Non-compact myocardium assessment by cardiac magnetic resonance: dependence on image analysis method.

To compare image analysis methods for the assessment of left ventricle non-compaction from cardiac magnetic resonance (CMR) imaging. CMR images were analyzed in 20 patients and 10 normal subjects. A reference model of the MR signal was introduced and validated based on image data. Non-compact (NC) myocardium size and distribution were assessed by tracing a single, continuous contour delimiting trabeculated region (Jacquier) or by one-by-one selection of trabeculae (Grothoff). The global non-compact/compact (NC/C) ratio, the NC mass, and the segmental NC/C ratio were assessed. Results were compared with the reference model. A significant difference between Grothoff and Jacquier approaches in the estimation of NC/C ratio (32.08 ± 6.63 vs. 19.81 ± 5.72, p < 0.0001) and NC mass (26.59 ± 8.36 vs. 14.15 ± 5.73 g/m2, p < 0.0001) was found. The Grothoff approach better matches the expected signal distribution. Inter-observer reproducibility of both Grothoff and Jacquier methods was adequate (9.71 and 8.22%, respectively) with no significant difference between observers. Jacquier and Grothoff approaches are not interchangeable so that specific diagnostic thresholds should be used for different image analysis methods. Grothoff method seems to better capture the true extension of trabeculated tissue.