Arrhythmogenic right ventricular cardiomyopathy/dysplasia (ARVC) is an inherited cardiomyopathy affecting approximately 1 in 1000 - 5000 individuals [1]. Ventricular structure (shape) in ARVC patients is an important feature of the disease, but is challenging to quantify. The aim of this study was to go beyond measures of volume and ejection fraction by automatically analysing structural abnormalities in ARVC patients from CMR using computational, quantifiable methods. A data-set of 27 patients diagnosed with ARVC following the 2010 Task Force Criteria were recruited from the Unit of genetic cardiac diseases, Department of Cardiology, Oslo University Hospital (OUH). Thirteen out of the 27 patients were male, with mean age plus/minus one standard deviation = 38 ± 14 years. All participants were scanned at OUH with a Siemens 1.5T scanner (8 with an Avanto scanner, 13 with a Sonata Vision scanner). The voxel size was between 1.33mm and 1.65mm, with slice thickness of 6mm for all subjects. Where possible, automatic methods for extracting the structure were used to minimise the bias in defining the ventricles in each subject, and fully automatic methods for pre-processing the data were developed to enable population-wide comparison and analysis of the RV shape. The shape models were extracted from the image sequences using image segmentation techniques. The shape models were pre-processed to firstly correct for slice misalignment, then to eliminate differences in orientation and pose due to differing image locations, and finally to account for temporal sampling of each image sequence. Once all subjects were aligned to a common space and temporal sampling, the end-diastolic (ED) and end-systolic (ES) phases were identified and the mean shape was computed individually on the surface model of both phases. Principal component analysis (PCA) was applied to extract the most common RV shapes in the ARVC population [2]. The mean (average) and modes (dominant shape patterns) in the data were computed from shape models representing the RV endocardium following a meta-modelling analysis, as summarised in Fig. 1. The first five RV shape modes are shown in Fig. 2 from different views to emphasise the largest shape difference observed for each mode (in terms of differences to the control group mean). Since the ES phase was found to be more indicative of shape correlations, only the ES phase is presented. The first mode, explaining 44% of the shape variance in the population, shows global dilation. The second mode, explaining 15% of the variance, shows elongation at the RV outflow tract. The third mode, explaining 10% of the variance, shows a tilting from the apex to the base at the septum. The fourth mode, explaining 7% of the variance, shows lengthening/shortening in the RV for this group. Finally, the fifth mode, explaining 5% of the variance, shows abnormal bulging at the RV inlet and outlet. The loadings of each mode for each subject, which quantifies structural abnormalities at a more regional level, are shown in Fig. 3 for each of the ARVC subjects. The loadings can be used to measure how “abnormal” each subject is with respect to the identified abnormalities defined from Fig. 2. CMR images of the patients with the largest loadings (circled in Fig. 3) are shown in Fig. 4 to visualise the abnormalities on the images directly.An automatic method for quantifying the RV shape in ARVC patients ispresented. The extracted RV shapes represented the most commonly known features in these patients, namely global RV dilation, elongation at the RVOT, tilting at the septum, shortening/lengthening of the ventricles, and bulging at the RV inlet and outlet. These shape abnormalities can now automatically be quantified in a new patient by measuring to define the degree of a given abnormality that is present in that specific patient.