Deep Learning for Quantification of Epicardial Fat from Non-Contrast CT

Epicardial Adipose Tissue is a visceral fat presently on the walls of the heart and pericardium.it has a direct correlation with coronary disease.it likewise possibly causes nearby inflammation and likely has direct effects on coronary atherosclerosis. This metabolically recognized by pro-inflammatory mediator. hence this fat must be quantified and segmented to diagonized the heart related disease..EAT can be manually measured but it is difficult task.it requires experienced technical members it is a time consuming process.so we propose a novel segmentation algorithm using k means clustering and further propose convolution neural network for fine segmentation of the image. We advise a fast and fully mechanized algorithm for Epicardial Adipsoe Tissue volume calculation from non-contrast calcium scoring Computed Tomography datasets using a machine learning Tecniques it is nothing but machine learnig techniques. Our proposed approach involves in preprocessing of CT images and Segment the image using K mean clustering algorithm. Then we fine tune the segmented image using CNN framework.CNN is an unsupervised calculation and machine learning method is utilized for how to adapt such great highlights automatically. The purpose behind we assessed the epicardial fat volume is the epicardial layer is closer to the heart. If may the layer thickness get expanded than specific range the patient may influenced by coronary illness. At long last it causes passing. We evaluated the performance of this process in terms of epicardial fat volume, Correlation Coeffecient, Mean Squared error, processing time.