Mini Review: Deep Learning for Atrial Segmentation From Late Gadolinium-Enhanced MRIs.

Segmentation and 3D reconstruction of the human atria is of crucial importance for precise diagnosis and treatment of atrial fibrillation, the most common cardiac arrhythmia. However, the current manual segmentation of the atria from medical images is a time-consuming, labor-intensive, and error-prone process. The recent emergence of artificial intelligence, particularly deep learning, provides an alternative solution to the traditional methods that fail to accurately segment atrial structures from clinical images. This has been illustrated during the recent 2018 Atrial Segmentation Challenge for which most of the challengers developed deep learning approaches for atrial segmentation, reaching high accuracy (>90% Dice score). However, as significant discrepancies exist between the approaches developed, many important questions remain unanswered, such as which deep learning architectures and methods to ensure reliability while achieving the best performance. In this paper, we conduct an in-depth review of the current state-of-the-art of deep learning approaches for atrial segmentation from late gadolinium-enhanced MRIs, and provide critical insights for overcoming the main hindrances faced in this task.

A V-Net Based Deep Learning Model for Segmentation and Classification of Histological Images of Gastric Ablation.

Gastric motility disorders are associated with bioelectrical abnormalities in the stomach. Recently, gastric ablation has emerged as a potential therapy to correct gastric dysrhythmias. However, the tissue-level effects of gastric ablation have not yet been evaluated. In this study, radiofrequency ablation was performed in vivo in pigs (n=7) at temperature-control mode (55-80°C, 5-10 s per point). The tissue was excised from the ablation site and routine H&E staining protocol was performed. In order to assess tissue damage, we developed an automated technique using a fully convolutional neural network to segment healthy tissue and ablated lesion sites within the muscle and mucosa layers of the stomach. The tissue segmentation achieved an overall Dice score accuracy of 96.18 ± 1.0 %, and Jacquard score of 92.77 ± 1.9 %, after 5-fold cross validation. The ablation lesion was detected with an overall Dice score of 94.16 ± 0.2 %. This method can be used in combination with high-resolution electrical mapping to define the optimal ablation dose for gastric ablation.Clinical Relevance-This work presents an automated method to quantify the ablation lesion in the stomach, which can be applied to determine optimal energy doses for gastric ablation, to enable clinical translation of this promising emerging therapy.