Purpose: To propose and validate anatomical criteria for the manual segmentation of right and left ventricles on cardiac MRI, to evaluate the interobserver reproducibility and tho evaluate the ease of implementation.
Materials and methods: Criteria for the segmentation and the choice of the basal slices were defined based on a review of the literature and cardiac anatomy. Cardiac volumetric measurements obtained from 20 consecutive patients using these criteria were compared to measurements obtained in reference to the orthogonal planes, and to those obtained with the classical criteria described in the literature. The interobserver reproducibility was statistically assessed for 6 observers (3 experts and 3 residents with no experience in cardiac imaging). The ease of implementation was determined by the post-processing time.
Results: Using orthogonal planes as a reference, the new criteria for the selection of the basal slice resulted in improved interobserver agreement and correlation compared to the classical criteria. Interobserver agreement was excellent (intra-class coefficient>0.93 for the left ventricle,>0.86 for the right ventricle). The mean post-processing time was similar or inferior to results from the literature.
Conclusion: The proposed criteria allow easy volumetric assessment of both ventricles, while providing reproducible measurements.