SEG.A.2023
- Machine Learning
SEG.A.2023 - Segmentation of the Aorta
Automatic segmentation of the aorta vessel tree.
2023-2024
Python, Pytorch, Lightning
Medical
AI Tool
About the Project
The project aimed at the automatic segmentation of the aortic vessel tree (AVT). The images came from a contrast-enhanced computed tomography scanner and depicted the AVT along with their corresponding manually performed segmentations from different hospitals. The prepared models were evaluated based on a carefully selected test set, using the Dice Similarity Coefficient (DSC) and Hausdorff Distance (HD). It was crucial for our AVT reconstructions to be artifact-free for subsequent visualization and use in hemodynamic simulation. 🩺💡
Team
Due to its interdisciplinary nature, the project required the knowledge of many experts. The current team composition allowed for capturing and solving its key problems. Collaboration among people dealing with algorithms, medicine, system design, cloud solutions, and, of course, artificial intelligence allowed for achieving satisfactory results and trying new approaches.
Patryk Rygiel
AI Engineer
Wojciech Cebula
Head of ML 23/24
Jakub Chojnacki
AI Engineer