Accurate Implantation Estimation of Implantable Cardioverter Defibrillators via Three-Stage Segmentation Strategy

Myocardial scar regions appear in cardiac magnetic resonance (CMR) images of patients with myocardial infarction. Implantable cardioverter defibrillators (ICDs) can effectively prevent arrhythmia and even death caused by myocardial infarction. Whether or not to implant an ICD and the precise location of implantation is a huge clinical challenge. This work proposes a three-stage segmentation strategy for accurate estimation of ICD implantation. We first achieve accurate mixed or unknown noise estimation of cardiac magnetic resonance images. A weighted regularization term is proposed by weighting mixed noise distribution and using spatial information constraints, thus forming weighted Fuzzy C-Means for CMR image segmentation. Through region growth and flood fill in order, the region and volume of myocardial scar are precisely obtained. Thus, the ICD implantation can be accurately estimated. Experimental results on myocardial infarction datasets show that the proposed algorithm is more effective and efficient than its peers.