Abstract
Nodule-to-capsule distance measurement in thyroid ultrasound provides important anatomical information for assessing the relationship between a thyroid nodule and the adjacent capsule. Recently, artificial intelligence-based methods have been explored to automate this measurement. However, segmentation-based methods rely on complete capsule delineation and are highly sensitive to local boundary errors, while landmark-based methods impose coordinate-level supervision on a single annotated landmark pair, incorrectly penalizing alternative anatomically valid pairs with equivalent measurement distances. To address these issues, we propose TNCMeasure (Thyroid Noduleto-Capsule Measure), an anatomy- and distance-aware vision-language framework that integrates: 1) a nodule-mask-prompted anatomical perception framework that uses the nodule segmentation mask as an explicit anatomical prompt to jointly localize the nodule- and capsuleside landmarks without requiring complete capsule segmentation; and 2) an anatomy-constrained geometric reward that evaluates candidate landmark pairs according to physical-distance accuracy and anatomical boundary validity, thereby accommodating multiple clinically equivalent landmark configurations. Experimental results on 113 held-out clinical thyroid ultrasound images demonstrate that TNCMeasure achieves a mean absolute error of 0.98 mm and an SDR@3mm of 93.81%, indicating its effectiveness for accurate and anatomically valid nodule-to-capsule distance measurement.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/MWM_108.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=zUBODLHMUq
BibTex
@InProceedings{CheLif_TNCMeasure_MICCAISAT2026,
author = { Chen, Lifan AND Chen, Haoyuan AND Ouyang, Xi AND Xue, Zhong AND Shen, Dinggang},
title = { { TNCMeasure: Anatomy-Guided and Geometry-Constrained Vision-Language Learning for Thyroid Nodule-to-Capsule Distance Measurement } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17262},
month = {pending},
page = {pending}
}