IEEE Transactions on Pattern Analysis and Machine Intelligence
Hi-index | 0.00 |
This paper presents a novel approach for segmentation of the infarct and peri-infarct tissue in the left ventricular wall of the heart. This paper is motivated by a recent finding that shows the infarct and peri-infarct zones to be independent predictors of post myocardial infarction. This paper proposes a method to segment the endocardial and epicardial contours of the left ventricle in the presence of the enhanced infarct and peri-infarct tissues. A level set method using shape priors, obtained from a 3D active appearance model of the ventricle wall on cine MR images is presented. From the extracted 3D cardiac ventricular wall, a method is proposed to segment the infarct and peri-infarct tissues using intensity, volume, shape and heart wall thickness features. The parameters of end-diastolic volume, end-systolic volume, myocardial mass, ejection fraction and infarct and peri-infarct mass are computed using the proposed method and compared with the gold standard provided by the cardiologists. Promising results and comparisons demonstrate the potential of our approach for a practical computer assisted diagnostic system.