Background Deep learning algorithms are increasingly used for automatic medical imaging analysis and cardiac chamber segmentation. Especially in congenital heart disease, obtaining a sufficient number of training images and data anonymity issues remain of concern. Methods Progressive generative adversarial networks (PG-GAN) were trained on cardiac magnetic resonance imaging (MRI) frames from a nationwide prospective study to generate synthetic MRI frames. These synthetic frames were subsequently used to train segmentation networks (U-Net) and the quality of the synthetic training images, as well as the performance of the segmentation network was compared to U-Net-based solutions trained entirely on patient data. Results Cardiac MRI data from 303 patients with Tetralogy of Fallot were used for PG-GAN training. Using this model, we generated 100,000 synthetic images with a resolution of 256 × 256 pixels in 4-chamber and 2-chamber views. All synthetic samples were classified as anatomically plausible by human observers. The segmentation performance of the U-Net trained on data from 42 separate patients was statistically significantly better compared to the PG-GAN based training in an external dataset of 50 patients, however, the actual difference in segmentation quality was negligible (< 1% in absolute terms for all models). Conclusion We demonstrate the utility of PG-GANs for generating large amounts of realistically looking cardiac MRI images even in rare cardiac conditions. The generated images are not subject to data anonymity and privacy concerns and can be shared freely between institutions. Training supervised deep learning segmentation networks on this synthetic data yielded similar results compared to direct training on original patient data.

No abstract

Introduction: Automated echocardiography image interpretation has the potential to transform clinical practice. However, neural networks developed in general cohorts may underperform in the setting of altered cardiac anatomy. Methods: Consecutive echocardiographic studies of patients with congenital or structural heart disease (C/SHD) were used to validate an existing convolutional neural network trained on 14,035 echocardiograms for automated view classification. In addition, a new convolutional neural network for view classification was trained and tested specifically in patients with C/SHD. Results: Overall, 9793 imaging files from 262 patients with C/SHD (mean age 49 years, 60% male) and 62 normal controls (mean age 45 years, 50.0% male) were included. Congenital diagnoses included among others, tetralogy of Fallot (30), Ebstein anomaly (18) and transposition of the great arteries (TGA, 48). Assessing correct view classification based on 284,250 individual frames revealed that the non-congenital model had an overall accuracy of 48.3% for correct view classification in patients with C/SHD compared to 66.7% in patients without cardiac disease. Our newly trained convolutional network for echocardiographic view detection based on over 139,910 frames and tested on 35,614 frames from C/SHD patients achieved an accuracy of 76.1% in detecting the correct echocardiographic view. Conclusions: The current study is the first to validate view classification by neural networks in C/SHD patients. While generic models have acceptable accuracy in general cardiology patients, the quality of image classification is only modest in patients with C/SHD. In contrast, our model trained in C/SHD achieved a considerably increased accuracy in this particular cohort.

Aims To test the hypothesis that deep learning (DL) networks reliably detect pulmonary arterial hypertension (PAH) and provide prognostic information. Methods and results Consecutive patients with PAH, right ventricular (RV) dilation (without PAH), and normal controls were included. An ensemble of deep convolutional networks incorporating echocardiographic views and estimated RV systolic pressure (RVSP) was trained to detect (invasively confirmed) PAH. In addition, DL-networks were trained to segment cardiac chambers and extracted geometric information throughout the cardiac cycle. The ability of DL parameters to predict all-cause mortality was assessed using Cox-proportional hazard analyses. Overall, 450 PAH patients, 308 patients with RV dilatation (201 with tetralogy of Fallot and 107 with atrial septal defects) and 67 normal controls were included. The DL algorithm achieved an accuracy and sensitivity of detecting PAH on a per patient basis of 97.6 and 100%, respectively. On univariable analysis, automatically determined right atrial area, RV area, RV fractional area change, RV inflow diameter and left ventricular eccentricity index (P &lt; 0.001 for all) were significantly related to mortality. On multivariable analysis DL-based RV fractional area change (P &lt; 0.001) and right atrial area (P = 0.003) emerged as independent predictors of outcome. Statistically, DL parameters were non-inferior to measures obtained manually by expert echocardiographers in predicting prognosis. Conclusion The study highlights the utility of DL algorithms in detecting PAH on routine echocardiograms irrespective of RV dilatation. The algorithms outperform conventional echocardiographic evaluation and provide prognostic information at expert-level. Therefore, DL methods may allow for improved screening and optimized management of PAH.