EUVIMEDEUVIMED entdecken
Crossref

Adaptive ensemble loss and multi-scale attention in breast ultrasound segmentation with UMA-Net

Mohsin Furkh Dar, Avatharam Ganivada

Medical & Biological Engineering & Computing · 2025-06 · 63 (6) · 1697-1713

DOI: 10.1007/s11517-025-03301-5 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Medical & Biological Engineering & Computing, online erschienen am 23.01.2025. Verfasst von Mohsin Furkh Dar und 1 weiteren Autor:innen. Beteiligte Einrichtung: Artificial Intelligence Lab, School of Computer and Information Sciences, University of Hyderabad, Hyderabad, 500046, India. 20mcpc02@uohyd.ac.in. Hauptthemen laut MEDLINE: Ultrasonography, Mammary, Breast Neoplasms, Image Processing, Computer-Assisted, Breast. Die Arbeit wurde bisher 9-mal zitiert (Crossref) und nennt 40 Literaturangaben.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The generalization of deep learning (DL) models is critical for accurate lesion segmentation in breast ultrasound (BUS) images. Traditional DL models often struggle to generalize well due to the high frequency and scale variations inherent in BUS images. Moreover, conventional loss functions used in these models frequently result in imbalanced optimization, either prioritizing region overlap or boundary accuracy, which leads to suboptimal segmentation performance. To address these issues, we propose UMA-Net, an enhanced UNet architecture specifically designed for BUS image segmentation. UMA-Net integrates residual connections, attention mechanisms, and a bottleneck with atrous convolutions to effectively capture multi-scale contextual information without compromising spatial resolution. Additionally, we introduce an adaptive ensemble loss function that dynamically balances the contributions of different loss components during training, ensuring optimization across key segmentation metrics. This novel approach mitigates the imbalances found in conventional loss functions. We validate UMA-Net on five diverse BUS datasets-BUET, BUSI, Mendeley, OMI, and UDIAT-demonstrating superior performance. Our findings highlight the importance of addressing frequency and scale variations, confirming UMA-Net as a robust and generalizable solution for BUS image segmentation.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Ultrasonography, MammaryBreast NeoplasmsImage Processing, Computer-AssistedBreastHumansFemaleDeep LearningAlgorithms

Stichwörter der Autor:innen

Breast cancerDeep learningLoss functionSegmentationUltrasoundUnet

Autor:innen und Institutionen

  • Artificial Intelligence Lab, School of Computer and Information Sciences, University of Hyderabad, Hyderabad, 500046, India. 20mcpc02@uohyd.ac.in.
  • Artificial Intelligence Lab, School of Computer and Information Sciences, University of Hyderabad, Hyderabad, 500046, India.

Zitieren

APA

Dar, M. F., & Ganivada, A. (2025). Adaptive ensemble loss and multi-scale attention in breast ultrasound segmentation with UMA-Net. Medical & Biological Engineering & Computing, 63(6), 1697-1713. https://doi.org/10.1007/s11517-025-03301-5

Vancouver

Dar MF, Ganivada A. Adaptive ensemble loss and multi-scale attention in breast ultrasound segmentation with UMA-Net. Medical & Biological Engineering & Computing. 2025;63(6):1697-1713. doi:10.1007/s11517-025-03301-5

BibTeX
@article{Dar2025,
  author = {Dar, Mohsin Furkh and Ganivada, Avatharam},
  title = {Adaptive ensemble loss and multi-scale attention in breast ultrasound segmentation with UMA-Net},
  journal = {Medical & Biological Engineering & Computing},
  year = {2025},
  volume = {63},
  number = {6},
  pages = {1697--1713},
  doi = {10.1007/s11517-025-03301-5}
}
Kostenlos starten →