No abstract

No abstract

Objective The aim of the study is to compare structure tensor imaging (STI) with diffusion tensor imaging (DTI) of the sheep heart (approximately the same size as the human heart). Materials and methods MRI acquisition on three sheep ex vivo hearts was performed at 9.4 T/30 cm with a seven-element RF coil. 3D FLASH with an isotropic resolution of 150 µm and 3D spin-echo DTI at 600 µm were performed. Tensor analysis, angles extraction and segments divisions were performed on both volumes. Results A 3D FLASH allows for visualization of the detailed structure of the left and right ventricles. The helix angle determined using DTI and STI exhibited a smooth transmural change from the endocardium to the epicardium. Both the helix and transverse angles were similar between techniques. Sheetlet organization exhibited the same pattern in both acquisitions, but local angle differences were seen and identified in 17 segments representation. Discussion This study demonstrated the feasibility of high-resolution MRI for studying the myocyte and myolaminar architecture of sheep hearts. We presented the results of STI on three whole sheep ex vivo hearts and demonstrated a good correspondence between DTI and STI.

Cardiac electrophysiology (EP) models achieved good progress in simulating cardiac electrical activity. However numerical issues and computational times hamper clinical applicability of such models. Moreover, personalisation can still be challenging and model errors can be difficult to overcome. On the other hand, deep learning methods achieved impressive results but suffer from robustness issues in healthcare due to their lack of physiological knowledge. We propose a novel approach which is based on deep learning in order to replace numerical integration of partial differential equations. This has the advantage to directly learn spatio-temporal correlations, which increases stability. Moreover, once trained, solutions are very fast to compute. We present first results in state estimation based on few measurements and evaluate the forecasting power of the trained network. The proposed method performed very well on this preliminary evaluation. It opens up possibilities towards data-driven personalisation, to overcome model error by learning from the data.