Semi-automatic quantification of myocardial infarction from delayed contrast enhanced magnetic resonance imaging
2005 (English)In: Scandinavian Cardiovascular Journal, ISSN 1401-7431, Vol. 39, no 5, 267-275 p.Article in journal (Refereed) Published
Objective. Accurate and reproducible assessment of myocardial infarction is important for treatment planning in patients with ischemic heart disease. This study describes a novel method to quantify myocardial infarction by semi-automatic delineation of hyperenhanced myocardium in delayed contrast enhanced (DE) magnetic resonance (MR) images. Design. The proposed method automatically detects the hyperenhanced tissue by first determining the signal intensity of non-enhanced myocardium. A fast level set algorithm was used to limit the heterogeneity of the hyperenhanced regions, and to exclude small regions that constitute noise rather than infarction. The method was evaluated in 40 patients, 20 with acute infarction and 20 with chronic healed infarction using scanners from two different manufacturers. Infarct size measured by the proposed semi-automatic method was compared with manual measurements from three experienced observers. The software used is freely available for research purposes at http://segment.heiberg.se. Results. The difference in infarct size between semi-automatic quantification and the mean of the three observers was 6.1 ± 6.6 ml (mean ± SD), and the interobserver variability (SD) was 4.2 ml. Conclusions. The method presented is a highly automated method for analyzing myocardial viability from DE-MR images. The bias of the method is acceptable and the variability is in the same order of magnitude as the interobserver variability for manual delineations. © 2005 Taylor & Francis.
Place, publisher, year, edition, pages
2005. Vol. 39, no 5, 267-275 p.
MRI, myocardial infarction, quantification, viability
National CategoryMedical and Health Sciences
IdentifiersURN: urn:nbn:se:liu:diva-33750DOI: 10.1080/14017430500340543Local ID: 19796OAI: oai:DiVA.org:liu-33750DiVA: diva2:254573