Yuan Jin, Antonio Pepe, Gian Marco Melito, Yuxuan Chen, Gege Ma, Yunsu Byeon, Hyeseong Kim, Kyungwon Kim, Doohyun Park, Euijoon Choi, Dosik Hwang, Andriy Myronenko, Dong Yang, Yufan He, Daguang Xu, Ayman El-Ghotni, Mohamed Nabil, Hossam El-Kady, Ahmed Ayyad, Amr Nasr, Marek Wodzinski, Henning Müller, Hyeongyu Kim, Yejee Shin, Abbas Khan, Muhammad Asad, Alexander Zolotarev, Caroline Roney, Anthony Mathur, Martin Benning, Gregory Slabaugh, Theodoros Panagiotis Vagenas, Konstantinos Georgas, George K Matsopoulos, Jihan Zhang, Zhen Zhang, Liqin Huang, Christian Mayer, Heinrich Mächler, Jan Egger
The automated analysis of the aortic vessel tree (AVT) from computed tomography angiography (CTA) is crucial for clinical applications but lacks shared, high-quality data. To address this, we launched the SEG.A. challenge, introducing a large, public, multi-institutional dataset for AVT segmentation and benchmarking automated algorithms. The challenge results showed a strong trend toward deep learning, with 3D U-Net architectures being most effective. The winning solution used an ensemble-based strategy, highlighting the value of model ensembling for robust AVT segmentation. Performance strongly correlated with algorithmic design, notably the use of customized post-processing and training data characteristics. This initiative establishes a new performance benchmark and provides a lasting resource to drive future innovation toward robust, clinically translatable AVT analysis tools.