EUVIMEDEUVIMED entdecken
Crossref

Robust Medical Image Segmentation for Consumer-Grade Tele-Diagnosis: A Geometric-Coupled Dual-Frequency Network

Shaoqiang Wang, Shuo Sun, Guiling Shi, Yuanyuan Zhang, Qiang Li, Xueyan Chen, Yawu Zhao, Yuchen Wang

IEEE Transactions on Consumer Electronics · 2026-08 · 72 (3) · 6007-6021

DOI: 10.1109/tce.2026.3675453 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2026-08. Verfasst von Shaoqiang Wang und 7 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 38 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In the rapidly evolving consumer healthcare sector, the proliferation of portable imaging devices—such as smartphone dermatoscope accessories and handheld ultrasound probes—has significantly advanced home health monitoring. Therefore, achieving precise medical image segmentation is crucial for providing reliable tele-diagnosis in home settings. However, diagnosing common conditions like skin lesions and breast nodules often presents significant challenges due to features such as blurred boundaries, low contrast, and high internal texture heterogeneity. To address these challenges, we designed a Geometric-Coupled Dual-Frequency Domain Network (DuF-Net). Our network utilizes wavelet transforms to adaptively suppress high-frequency noise, prioritizing the retention of diagnostically valuable texture details during feature extraction. In the decoding stage, we further introduce a spectral gated aggregation module. This module guides feature upsampling by leveraging globally compressed frequency information from the discrete cosine transform, effectively resolving boundary blurring caused by poor image quality. Finally, geometric coupling supervision enhances semantic consistency through multi-task learning, boosting overall segmentation accuracy. Extensive experiments demonstrate DuF-Net’s robust segmentation performance across six public datasets. Notably, our method achieves significant segmentation gains and maintains precision on low-quality images, making it well-suited for delivering clinically accurate diagnostic services in home healthcare settings. Our code is open-sourced at: https://github.com/814398794a-ops/DuF-Net

Zitieren

APA

Wang, S., Sun, S., Shi, G., Zhang, Y., Li, Q., Chen, X., Zhao, Y., & Wang, Y. (2026). Robust Medical Image Segmentation for Consumer-Grade Tele-Diagnosis: A Geometric-Coupled Dual-Frequency Network. IEEE Transactions on Consumer Electronics, 72(3), 6007-6021. https://doi.org/10.1109/tce.2026.3675453

Vancouver

Wang S, Sun S, Shi G, Zhang Y, Li Q, Chen X, et al. Robust Medical Image Segmentation for Consumer-Grade Tele-Diagnosis: A Geometric-Coupled Dual-Frequency Network. IEEE Transactions on Consumer Electronics. 2026;72(3):6007-6021. doi:10.1109/tce.2026.3675453

BibTeX
@article{Wang2026,
  author = {Wang, Shaoqiang and Sun, Shuo and Shi, Guiling and Zhang, Yuanyuan and Li, Qiang and Chen, Xueyan and Zhao, Yawu and Wang, Yuchen},
  title = {Robust Medical Image Segmentation for Consumer-Grade Tele-Diagnosis: A Geometric-Coupled Dual-Frequency Network},
  journal = {IEEE Transactions on Consumer Electronics},
  year = {2026},
  volume = {72},
  number = {3},
  pages = {6007--6021},
  doi = {10.1109/tce.2026.3675453}
}
Kostenlos starten →