Application of Pre-processing and Segmentation Methods on Cardiac MR Images G. N. Beena Bethel, Assoc. Professor, CSE Dept, GRIET, Hyderabad, India. email_id: beenabethel@gmail.com Prof. T. V. Rajinikanth, Professor, CSE Dept., SNIST, Hyderabad, India. email_id: rajinitv@gmail.com Prof. S. Viswanadha Raju Professor, CSE Dept., JNTUH (Jagityal), Karimnagar, AP., India. email_id: svraju.jntu@gmail.com Abstract Cardiac MRI (Magnetic Resonance Imaging), is a non-invasive technique in the field of medical imaging technology for assessing the heart function and also to diagnose and analyse the morphological features of the cardiovascular system of a human heart. It gives a clear picture of heart’s chambers and valves, without the patient having to undergo cardiac catheterization for most cases. Analysing the functionality of heart and diagnosing its varieties of ailments at the right time without causing any punctures is a challenge in today’s medical community. Locating the exact region of ailment or interest, especially in a sensitive organ such as heart is quite cumbersome even with the help of imaging techniques. Application of Image pre-processing techniques to reduce noise in the heart MRI images apart from enhancement of the MRI images and further followed by segmentation methods in order to locate the problem area in the heart will be a boon to the Cardiologist / Cardio Surgeon to carry out better diagnosis. The objective of this paper is to find a better filtering technique and a segmentation method which would help fast and accurate tracing of the portion of the heart ailment in a better manner with more clarity. Keywords: Heart images; Cardiac Magnetic Resonance Imaging; segmentation methods; noise filtering; threshold; watershed. 1 Introduction Cardiac Image processing has gained its attention after the heart failures have been increasing day after day. A quick diagnosis and immediate attention to the various types of heart ailments could save lives and lot of research is heading towards these goals. According to World Health Organization’s report 17.5 million deaths out of 31 million per year are due to heart diseases. Almost 80% of them are a result of heart attacks or strokes. Enhancing an image, particularly medical images leads to correct diagnosis and can be done both in spatial domain and time domain. Representation of an image in two-dimensional space f(x, y) can be represented by the gray-level intensity at that point (x, y). This part of the work is done considering the gray level images of cardiac MRI taken from 33 subjects affected with Congenital Heart diseases. The images were first pre-processed using different filters. In literature, Christine Guillemot, et al., in [1] have worked on transforms relying on signal extensions, shape-adaptive block transforms and transforms relying on time-varying multi-rate filter banks. V. Behar et al., in [2] have proved the capacity of a nonlinear filter to improvise the images, disturbed by spectrally and spatially correlated Gaussian noise, and an average Improvement factor in the Peak Signal to Noise Ratio called IPSNR was evaluated. Prodip Biswas et al., have worked on de-blurring of images using wiener filter in [3]. An automatic threshold selection based on nonparametric and unsupervised method was first introduced by N. Otsu [4] and was named after him as Otsu thresholding. Later OTSU Method of filtering was extended to an Iterative Triclass Thresholding Technique, by P. D. Honawadajkar et al., in [5]. Ch. Hima Bindu et al., [6] applied an improved Otsu thresholding on brain images which could reduce the computational complexity of the G. N. Beena Bethel et al. / Indian Journal of Computer Science and Engineering (IJCSE) ISSN : 0976-5166 Vol. 8 No. 3 Jun-Jul 2017 210