Abstract
The main pulmonary artery-to-ascending aorta diameter ratio (PA:A) is an established imaging biomarker for pulmonary hypertension and COPD exacerbation risk. However, manual measurement performed on axial CT slices is time-consuming, observer-dependent, and geometrically biased when the vessel is oriented obliquely to the imaging plane. We present a fully automated pipeline utilizing a multi-task attention U-Net to predict vessel segmentation and landmarks that guide centerline extraction, allowing PA:A computation on cross-sections strictly perpendicular to the 3D vessel centerline. Evaluated on 100 CTPA cases against a validated orthogonal reference, the pipeline achieved a clinically viable ratio mean absolute error (MAE) of 0.034, less than one-tenth of the standard 1.0 diagnostic threshold. End-to-end inference completes in 50.9 seconds on average, compared to approximately 5 minutes for manual workflows, establishing a robust computational foundation for high-throughput clinical screening.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/TIA_008.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: Not Available
BibTex
@InProceedings{HosRai_Pulmonary_MICCAISAT2026,
author = { Hossain, Raisa Bentay AND Gerard, Sarah E.},
title = { { Pulmonary Artery-to-Aorta Ratio Estimation via Landmark-Guided Orthogonal Slicing } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17269},
month = {pending},
page = {pending}
}