Objective: To assess combined analysis of coronary arteries and delayed myocardial contrast enhancement based on co-registration of coronary CT angiography and late-phase CT and automatic segmentation.
Materials and methods: Co-registration and late enhancement segmentation were applied to coronary CT angiography and late-phase CT images from six pigs with acute myocardial infarction (MI) and six patients with chronic MI. MI size was quantified by manual delineation, the established 3SD method, and a new mixture model approach. Correspondence between coronary artery lesions and MI was assessed visually from fused segmentation results.
Results: Co-registration was successful in all cases. There was substantial agreement in the number of segments diagnosed with MI, comparing manual delineation and the mixture model for animal (κ = 0.839) and patient studies (κ = 0.770). There were no significant differences between the two methods (P > 0.05). In patients there was a discrepancy between the segmental distribution of MI and empirical coronary artery perfusion in 10/96 segments when compared with the true coronary branching pattern.
Conclusion: The mixture model approach is well suited for automated assessment of MI size from late-phase cardiac CT. Fusion imaging eliminates the need for empirical knowledge of the anatomical relationship between the coronary artery lesion and the area of myocardial ischaemia.