EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Dense U-Net for Prostate Zonal Segmentation on T2-Weighted MRI

Tawhidur Rahman, Shamsad Binte Ehsan

Journal of Computational Science and Applications · 11.12.2025 · 2 (2) · 11-20

DOI: 10.51846/jcsa.v2i2.4683 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Computational Science and Applications, online erschienen am 11.12.2025. Verfasst von Tawhidur Rahman und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The magnetic resonance imaging (MRI) offers high resolution of the visualization of the prostate and its zonal structure, and more importantly, it is non-invasive important background knowledge in the diagnosis and treatment plan of prostate cancer. Nevertheless, prostate automatic segmentation and its regions, particularly peripheral and central regions, is a difficult one because of blurred edges, tissue heterogeneity, and great inter-patient anatomical variability. In order to overcome these issues, we suggest the Dense U-Net, a new convolutional neural network incorporating the efficient feature propagation of DenseNet and the exact U-Net localization capability. The training and testing of the model were done on 141 and 47 T2-weighted MRI images, respectively, with four-fold cross-validation. The network was tested on quality and poor segmentation masks to determine the robustness. The better one was Dense-2 U-Net, which outperformed the standard U-Net with an average Dice Similarity Coefficient of 92.3(±0.7)% and the overall accuracy of 99% in the entire prostate, central zone, and peripheral zone. These findings show that Dense-2 U-Net improves performance of segmentation, especially in the challenging peripheral zone, with imperfect annotations. The current work highlights that the potential of Dense-2 U-Net, as a powerful computer-aided diagnosis system, can be used to accelerate clinical workflows and enhance patient outcome in MRI-based prostate imaging.

Zitieren

APA

Rahman, T., & Ehsan, S. B. (2025). Dense U-Net for Prostate Zonal Segmentation on T2-Weighted MRI. Journal of Computational Science and Applications, 2(2), 11-20. https://doi.org/10.51846/jcsa.v2i2.4683

Vancouver

Rahman T, Ehsan SB. Dense U-Net for Prostate Zonal Segmentation on T2-Weighted MRI. Journal of Computational Science and Applications. 2025;2(2):11-20. doi:10.51846/jcsa.v2i2.4683

BibTeX
@article{Rahman2025,
  author = {Rahman, Tawhidur and Ehsan, Shamsad Binte},
  title = {Dense U-Net for Prostate Zonal Segmentation on T2-Weighted MRI},
  journal = {Journal of Computational Science and Applications},
  year = {2025},
  volume = {2},
  number = {2},
  pages = {11--20},
  doi = {10.51846/jcsa.v2i2.4683}
}
Kostenlos starten →