PhD Defense Zhou ZHAO
Atrial fibrillation is the most common heart rhythm disease. Due to a lack of understanding in matter of underlying atrial structures, current treatments are still not satisfying. Recently, with the popularity of deep learning, many segmentation methods based on deep learning have been proposed to analyze atrial structures, especially from late gadolinium-enhanced magnetic resonance imaging. However, two problems still occur: 1) segmentation results include the atrial-like background; 2) boundaries are very hard to segment. Most segmentation approaches design a specific network that mainly focuses on the regions, to the detriment of the boundaries. Therefore, in this dissertation, we propose two different methods to segment the heart, one two-stage and one end-to-end trainable method. For the two-stage method, it can be decomposed in three main steps: a localization step, a Gaussian-based contrast enhancement step, and a segmentation step. This architecture is supplied with a hybrid loss function that guides the network to study the transformation relationship between the input image and the corresponding label in a three-level hierarchy (pixel-, patch- and map-level), which is helpful to improve segmentation and recovery of the boundaries. We demonstrate the efficiency of our approach on three public datasets in terms of regional and boundary segmentations. For the end-to-end trainable method. we propose an attention full convolutional network framework based on the ResNet-101 architecture, which focuses on boundaries as much as on regions. The additional attention module is added to have the network pay more attention on regions and then to reduce the impact of the misleading similarity of neighboring tissues. We also use a hybrid loss composed of a region loss and a boundary loss to treat boundaries and regions at the same time. The efficiency of proposed approach is verified on three public datasets. Finally, for evaluating the fibrosis degree, we proposed two methods, one is to combine deep learning with morphology, and the other is to use deep learning directly. For the first method, we calculate the left atrial wall based on the segmentation results in the previous chapter by morphologically dilating, and then thresholds to evaluate the fibrosis degree. For the second method, we provide one cascaded UNet architecture and uses multi-modalities information to complete the segmentation of the myocardium, scar and edema. We demonstrate the efficiency of our approach on one public dataset.