Comparison of machine learning and deep learning for view identification from cardiac magnetic resonance images.

BACKGROUND: Artificial intelligence is increasingly utilized to aid in the interpretation of cardiac magnetic resonance (CMR) studies. One of the first steps is the identification of the imaging plane depicted, which can be achieved by both deep learning (DL) and classical machine learning (ML) techniques without user input. We aimed to compare the accuracy of ML and DL for CMR view classification and to identify potential pitfalls during training and testing of the algorithms. METHODS: To train our DL and ML algorithms, we first established datasets by retrospectively selecting 200 CMR cases. The models were trained using two different cohorts (passively and actively curated) and applied data augmentation to enhance training. Once trained, the models were validated on an external dataset, consisting of 20 cases acquired at another center. We then compared accuracy metrics and applied class activation mapping (CAM) to visualize DL model performance. RESULTS: The DL and ML models trained with the passively-curated CMR cohort were 99.1% and 99.3% accurate on the validation set, respectively. However, when tested on the CMR cases with complex anatomy, both models performed poorly. After training and testing our models again on all 200 cases (active cohort), validation on the external dataset resulted in 95% and 90% accuracy, respectively. The CAM analysis depicted heat maps that demonstrated the importance of carefully curating the datasets to be used for training. CONCLUSIONS: Both DL and ML models can accurately classify CMR images, but DL outperformed ML when classifying images with complex heart anatomy.

Cell-Laden Gradient Hydrogel Scaffolds for Neovascularization of Engineered Tissues.

Gradients in mechanical properties, physical architecture and biochemical composition exist in a variety of complex tissues, yet 3D in vitro models that enable investigation of these cues on cellular processes, especially those contributing to vascularization of engineered tissues are limited. Here, a photopolymerization approach to create cell-laden hydrogel biomaterials with decoupled and combined gradients in modulus, immobilized cell adhesive peptide (RGD) concentration, and proteolytic degradation enabling spatial encapsulation of vascular spheroids is reported to elucidate their impact on vascular sprouting in 3D culture. Vascular spheroids encapsulated in these gradient scaffolds exhibit spatial variations in total sprout length. Scaffolds presenting an immobilized RGD gradient promote biased vascular sprouting toward increasing RGD concentration. Importantly, biased sprouting is found to be dependent on immobilized RGD gradient characteristics, including magnitude and slope, with increases in these factors contributing to significant enhancements in biased sprouting responses. Conversely, reduction in biased sprouting responses is observed in combined gradient scaffolds possessing opposing gradients in RGD and modulus. The presented work is the first to demonstrate the use of a cell-laden biomaterial platform to systematically investigate the role of multiple scaffold gradients as well as gradient slope, magnitude and orientation on vascular sprouting responses in 3D culture.