Skip to main content
Fig. 4 | Journal of Cardiovascular Magnetic Resonance

Fig. 4

From: MVnet: automated time-resolved tracking of the mitral valve plane in CMR long-axis cine images with residual neural networks: a multi-center, multi-vendor study

Fig. 4

Accuracy of each model (MVnet) trained and evaluated on its own dataset, by the mean a Euclidean and b angular distance error, and the agreement with ICC in c MAPSE, d LV e’ and e MV displacement, stratified by the output of the first stage (stage 1), second stage (stage 1+2), and an iteration of the second stage (stage 1+2+2). For (a, b), each bar represents the mean, and error bar the standard deviation of each accuracy metric. For (cde), each bar represents the ICC, and error bar the confidence interval (95%) of each accuracy metric. The output of the iteration (stage 1+2+2) achieved the best accuracy and was chosen for the proposed workflow. ICC intra-class correlation coefficient, MAPSE mitral annular plane systolic excursion, LV left ventricle, MV mitral valve

Back to article page