Abstract
EvaluatingsyntheticmedicalimagesremainschallengingbecausewidelyusedmetricssuchasFréchetInceptionDistance(FID)and Multi-Scale Structural Similarity (MS-SSIM) primarily assess perceptual or feature-space similarity rather than anatomical plausibility. We presentWAND-CXR,amorphometricWassersteinmetricforevaluating theanatomicalplausibilityofsyntheticchestradiographsatthedataset level. The proposed approach extracts multi-organ morphometric measurements from segmentation masks and compares their distributions using the empirical 2-Wasserstein distance. Experiments on real and synthetic chest X-ray cohorts demonstrate that WAND-CXR provides clinicallymeaningfulstratificationofanatomicalvariation,distinguishing within-cohortvariability,cross-datasetdifferences,andgeometry-altering pathologies.AdditionalmetricanalysisshowsthattheselectedanatomicalmeasurementsprovidecomplementaryinformationandthatWAND- CXR remains robust to the removal of individual anatomical feature groups.Theseresultssuggestthatmorphometricdistributioncomparison offers a clinically interpretable and complementary perspective to conventionalimage-levelandsemanticevaluationmetricsforsyntheticchest radiographs. Code is available at: https://github.com/FrexG/wand-cxr.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/SASHIMI_057.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: https://openreview.net/forum?id=P3oDX7A3QH
BibTex
@InProceedings{GelFra_WANDCXR_MICCAISAT2026,
author = { Gelana, Fraol AND Hines, Andrew AND Becerra, Helard},
title = { { WAND-CXR: A Morphometric Wasserstein Metric for Evaluating Anatomical Plausibility in Synthetic Chest Radiographs } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17258},
month = {pending},
page = {pending}
}