Abstract
Flow analysis carried out using phase contrast cardiac magnetic resonanceimaging (PC-CMR) enables the quantification of important parameters that areused in the assessment of cardiovascular function. An essential part of thisanalysis is the identification of the correct CMR views and quality control(QC) to detect artefacts that could affect the flow quantification. We proposea novel deep learning based framework for the fully-automated analysis of flowfrom full CMR scans that first carries out these view selection and QC stepsusing two sequential convolutional neural networks, followed by automatic aortaand pulmonary artery segmentation to enable the quantification of key flowparameters. Accuracy values of 0.958 and 0.914 were obtained for viewclassification and QC, respectively. For segmentation, Dice scores were$>$0.969 and the Bland-Altman plots indicated excellent agreement betweenmanual and automatic peak flow values. In addition, we tested our pipeline onan external validation data set, with results indicating good robustness of thepipeline. This work was carried out using multivendor clinical data consistingof 986 cases, indicating the potential for the use of this pipeline in aclinical setting.