Objectives To examine the factors affecting radiomic analysis of the myocardium and to identify the optimal approach to image analysis. Materials and methods Texture features were extracted from left ventricular (LV) non-contrast cine images. Texture variable reduction work was performed by studying the region of interest (ROI) size influence and intra- and inter-observer repeatability. The texture analysis (TA) process was applied to a healthy population (n = 600 individuals) for further analysis. ROI were defined at end diastole (ED) and end systole (ES), with separate ROI drawn encompassing the whole myocardial wall, septal wall, and the free wall. Differences between ED and ES; females and males; different ages; and myocardial regions were compared. Results Fifteen TA variables were tested; the majority were affected by the phase of imaging. Texture varied significantly with age, with ED images and with a whole wall ROI best differentiating between age groups. In the detection of physiological differences between men and women, the combination of ED with whole myocardium quantification resulted in a greater number of features showing a significant difference between the sexes. Consistent TA results were obtained for ROIs placed on the whole myocardial wall and on the septal wall. Conclusions In a cohort of healthy volunteers, we have identified a subset of features with high reproducibility that are invariant to the region of interest size. Use of the whole myocardium at ED results in the greatest discrimination of physiological changes associated with age and sex. Key Points <Emphasis Type="BoldItalic">Question</Emphasis> The characteristics of healthy population myocardium texture using cardiac MR texture analysis are unclear. <Emphasis Type="BoldItalic">Findings</Emphasis> A reasonable number of robust texture variables can discriminate physiological changes associated with age and sex. <Emphasis Type="BoldItalic">Clinical relevance</Emphasis> This work develops a deeper understanding of healthy population myocardium variability, thus helping to increase awareness and recognition of myocardial disease. It is also achievable without the need for contrast agent injections or additional scanning. Graphical Abstract

Cardiac left ventricular MRI texture analysis to derive texture characteristics of a healthy population: clinical implications

期刊论文
出版模式：

开放获取
发表日期：

2025年7月26日

查看全文

Ping Tie,
Stephen J. Gandy,
Jonathan Weir-McCall,
John Graeme Houston

摘要

Objectives

To examine the factors affecting radiomic analysis of the myocardium and to identify the optimal approach to image analysis.

Materials and methods

Texture features were extracted from left ventricular (LV) non-contrast cine images. Texture variable reduction work was performed by studying the region of interest (ROI) size influence and intra- and inter-observer repeatability. The texture analysis (TA) process was applied to a healthy population (n = 600 individuals) for further analysis. ROI were defined at end diastole (ED) and end systole (ES), with separate ROI drawn encompassing the whole myocardial wall, septal wall, and the free wall. Differences between ED and ES; females and males; different ages; and myocardial regions were compared.

Results

Fifteen TA variables were tested; the majority were affected by the phase of imaging. Texture varied significantly with age, with ED images and with a whole wall ROI best differentiating between age groups. In the detection of physiological differences between men and women, the combination of ED with whole myocardium quantification resulted in a greater number of features showing a significant difference between the sexes. Consistent TA results were obtained for ROIs placed on the whole myocardial wall and on the septal wall.

Conclusions

In a cohort of healthy volunteers, we have identified a subset of features with high reproducibility that are invariant to the region of interest size. Use of the whole myocardium at ED results in the greatest discrimination of physiological changes associated with age and sex.

Key Points

Question The characteristics of healthy population myocardium texture using cardiac MR texture analysis are unclear.

Findings A reasonable number of robust texture variables can discriminate physiological changes associated with age and sex.

Clinical relevance This work develops a deeper understanding of healthy population myocardium variability, thus helping to increase awareness and recognition of myocardial disease. It is also achievable without the need for contrast agent injections or additional scanning.

Graphical Abstract