Journal

Exploring the limit of image resolution for human expert classification of vascular ultrasound images in giant cell arteritis and healthy subjects: the GCA-US-AI project

Claus-Jürgen Bauer, Stavros Chrysidis, Christian Dejaco, Matthew J. Koster, Minna J. Kohler, Sara M. Monti, Wolfgang Schmidt, Chetan B. Mukhtyar, Pantelis Karakostas, Marcin Milchert, Cristina Ponte, Christina Duftner, Eugenio de Miguel, Alojzija Hočevar, Annamaria Iagnocco, Lene Terslev, Uffe Møller Døhn, Berit Dalsgaard Nielsen, Aaron Juche, Luca Seitz, Kresten Krarup Keller, Rositsa Valerieva Karalilova, Thomas D. Daikeler, Sarah Louise Mackie, Karina D. Torralba, Kornelis S. M. van der Geest, Dennis Boumans, Philipp Bösch, Alessandro Tomelleri, Markus Aschwanden, Tanaz A. Kermani, Andreas P. Diamantopoulos, Ulrich Fredberg, Nevsun İnanç, Simon Michael Petzinna, Shadi Albarqouni, Charlotte Behning, Valentin Schäfer

Annals of the rheumatic diseases, 2025

PublisherPDFInteractive explainer · soonBibTeX

Abstract

OBJECTIVES Prompt diagnosis of giant cell arteritis (GCA) with ultrasound is crucial for preventing severe ocular and other complications, yet expertise in ultrasound performance is scarce. The development of an artificial intelligence (AI)-based assistant that facilitates ultrasound image classification and helps to diagnose GCA early promises to close the existing gap. In the projection of the planned AI, this study investigates the minimum image resolution required for human experts to reliably classify ultrasound images of arteries commonly affected by GCA for the presence or absence of GCA.

METHODS Thirty-one international experts in GCA ultrasonography participated in a web-based exercise. They were asked to classify 10 ultrasound images for each of 5 vascular segments as GCA, normal, or not able to classify. The following segments were assessed: (1) superficial common temporal artery, (2) its frontal and (3) parietal branches (all in transverse view), (4) axillary artery in transverse view, and 5) axillary artery in longitudinal view. Identical images were shown at different resolutions, namely 32 × 32, 64 × 64, 128 × 128, 224 × 224, and 512 × 512 pixels, thereby resulting in a total of 250 images to be classified by every study participant.

RESULTS Classification performance improved with increasing resolution up to a threshold, plateauing at 224 × 224 pixels. At 224 × 224 pixels, the overall classification sensitivity was 0.767 (95% CI, 0.737-0.796), and specificity was 0.862 (95% CI, 0.831-0.888).

CONCLUSIONS A resolution of 224 × 224 pixels ensures reliable human expert classification and aligns with the input requirements of many common AI-based architectures. Thus, the results of this study substantially guide projected AI development.

Interactive explainer

Coming soon. An interactive Slidev walkthrough of this paper — with live, explorable figures — will appear here. Authors: set slidev: in this page's front matter to the deployed deck URL.

Research themes

Medical ImagingResponsible AI

Lab authors

Claus-Jürgen BauerShadi AlbarqouniValentin Schäfer
Cite (BibTeX)
@article{bauer2025exploring,
 author = {Bauer, Claus-Juergen and Chrysidis, Stavros and Dejaco, Christian and Koster, Matthew J and Kohler, Minna J and Monti, Sara and Schmidt, Wolfgang A and Mukhtyar, Chetan B and Karakostas, Pantelis and Milchert, Marcin and others},
 journal = {Annals of the rheumatic diseases},
 publisher = {Elsevier},
 title = {Exploring the limit of image resolution for human expert classification of vascular ultrasound images in giant cell arteritis and healthy subjects: the GCA-US-AI project},
 year = {2025}
}