Improved segmentation technique to detect cardiac infarction in MRI C-SENC images

Composite Strain Encoding (C-SENC) is a new MRI technique that acquires cardiac functional and viability images simultaneously. It combines the use of Delayed Enhancement (DE) imaging to identify the infracted (dead) tissue inside the heart muscle and the ability to image myocardial deformation from the Strain Encoding (SENC) imaging technique. In this work, a new multi-stage technique is proposed to objectively identify infarcted heart tissues in the functional and viability images provided by C-SENC MRI. The proposed technique is based on sequential application of Bayes classifier, Otsu's thresholding, morphological opening, radial sweep boundary tracing and the fuzzy C-means (FCM) clustering algorithm. This technique is tested on images of eleven patients suffering myocardial infarction (MI). The resulting clustered images are compared with those marked up by expert cardiologists who assisted in validating results coming from the proposed technique. Infarcted myocardium is correctly identified using the proposed technique with high levels of accuracy and precision.