Segmentation and morphological analysis of the left atria from CT images with machine-learning tools Segmentation and morphological analysis of the left atria from CT images with machine-learning tools

Segmentation and morphological analysis of the atria can provide biomarkers to help preventing the outcome of critical surgical interventions such as radiofrequency ablation in patients with atrial fibrillation. This project aims to apply advanced deep learning techniques to segment the left atria in patient-specific CT images and obtain robust shape indices to classify patients in different classes. A particular challenging case concerns the morphological characterization of the left atrial appendage. Tools like Shapeworks, spectral techniques (SPHARM), Deformetrica, and (geometrical) deep learning-based tools will be used. The project will be performed in collaboration with national and international clinicians from Hospital Clínic de Barcelona, Hospital Sant Pau and Righospitalet from Denmark.


Supervisors: Oscar Camara, Jordi Mill, Xabier Morales