一种基于灰度修正的心肌瘢痕阈值分割方法

A Myocardial Scar Threshold Segmentation Algorithm Based on Grey-level Correction

摘要:

将心肌瘢痕从心脏磁共振延迟强化图像中分离出来并进行定性定量分析，对缺血性心脏病患者的早期诊断和准确预后评估具有十分重要的意义。针对心脏左心室内外膜边缘周围常见的噪声容易导致心肌瘢痕错分的问题，构建了一种基于灰度修正的心肌瘢痕阈值分割方法。方法依次采用像素标记、影响力计算和灰度更新三个步骤对心肌区域进行灰度修正，然后结合OSTU算法对心肌瘢痕进行阈值分割。实验结果表明，本文方法较以往的心肌瘢痕分割算法更接近有经验医生的人工分割结果，有效避免了左心室内外膜边缘噪声可能导致的心肌瘢痕的错分。

Abstract:

Isolation of the myocardial scar from delayed enhancement cardiac magnetic resonance image (DEMRI) for a qualitative and quantitative assessment is of significant early diagnostic and accurate prognostic value for patients with ischemic heart diseases. To address the issue of the common error in the segmentation of the myocardial scar as a result of noise due to endocardial contours of the left ventricle, the threshold segmentation algorithm for myocardial scar based on the grey-level correction is proposed. In this algorithm, the grey-level correction of the myocardial region is first performed in the order of the following three steps: pixel labeling, influence calculation and grey updating. Then the threshold segmentation of the myocardial scar is conducted in combination with the OSTU algorithm. The experimental results indicate that the proposed method in the paper is better agreement with the results obtained with manual tracing by experience physicians compared to the previous myocardial scar segmentation algorithm, effectively preventing the error in the segmentation of the myocardial scar as a result of noise due to endocardial contours of the left ventricle.