2017/08/03 by Jelmer M. Wolterink, Wolterink, Jelmer M., Tim Leiner +5 · 1 citation
Engineering · Medicine · #Advanced X-ray and CT Imaging #Cardiac Imaging and Diagnostics #Cardiovascular Disease and Adiposity #Cardiovascular Function and Risk Factors #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences
paper · pdf · doi:10.48550/arxiv.1708.01141
openalex publication_date 2017/08/03 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Segmentation of the heart in cardiac cine MR is clinically used to quantify\ncardiac function. We propose a fully automatic method for segmentation and\ndisease classification using cardiac cine MR images. A convolutional neural\nnetwork (CNN) was designed to simultaneously segment the left ventricle (LV),\nright ventricle (RV) and myocardium in end-diastole (ED) and end-systole (ES)\nimages. Features derived from the obtained segmentations were used in a Random\nForest classifier to label patients as suffering from dilated cardiomyopathy,\nhypertrophic cardiomyopathy, heart failure following myocardial infarction,\nright ventricular abnormality, or no cardiac disease. The method was developed\nand evaluated using a balanced dataset containing images of 100 patients, which\nwas provided in the MICCAI 2017 automated cardiac diagnosis challenge (ACDC).\nThe segmentation and classification pipeline were evaluated in a four-fold\nstratified cross-validation. Average Dice scores between reference and\nautomatically obtained segmentations were 0.94, 0.88 and 0.87 for the LV, RV\nand myocardium. The classifier assigned 91% of patients to the correct disease\ncategory. Segmentation and disease classification took 5 s per patient. The\nresults of our study suggest that image-based diagnosis using cine MR cardiac\nscans can be performed automatically with high accuracy.\n