EUVIMEDEUVIMED entdecken
Crossref

An Attention-Based Multi-Headed U-Net for Nuclei Segmentation Using Shearlet-Based Unsharp Masking

Shivam Mishra, Amit Vishwakarma, Anil Kumar

IEEE Transactions on Consumer Electronics · 2026-02 · 72 (1) · 416-427

DOI: 10.1109/tce.2025.3645832 ↗

Kurz gefasst

Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2026-02. Verfasst von Shivam Mishra und 2 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref) und nennt 57 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Automated nuclei segmentation plays a vital role in computer-assisted digital pathology, facilitating early disease diagnosis and cellular analysis in consumer healthcare applications. However, challenges such as variations in nuclei morphology, indistinct boundaries, and multi-modal imaging hinder segmentation accuracy. To address these issues, we propose an Attention-based Multi-Headed U-Net (AM-UNet), optimized for consumer-grade digital pathology and medical imaging devices. The model employs multi-resolution feature extraction using convolutional heads with varying kernel sizes, while an attention mechanism enhances focus on nuclei morphology and boundaries. Additionally, we introduce a Shearlet-Based Unsharp Masking (SBUM) technique to amplify high-frequency image details, improving clarity in low-power and real-time medical imaging applications. To further enhance segmentation accuracy, a hybrid loss function combining Intersection over Union (IoU) loss, Log-Dice loss, and Binary Cross-Entropy (BCE) loss is proposed. Experimental results on DSB-2018 and TNBC datasets demonstrate that AM-UNet surpasses state-of-the-art methods in segmentation performance and computational efficiency, making it well-suited for portable and AI-driven consumer healthcare technologies.

Zitieren

APA

Mishra, S., Vishwakarma, A., & Kumar, A. (2026). An Attention-Based Multi-Headed U-Net for Nuclei Segmentation Using Shearlet-Based Unsharp Masking. IEEE Transactions on Consumer Electronics, 72(1), 416-427. https://doi.org/10.1109/tce.2025.3645832

Vancouver

Mishra S, Vishwakarma A, Kumar A. An Attention-Based Multi-Headed U-Net for Nuclei Segmentation Using Shearlet-Based Unsharp Masking. IEEE Transactions on Consumer Electronics. 2026;72(1):416-427. doi:10.1109/tce.2025.3645832

BibTeX
@article{Mishra2026,
  author = {Mishra, Shivam and Vishwakarma, Amit and Kumar, Anil},
  title = {An Attention-Based Multi-Headed U-Net for Nuclei Segmentation Using Shearlet-Based Unsharp Masking},
  journal = {IEEE Transactions on Consumer Electronics},
  year = {2026},
  volume = {72},
  number = {1},
  pages = {416--427},
  doi = {10.1109/tce.2025.3645832}
}
Kostenlos starten →