Characterization Of Normal And Scarred Myocardium Based On Texture Analysis Of Cardiac Computed Tomography Images

An accurate detection of myocardial scar using Cardiac CT may have a strong clinical impact; however, the main drawback is the insufficient contrast to noise ratio of delayed iodine enhanced (DIE) CT images, which makes its accurate segmentation (manual as well as automatic) difficult. In this work, we investigate texture parameters applied on the different scans in order to obtain the scans and features that best differentiates normal from scarred myocardium. The experiments on 7 cases of myocarditis show the accuracy of the parameter energy in all scans, as well as the good performance of the angiographic scan (having higher spatial resolution) with different parameters for the segmentation propose. Moreover, the best performance was obtained on the baseline scan for the energy feature, with an accuracy of 94%.