Correcting Aortic Diameter Measurement Errors from Severe Angulations

Joris Fournel, Charlotte Lu, Axel Bartoli, Aasa Feragen, Alexis Jacquier

Keywords: Computed Tomography Angiography, Maximum aortic diameter, Aneurysm, Aortic dissection, Deep Learning

TL;DR: When angulated aortas distort automatic diameter measurements, CorrectNet corrects them—bringing reliability back to surgical planning.

Abstract: Surgical decisions often hinge on the maximum aortic diameter—but current automated methods fail in anatomies with severe angulations. We show that sharp bends distort the aorta in orthogonal measurement planes, leading to unreliable values. Our model, CorrectNet, learns to segment around these distortions, restoring measurement reliability and strengthening surgical planning.

Submission Number: 20