User-dependence of myocardial infarct identification using semi-automated thresholding techniques: implications for CRT response predictions based on scar burden

Jonathan Suever,John Oshinski,Matthew Janick,Stamatios Lerakis

doi:10.1186/1532-429x-13-s1-p263

Jonathan Suever, John Oshinski + Show 2 more

Open Access

PDF Available

https://doi.org/10.1186/1532-429x-13-s1-p263

Copy DOI

Export

Save

Cite

Abstract
Highlights/Summary
Full-Text PDF
Similar Papers

Abstract

Listen

Background Semi-automatic segmentation of infarcted tissue in late gadolinium-enhanced (LGE) MR images is usually done using a thresholding technique. In the threshold technique, an observer identifies a region of remote “normal” myocardium. Based on the distribution of pixel intensities within this “normal” region, pixels that lay more than two standard deviations outside the mean of this distribution are classified as infarcted myocardium. Having a left ventricular scar burden greater than 33% has previously been used as a cut-off for predicting nonresponse to cardiac resynchronization therapy (CRT) (Chalil et al. Europace, 2007). We hypothesized that the user-defined region of “normal” myocardium would have a significant effect on the classification of patients as non-responders to CRT (>33% LV scar burden).

Highlights

Semi-automatic segmentation of infarcted tissue in late gadolinium-enhanced (LGE) MR images is usually done using a thresholding technique
User-dependence of myocardial infarct identification using semi-automated thresholding techniques: implications for cardiac resynchronization therapy (CRT) response predictions based on scar burden
Having a left ventricular scar burden greater than 33% has previously been used as a cut-off for predicting nonresponse to cardiac resynchronization therapy (CRT) (Chalil et al Europace, 2007)

Summary

Introduction

Semi-automatic segmentation of infarcted tissue in late gadolinium-enhanced (LGE) MR images is usually done using a thresholding technique. User-dependence of myocardial infarct identification using semi-automated thresholding techniques: implications for CRT response predictions based on scar burden Objective To assess the effect of inter-observer variability on myocardial infarct identification and quantification for the prediction of response to CRT. Background Semi-automatic segmentation of infarcted tissue in late gadolinium-enhanced (LGE) MR images is usually done using a thresholding technique.

Results

Conclusion