- Poster presentation
- Open Access
Automated quantitative assessment of myocardial infarction in late enhancement MRI
https://doi.org/10.1186/1532-429X-12-S1-P111
© Tao et al; licensee BioMed Central Ltd. 2010
- Published: 21 January 2010
Keywords
- Myocardial Infarction
- Automate Method
- Short Axis Slice
- Late Enhancement
- Viability Assessment
Introduction
Accurate quantitative assessment of the size and distribution of myocardial infarction (MI) from late enhancement (LE) MRI is of significant prognostic value for post-infarction patients. Manual processing of the data is labor-intensive and simple processing methods, like thresholding, tend to produce unreliable results.
Purpose
The purpose of this study was to design an automated, robust, and systematic method for labeling the MI in LE MR imaging for quantitative MI assessment.
Methods
Twenty patients with known chronic myocardial infarction (all male, mean age 64 ± 8, range 45-82 years) referred for viability assessment were included. LE MR was performed in multiple short axis slices covering the entire LV (slice thickness 10 mm, 5 mm overlap). Endocardial and epicardial LV contours were derived semi-automatically taking into account corresponding cine MR data. Two independent observers manually outlined the MI regions from a total of 348 slices.
The automated method started with finding a reliable and robust threshold on the image intensity, to discriminate the hyperenhanced MI from the normal myocardial tissue. The identified regions were subsequently processed with respect to their size and geometry to preclude falsely identified MI regions caused by noise or contour tracing error. Finally, the remaining MI regions were further refined by region-growing to achieve an explicit delineation of the entire MI region.
Results
Comparison of MI labeling results. Column (a): the original LE MR images, (b): MI labeled by the automated method, (c): MI labeled by observer 1, and (d): MI labeled by observer 2.
Bland-Altman analysis of the percentage infarction between the automated method and manual tracing of observer 1 and observer 2, respectively.
Conclusion
An automated MI labeling method is proposed in this study. Validation results demonstrated that the method can provide accurate quantitative assessment of the MI for post-infarction patients. More extensive study like heterogeneity and transmurality analysis can be done on this basis.
Authors’ Affiliations
Copyright
This article is published under license to BioMed Central Ltd.