- Poster presentation
- Open Access
Automated myocardium segmentation in late gadolinium enhanced MR images
© Tao et al.; licensee BioMed Central Ltd. 2014
- Published: 16 January 2014
- Myocardial Scar
- Late Gadolinium Enhance
- Endocardial Contour
- Epicardial Contour
- Automate Contour
Late Gadolinium Enhanced (LGE) MRI has proven clinical value for diagnosis and prognosis of post-infarct patients. A prerequisite for accurate myocardial scar assessment is the reliable segmentation of the myocardium. However, in post-infarct patients, the myocardial scar is often connected to the blood pool, while the contrast between them is typically poor. The ambiguous border between myocardial scar and blood pool substantially complicates manual contouring, and potentially gives rise to over- or under-estimation of the scar.
We propose an automated LGE segmentation algorithm, which utilizes the myocardium morphology information from the cine sequence in the same study. In practice, the cine contours of the end-diastolic and end-systolic phases are routinely annotated for functional analysis. From the available cine contours, the potential endo- and epicardial contour sets for each LGE slice were estimated by cine-LGE registration and linear interpolation. Secondly, the match between the dual contour sets and each LGE slice was evaluated by correlating the LGE slice with the contour image constituted of edge filters adaptive to the edge direction. Finally, the global optimal contour was established by dynamic programming through the entire LGE stack. The method was validated on the LGE MR of 30 post-infarct patients with an in-plane resolution of 1.56 mm, and compared to the manual contouring results.
We proposed an automated method to segment the myocardium in LGE for myocardial scar assessment. The method resulted in comparable performance with manual scar segmentation by experienced observers, in terms of contouring accuracy and myocardial scar size. The method does not require extra cine contour analysis than routinely performed, and reduces the observer-dependency by avoiding visual interpretation of LGE.
The work was financially supported by the EU MEDIATE project (ITEA2-09039).
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.