A sequential search-space shrinking using CNN transfer learning and a Radon projection pool for medical image retrieval. (15th June 2018)

• Record Type: Journal Article
• Authors: Khatami, Amin; Babaie, Morteza; Tizhoosh, H.R.; Khosravi, Abbas; Nguyen, Thanh; Nahavandi, Saeid
• Is Part Of: Expert systems with applications. Volume 100(2018)
• Page Start: 224
• Page End: 233
• Publication Date: 2018-06-15
• Subjects: Content-based image retrieval -- CBIR -- Medical imaging -- Deep learning -- Radon; Expert systems (Computer science) -- Periodicals; Systèmes experts (Informatique) -- Périodiques; Electronic journals; 006.33
• DOI: 10.1016/j.eswa.2018.01.056
• Languages: English
• ISSNs: 0957-4174

Abstract

Highlights: Propose a two-step hierarchical shrinking search space when a descriptoris used. Transfer learning via CNNs is utilized for the first stage shrinking. A selection pool using Radon transform is created for further shrinking. Difference between two orthogonal Radon projections is considered in the pool. A state-of-the-art result is achieved on the IRMA challenging dataset. Abstract: Closing the semantic gap in medical image analysis is critical. Access to large-scale datasets might help to narrow the gap. However, large and balanced datasets may not always be available. On the other side, retrieving similar images from an archive is a valuable task to facilitate better diagnosis. In this work, we concentrate on forming a search space, consisting of the most similar images for a given query, to be used for a similarity-based search technique in a retrieval system. We propose a two-step hierarchical shrinking search space when local binary patterns are used. Transfer learning via convolutional neural networks is utilized for the first stage of search space shrinking, followed by creating a selection pool using Radon transform for further reduction. The difference between two orthogonal Radon projections is considered in the selection pool to extract more information. The IRMA dataset, from ImageCLEF initiative, containing 14, 400 X-ray images, is used to validate the proposed scheme. We report a total IRMA error of 168.05 (or 90.30% accuracy) which is the best result compared with existing methods in the literature for this dataset when real-time processing is considered.