False-Positive Scar Control in Scribble-Supervised LGE-CMR Segmentation Using an Asymmetric Intensity Prior
Year: 2026
Type: article
Abstract: Background: Scribble-supervised LGE-CMR segmentation can overestimate myocardial scar. Methods: We introduced an asymmetric intensity prior that penalizes predicted scar darker than a scribble-derived reference. We evaluated 36 patient-disjoint test cases using reference myocardium-guided crops and performed separate reference-free central-crop and automatic-myocardial-wall analyses. The main prior-specific comparison matched the 2.5D backbone with and without the prior across three seeds. Results: On the pre-cropped images, mean Dice increased from 0.5328±0.0108 to 0.5472±0.0060, while recall fell from 0.6924 to 0.6399. The predicted-to-reference scar ratio fell from 1.6796±0.1834 to 1.4039±0.1389, and absolute scar-volume error fell from 21.09±5.27 to 14.08±2.47 mL on the common crop grid. However, restriction to the reference myocardial wall increased absolute percent-LV-scar error from 5.55±1.14 to 6.58±0.99 percentage points. With additional healthy-myocardium background points (seed 2022), the matched prior model reduced Pred/GT from 1.5789 to 1.1500 but Dice fell from 0.5441 to 0.5366. With one fixed-rule reference-free central crop and retraining (seed 2022), Dice was 0.4892 without and 0.5047 with the prior. In a paired automatic-wall analysis, the prior did not improve three-seed wall-restricted burden error (6.14 versus 8.16 percentage points). Conclusions: The prior reduced full-output overestimation on the pre-cropped images, but did not improve intramyocardial burden agreement or retain a Dice advantage in the single-seed healthy-background sensitivity analysis. Its effects depend on localization and wall quality and require independent prospective reference-free evaluation.
DOI: 10.3390/jcdd13110605
Authors: Xin Lan(Conceptualization, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – original draft), Xinchen Xie(Formal analysis, Validation, Writing – review & editing), Jinwei Dong(Writing – review & editing), Chenming Zhao(Data curation, Investigation), Qitong Zhang(Investigation, Software, Validation), Wangbin Ding(Conceptualization, Methodology, Project administration, Resources, Supervision, Writing – review & editing), Shimei Fu(Data curation, Funding acquisition, Resources, Supervision, Writing – review & editing)
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $3,258
Funder: Fujian Medical University
Awards: G6995357663, G9001396081
OpenAlex ID: W7235453638
