retrieval. Through training on a comprehensive dataset of medical images, CNN learns to extract discriminative and representative features effectively. This ability to capture intricate patterns and context in medical images enriches the feature representation, preserving essential diagnostic information. After feature extraction, Hashing technique is applied to convert the high-dimensional feature vectors into compact binary codes. This not only reduces storage requirements but also significantly speeds up retrieval times, making the proposed CBMIR system highly suitable for real-time applications. In conclusion, the fusion of CNN-based feature extraction and Hashing technique presents a potent solution for Content-Based Medical Image Retrieval (CBMIR). The model’s superior performance compared to existing methods demonstrates its effectiveness and highlights its potential to enhance medical image analysis and clinical decision-making processes. As a result, the proposed CBMIR system holds immense promise as an invaluable tool for medical professionals, facilitating rapid and accurate access to critical medical information.