Magnetic Resonance Imaging (MRI) scan are the most widely used radiographic techniques in diagnosis, clinical studies and treatment planning. It is important to be able to detect when an automatic segmentation method fails to avoid inclusion of wrong measurements into subsequent analysis which could otherwise lead to incorrect conclusion. Sometimes due to absence of Ground Truth (manually labelled) images it is difficult to detect the failure of automatic segmentation methods. Before deployment, performance is quantised using different metrics for which the predicted segmentation is compared to a reference segmentation also known as Ground Truth (GT), which is manually obtained by an expert. In some exceptional cases it becomes difficult to know about its real performance after deployment when a reference is unavailable. To that end, this paper aims to develop an improved and advanced technique of Reverse Classification Accuracy (RCA) on new data which enables us to discriminate between the successful and failed cases. The ideal concept is that to rank the ‘best’ segmentation results in the database without knowing the manual label. Then ‘match’ the rank between the prediction and the truth image saved in database. Further, for correctly and accurately segmented and classified brain MRI images, diseases are being detected using Random forest algorithm and Deep Learning.

Machine learning, Deep Learning, image Segmentation, MRI images, classification, performance evaluation

[1] Vanya V. Valindria, Ioannis Lavdas, Wenjia Bai, Konstantinos Kamnitsas, Eric O. Aboagye,Andrea G. Rockall, Daniel Rueckert, and Ben Glocker,“Reverse Classificaton Accuracy: Predicting Segmentation Performance in the absence of Ground Truth,” IEEE Transactions on Medical Imaging, Vol. 36, No. 8, 2017.
[2] T. Heimann and H.-P. Meinzer, “Statistical shape models for 3D medical image segmentation: A review” Med. Image Anal., vol. 13, no. 4, pp. 543-563, 2009.
[3] J. E. Iglesias and M. R. Sabuncu, “Multi-atlas segmentation of biomedical images: A survey” Med. Image Anal., vol. 24, no. 1, pp. 205-219,2015.
[4] E. Geremia et al., “Classification forests for semantic segmentation of brain lesions in multi-channel MRI” in Decision Forests for Computer Vision and Medical Image Analysis. London, U.K.: Springer, pp. 245260, 2013.
[5] E. Zhong, W. Fan, Q. Yang, O. Verscheure, and J. Ren, “Cross validation framework to choose amongst models and datasets for transfer learning” in Proc. Joint Eur. Conf. Mach. Learn. Knowl. Discovery Databases, pp. 547-562, 2010.
[6] W. Fan and I. Davidson, “Reverse testing: An efficient framework to select amongst classifiers under sample selection bias,” in Proc. 12th ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, pp. 147-156, 2006.
[7] F. Ge, S. Wang, and T. Liu, “New benchmark for image segmentation evaluation,” J. Electron. Image. vol. 16, no. 3, pp. 033011, 2007.
[8] L. Goldmann et al., “Towards fully automatic image segmentation evaluation,” in Proc. Int. Conf. Adv. Concepts Intell. Vis. Syst., pp. 566-577, 2008.
[9] H. Li, J. Cai, T. N. A. Nguyen, and J. Zheng, “A benchmark for semantic image segmentation”,in Proc. IEEE Int. Conf. Multimedia Expo, pp. 16,2013.
[10] K. Sikka and T. M. Deserno“Comparison of algorithms for ultrasound image segmentation without ground truth”, Proc. SPIE, vol. 7627, pp. 76271C176271C9, 2010.
[11] Leo Breimen, “Random Forest”, 2001.
[12] K.Yogeswara Rao, M.James Stephen, D.Siva Phanindra,“Classification Based Image Segmentation Approach”, IJCST, Vol 3, Issue 1, 2012.
[13] Xiaofeng Ren and Jitendra Malik,“Learning A Classification Model for Segmentation”.
[14] Dr Stefan Teipel, Michel J Grothe, , Henryk Barthel,“Multimodal imaging in Alzheimer’s disease: validity and usefulness for early detection ”,Lanclet Neurology,2015.
[15] B. van Ginnekan, T. Heimann, M.Styner,“3D Segmentation in the clinic: A Grand Challenge ”, in proc. MICCAI Workshop 3D Segmentation Clinic, pp. 7-15, 2007.
[16] C. Sudlow et al.,“UK Biobank: An open access resource for identifying the causes of a wide range of complex diseases of middle and old age ”, PLoS Med., vol. 12, no. 3, p.e1001779, 2015.
[17] H. Zhang, S. Cholleti, S.A. Goldman, J.E.Fritts,“Meta-evaluation of Image Segmentation using Machine Learning ”,in Proc. IEEEE Comput. Soc. Conf. Compute. Vis. Pattern REcognit. (CVPR). vol. 1, pp. 1138-1145,2006.
[18] B.Lamiroy and T. Sun,“Computing Precision and Recall with Missing or Uncertain Ground Truth”, in Graphics Recognition. New Trends and Challenges. Berlin, Germany: Springer- Verlag, pp. 149-162,2013
[19] S. Chabrier, B. Emile, C. Rosenberger, H. Laurent,“Unsupervised Performance Evaluation of Image Segmentataion”, EURASIP J. Appl.Signal Process., vol. 2006, pp. 217-229, 2006.
[20] H. Zhang, J. E. Fritts, and S. A. Goldman, “Image Segmentation Evaluation: A survey of Unsupervised Methods”,Computer Vision and Image Understanding , vol. 110, no. 2, pp. 260-280, 2008.
[21] K. Kamnitsas, C. Ledig, V. F. Newcombe, J. P. Simpson, A. D. Kane, D. K. Menon, D. Rueckert, B. Glocker, “Efficient Multi-scale 3D CNN with Fully Connected CRF for Accurate Brain Lesion Segmentation,” Medical Image Analysis, 2016.
[22] W. Bai, W. Shi, D. P. ORegan, T. Tong, H. Wang, S. Jamil- Copley, N. S. Peters, and D. Rueckert,“A Probabilistic Patch-Based Label Fusion Model for Multi-Atlas Segmentation with Registration Refinement: Application to Cardiac MR Images”, IEEE Transactions on Medical Imaging, vol. 32,no. 7, pp. 1302-1315, 2013