External validation of multitask deep learning for carotid plaque detection: anatomical auxiliary tasks shape performance trade-offs
Journal of Medical Ultrasonics · 28.10.2026
DOI: 10.1007/s10396-026-01701-0 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Medical Ultrasonics, online erschienen am 28.10.2026. Verfasst von Haichao Shi und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Ultrasound Medicine, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, 310014, Z…. Die Arbeit nennt 45 Literaturangaben.
Abstract (Originalsprache)
Purpose
Carotid plaque screening using ultrasound is challenged by class imbalance and ambiguous lesion boundaries. However, how different anatomical auxiliary tasks influence carotid plaque detection performance and clinical operating characteristics remains insufficiently investigated. We aimed to externally evaluate multitask deep learning models with different anatomical auxiliary tasks and investigate their effects on clinical performance profiles.
Methods
We developed a multitask network incorporating three anatomical auxiliary segmentation targets (vessel lumen, intima-media thickness (IMT), and plaque), together with a classification-only baseline. The model was developed on 1297 images from 262 patients and evaluated on an external cohort of 1091 images from 263 patients. To ensure fair comparison, we froze the encoder of each variant and retrained an identical classification head on the internal training set, and then applied a fixed threshold determined via Youden index maximization for each model on the internal validation set to the external cohort.
Results
In external validation (263 patients, 94 plaque-positive), all multitask models outperformed the classification-only baseline (AUC 0.615, p 0.76). Despite comparable discrimination, they exhibited distinct clinical profiles: the IMT-guided model achieved the highest sensitivity (SEN 0.830), the lumen-guided model achieved the highest specificity (SPE 0.864), and the plaque-guided model offered more balanced performance (SEN 0.723, SPE 0.822). Radiofrequency balloon ablation significantly reduced external AUC for the lumen-guided (0.852 vs 0.790; p =0.0145) and plaque-guided (0.858 vs 0.796; p =0.0091) models, but not for the IMT-guided model ( p =0.2171), indicating task-dependent effects.
Conclusion
The choice of auxiliary task does not confer universal superiority in discrimination but instead determines the model’s clinical orientation, allowing task-specific adaptation to different screening or confirmatory scenarios.
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Department of Ultrasound Medicine, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, 310014, Zhejiang, China.
- School of Mathematical Sciences, Zhejiang University, Zijingang Campus, Yuhangtang Road, Hangzhou, Zhejiang, China.
- Department of Ultrasound Medicine, Haining Central Hospital, 758 Chang'an Road, Haining, Jiaxing, Zhejiang, China.
- Department of Ultrasound Medicine, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, 310014, Zhejiang, China. hehongfeng132@126.com.
Zitieren
APA
Shi, H., Lai, Z., Shen, F., Du, P., & He, H. (2026). External validation of multitask deep learning for carotid plaque detection: anatomical auxiliary tasks shape performance trade-offs. Journal of Medical Ultrasonics. https://doi.org/10.1007/s10396-026-01701-0
Vancouver
Shi H, Lai Z, Shen F, Du P, He H. External validation of multitask deep learning for carotid plaque detection: anatomical auxiliary tasks shape performance trade-offs. Journal of Medical Ultrasonics. 2026. doi:10.1007/s10396-026-01701-0
BibTeX
@article{Shi2026,
author = {Shi, Haichao and Lai, Zeyu and Shen, Fengqin and Du, Pei and He, Hongfeng},
title = {External validation of multitask deep learning for carotid plaque detection: anatomical auxiliary tasks shape performance trade-offs},
journal = {Journal of Medical Ultrasonics},
year = {2026},
doi = {10.1007/s10396-026-01701-0}
}
EUVIMED