arXiv reaDer
RotCAtt-TransUNet++: Novel Deep Neural Network for Sophisticated Cardiac Segmentation
Cardiovascular disease is a major global health concern, contributing significantly to global mortality. Accurately segmenting cardiac medical imaging data is crucial for reducing fatality rates associated with these conditions. However, current state-of-the-art (SOTA) neural networks, including CNN-based and Transformer-based approaches, face challenges in capturing both inter-slice connections and intra-slice details, especially in datasets featuring intricate, long-range details along the z-axis like coronary arteries. Existing methods also struggle with differentiating non-cardiac components from the myocardium, resulting in segmentation inaccuracies and the "spraying" phenomenon. To address these issues, we introduce RotCAtt-TransUNet++, a novel architecture designed for robust segmentation of intricate cardiac structures. Our approach enhances global context modeling through multiscale feature aggregation and nested skip connections in the encoder. Transformer layers facilitate capturing intra-slice interactions, while a rotatory attention mechanism handles inter-slice connectivity. A channel-wise cross-attention gate integrates multiscale information and decoder features, effectively bridging semantic gaps. Experimental results across multiple datasets demonstrate superior performance over current methods, achieving near-perfect annotation of coronary arteries and myocardium. Ablation studies confirm that our rotatory attention mechanism significantly improves segmentation accuracy by transforming embedded vectorized patches in semantic dimensional space.
updated: Mon Sep 09 2024 02:18:50 GMT+0000 (UTC)
published: Mon Sep 09 2024 02:18:50 GMT+0000 (UTC)
参考文献 (このサイトで利用可能なもの) / References (only if available on this site)
被参照文献 (このサイトで利用可能なものを新しい順に) / Citations (only if available on this site, in order of most recent)
Amazon.co.jpアソシエイト