RetrieveNet: a novel deep network for medical image retrieval

• Published in: Evolutionary intelligence 2021-12, Vol.14 (4), p.1449-1458
• Main Authors: Hussain, Chesti Altaff, Rao, Dhulipalla Venkata, Mastani, S. Aruna
• Format: Article
• Language: English
• Subjects: Applications of Mathematics; Artificial Intelligence; Bioinformatics; Computed tomography; Control; Engineering; Face recognition; Feature extraction; Filter banks; Image classification; Image management; Image retrieval; Magnetic resonance imaging; Mathematical and Computational Engineering; Mechatronics; Medical imaging; Object recognition; Performance evaluation; Research Paper; Robotics; Statistical Physics and Dynamical Systems
• ISSN: 1864-5909; 1864-5917
• DOI: 10.1007/s12065-020-00401-z

Content-Based Image Retrieval is an accurate characterization of visual information used for medical and natural image classification, retrieval, face recognition, etc. In recent years, deep networks achieved state-of-the-art accuracy in various vision tasks. In this paper, we propose an end-to-end approach for medical image retrieval. The proposed approach comprises a novel deep network for classification of input query image followed by the retrieval module to retrieve the images belongs to the query image class. The proposed network comprises of multi-scale filter bank for robust feature extraction. We make use of skip connections to share the initially learned features across the network. The performance of the proposed network for medical image retrieval is validated in terms of precision and recall on three publicly available medical image databases. We validated the proposed approach for two different image modalities, namely CT scans and MRI scans. Performance evaluation depicts that proposed RetrieveNet outperforms other existing methods for medical image retrieval.