Attention-driven tree-structured convolutional LSTM for high dimensional data understanding

Modeling the sequential information of image sequences has been a vital step various vision tasks and convolutional long short-term memory (ConvLSTM) demonstrated its superb performance in such spatiotemporal problems. Nevertheless, hierarchical data structures significant amount (e.g., human body parts vessel/airway tree biomedical images) cannot be properly modeled by models. Thus, ConvLSTM is not suitable for tree-structured analysis. In order to address these limitations, we present models analysis which can trained end-to-end. To demonstrate effectiveness proposed model, segmentation framework consists an attention fully network (FCN) model. The extensively validated on four large-scale coronary artery datasets. results efficiency method.