Multistage hybrid active appearance model matching: Segmentation of left and right ventricles in cardiac MR images

A fully automated approach to segmentation of the left and right cardiac ve ntricles from magnetic resonance (MR) images is reported. A novel multistag e hybrid appearance model methodology is presented in which a hybrid active shape model/active appearance model (AAM) stage helps avoid local minima o f the matching function. This yields an overall more favorable matching res ult, An automated initialization method is introduced making the approach f ully automated. Our method was trained in a set of 102 MR images and tested in a separate s et of 60 images. In all testing cases, the matching resulted in a visually plausible and accurate mapping of the model to the image data, Average sign ed border positioning errors did not exceed 0.3 mm in any of the three dete rmined contours-left-ventricular (LV) epicardium, LV and right-ventricular (RV) endocardium. The area measurements derived from the three contours cor related well with the independent standard (r = 0.96, 0.96, 0.90), with slo pes and intercepts of the regression lines close to one and zero, respectiv ely. Testing the reproducibility of the method demonstrated an unbiased per formance with small range of error as assessed via Bland-Altman statistic. In direct border positioning error comparison, the multistage method signif icantly outperformed the conventional AAM (p < 0.001), The developed method promises to facilitate fully automated quantitative analysis of LV and RV morphology and function in clinical setting.