Learning image representation from image reconstruction for a content-based medical image retrieval
- PDF / 721,447 Bytes
- 8 Pages / 595.276 x 790.866 pts Page_size
- 10 Downloads / 291 Views
ORIGINAL PAPER
Learning image representation from image reconstruction for a content-based medical image retrieval Rohini Pinapatruni1 · C. Shoba Bindu1 Received: 7 September 2019 / Revised: 20 January 2020 / Accepted: 6 March 2020 © Springer-Verlag London Ltd., part of Springer Nature 2020
Abstract In this paper, we propose a novel approach of feature learning through image reconstruction for content-based medical image retrieval. We propose an image reconstruction network to encode the input image into a set of features followed by the reconstruction of the input image from the encoded features. The robust reconstruction of the input image from encoded features shows that the encoded features can be used as an abstract version of an input image. Thus, we make use of these encoded features for medical image retrieval task. The performance of the proposed method has been analyzed with the help of three benchmark medical image databases. Average retrieval rate and average precision rate are used to evaluate the performance of proposed and existing state-of-the-art methods for medical image retrieval task. Experimental analysis shows that the proposed approach for image retrieval outperforms the other existing methods. Keywords Convolution neural network · Medical image retrieval · Image reconstruction
1 Introduction In the field of biomedical imaging, there is a large increment in biomedical data because of the use of advanced techniques like X-ray, magnetic resonance imaging, computed tomography. The handling of these enormous data with human efforts is difficult. Thus, there is a need to develop an automatic system for efficient access, search, indexing, and retrieval. Content-based medical image retrieval (CBMIR) aimed at retrieving the medical images which are having similar content to that of input query medical image. CBMIR system comprises two steps: feature extraction followed by index matching and retrieval task. Initially, researchers make use of local feature extraction [1–15] for computer vision applications. Ojala et al. [16] used a grayscale relationship between groups of pixels to extract features named as a local binary pattern (LBP) for texture classification. The extension of LBP descriptor is LTP [7] proposed for face recognition, which improves the encoding mechanism of pixels by using three quantization levels. Murala
B
Rohini Pinapatruni [email protected] C. Shoba Bindu [email protected]
1
JNTUA, Anatapuramu, India
et al. [17] proposed a new technique using edge information by taking the magnitude difference between center pixel and its neighborhood pixels named as local maximum edge binary pattern for CBIR. Murala et al. [8–10] have proposed several local descriptors for both natural and medical image retrievals. Further, Jhanwar et al. [11] used the concept of local gradient and 2 × 2 non-overlapping motif grids for image retrieval. To estimate the attribute of the image, motif occurrence probability is calculated using motif cooccurrence matrix (MCM). Vipparthi et al. [15] propo
Data Loading...
