Optimizing Aortic Segmentation with an Innovative Quality Assessment: The Role of Global Sensitivity Analysis

Precise aortic vessel tree segmentation is critical in the continuously evolving medical imaging domain. This study highlights the role of global sensitivity analysis in stimulating innovation in quality assessment techniques for aortic segmentation. In this methodology paper, we propose a novel method that integrates global sensitivity analysis with data augmentation techniques, aiming to enhance the reliability and robustness of segmentation algorithms. This approach aims to quantify the challenges posed by image variations and aspires to establish a methodology capable of managing a spectrum of image scenarios. The study also explores the implications of achieving accurate segmentations for clinical monitoring and computational fluid dynamics simulations of the aortic vessel tree. The presented approach was used for the final ranking of the MICCAI 2023 SEG.A. challenge to account for image variations in evaluating the submitted algorithms.