The rapid growth in digital imaging techniques have resulted in the generation of large volume of diverse medical images. Most of these image corpus is either unlabeled or partially annotated. To ex-tract relevant information from such large-scale image corpus, it is necessary to have an efficient and scalable image retrieval techniques. In this article, we present an effective approach for retrieving images from large-scale Chest X-Ray dataset that have the similar disease conditions or severity as that of the query image. We tested our approach on NIH chest x-ray image dataset, that contains images of pneumonia affected patients. The Histogram of Gradients (HoG) features are found to give better results in classifying the disease. The dimensionality of dense HoG features is reduced by using level decomposition of Haar wavelet and using random projection. The performance degradation happened due to the feature reduction is rectified by using a hybrid approach. The proposed features are compact and capable of conveniently outperforming several existing approaches in image retrieval. To find the nearest match to the query image, the feature space is reduced further by applying k-means clustering. The implementation results are presented to test efficacy of the proposed approach.

Medical image retrieval, pneumonia detection, hand-crafted features, classification, Histogram of Gradient, feature reduction, clustering

[1] B. Venkataramana, L. Padmasree, M. Srinivasa Rao, G. Ganesan, K. Rama Krishna, “Implementation of Clustering Algorithms for real datasets in Medical Diagnostics using MATLAB”, Journal of Soft Computing and Applications, Vol.2017, Issue.1, pp.53-66, 2017.
[2] N. Zayed, H. A. Elnemr, “Statistical Analysis of Haralick Texture Features to Discriminate Lung Abnormalities”, International Journal of Biomedical Imaging, Hindawi, Volume 2015.
[3] N. R. Ratnasari, A. Susanto, Indah Soesanti, Maesadji, “Thoracic X-ray features extraction using thresholding-based ROI template and PCA-based features selection for lung TB classification purposes”, In the Proceedings of 3rd International Conference on Instrumentation, Communications, Information Technology and Biomedical Engineering (ICICI-BME) Bandung, Indonesia, Nov. 2013.
[4] P. M. Gordaliza, J.J. Vaquero, S. Sharpe, M. Desco, A. M. Barrutia, “Towards an Informational Model for Tuberculosis Leision Discrimination on X-RAY CT Images”, In the Proceedings of 15th International Symposium on Biomedical Imaging (ISBI 2018), Washington, DC, USA, April 2018.
[5] Y. Chan, Y. Zeng, H. Wu M. Wu, H. Sun, “Effective Pneumothorax Detection for Chest X-Ray Images Using Local Binary Pattern and Support Vector Machine”, Journal of Healthcare Engineering, Hindawi, Volume 2018, April 2018.
[6] B. C. Ko, S. H. Kim, J. Y. Nam. “X-ray image classification using random forests with local wavelet-based CS-local binary patterns”, Journal of Digital Imaging, Vol. 24, Issue. 6, pp. 1141-1151, 2011.
[7] K. T. Vo, A. Sowmya, “Diffuse lung disease classification in HRCT lung images using generalized Gaussian density modeling of wavelets coefficients”, In the Proceedings of 16th IEEE International Conference on Image Processing (ICIP), Cairo, Egypt, November 2009.
[8] C. Ray, K. Sasmal, “A New Approach for Clustering of X-ray Images”, International Journal of Computer Science Issues, Vol.7, Issue.4, No.8, July 2010.
[9] C. Reta, I. S. Moreno, J. A. C. Ceballos, “Improving content-based image retrieval for heterogeneous datasets using histogram-based descriptors”, Multimedia Tools and Applications, Springer, Vol. 77, Issue. 7, pp. 8163–8193, April 2018.
[10] M. Muja, D. G. Lowe, “Fast Matching of Binary Features”, In the Proceedings of Ninth Conference on Computer and Robot Vision, CRV `12, pp. 404-410, 2012.
[11] J. Ahmad, K. Muhammad, S.W. J. Baik, “Medical Image Retrieval with Compact Binary Codes Generated in Frequency Domain Using Highly Reactive Convolutional Features”, Journal of Medical Systems, Vol. 42, Issue 24, February 2018.
[12] Z. Li, X. Zhang, H. Müller, S. Zhang, “Large-scale retrieval for Medical Image Analytics: A Comprehensive Review”, Medical Image Analysis, Vol.43 , pp. 66–84, 2018.
[13] W. Weihong , W. Song, “A Scalable Content-based Image Retrieval Scheme using Locality-sensitive Hashing”, International Conference on Computational Intelligence and Natural Computing, Wuhan, China, 2009.
[14] T. Reato, B. Demir L. Bruzzone, “Primitive cluster sensitive hashing for scalable content-based image retrieval in remote sensing archives”, In the proceedings of IEEE International Geoscience and Remote Sensing Symposium (IGARSS), Fort Worth, TX, USA, July 2017.
[15] M. Bruijne, “Machine Learning Approaches in Medical Image Analysis: From Detection to Diagnosis”, Medical Image Analysis, Vol. 33, 94–97, 2016.
[16] Y. D. Cid et al. “Making Sense of Large Data Sets without Annotations: Analyzing Age–related Correlations from Lung CT Scans”, In the Proceedings of Imaging Informatics for Healthcare, Research, and Applications, Medical Imaging, Vol. 10138, Orlando, Florida, United States, March 2017.
[17] I. E. Livieris, A. Kanavos, V. Tampakas, P. Pintelas, “An Ensemble SSL Algorithm for Efficient Chest X-Ray Image Classification”, Journal of Imaging, Vol.4, Issue.7, 2018.
[18] J. G. Dy, C. E. Brodley, A. Kak, L. S. Broderick, A. M. Aisen, “Unsupervised Feature Selection Applied to Content-Based Retrieval of Lung Images”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 25, No. 3, March 2003.
[19] E. M.F. E. Houby, “Medical Images Retrieval using Clustering Technique”, International Journal on Recent and Innovation Trends in Computing and Communication, Vol.3, Issue.5, pp.3134-3141, October 2015.
[20] P. Premalatha, S. Subasree, “Performance Analysis of Clustering Algorithms in Medical Datasets”, In the Proceedings of Second International Conference on Electrical, Computer and Communication Technologies (ICECCT), Coimbatore, India, 2017.