EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-SA 4.0

Early Breast Cancer Detection for Mammogram Images using U-Net Deep Learning Model.

Hardi Hiwa, Alan Anwer Abdulla

Zanin Journal of Science and Engineering · 2025

DOI: 10.64362/zjse.182 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Zanin Journal of Science and Engineering, 2025. Verfasst von Hardi Hiwa und 1 weiteren Autor:innen. Die Arbeit nennt 36 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-SA 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Since manually diagnosing diseases and determining treatments is time-consuming, costly, the advancement of the integration of convolutional neural networks (CNN) and image processing in medical imaging has improved diagnostic performance, particularly in breast cancer detection, where early diagnosis remains critical for patient outcomes. However, dense breast tissue continues to affect mammographic mass detection by overlapping with normal breast structures, which increases false-positive rates in traditional segmentation methods and classical machine learning. These approaches often fail under such conditions since mass features closely resemble dense tissue, creating a critical gap in tumor detection and compromising further diagnosis and classification. This study presents a deep learning-based computer-aided diagnostic (CAD) system for early breast cancer detection. The proposed system includes three stages. In the preprocessing stage, mammogram images are resized to reduce computation time, enhanced using contrast-limited adaptive histogram equalization (CLAHE), and normalized to help the model converge faster. In the second stage, several data augmentations are applied, including rotation, flipping, scaling, brightness adjustment, and contrast modification. The third stage introduces a modified U-Net model. This architecture incorporates batch normalization and dropout for improved regularization, expands the network from 23 to over 35 layers to capture hierarchical features for distinguishing irregular tumor boundaries from dense tissue, utilizes combined loss functions such as BCE and DICE to better capture complex features in dense breast tissue and improve segmentation accuracy. Experimental results show that the proposed CAD system achieved 95.5% accuracy in detecting and localizing abnormalities using the CBIS-DDSM dataset 94.2% sensitivity, 96.8% specificity.

Zitieren

APA

Hiwa, H., & Abdulla, A. A. (2025). Early Breast Cancer Detection for Mammogram Images using U-Net Deep Learning Model. Zanin Journal of Science and Engineering. https://doi.org/10.64362/zjse.182

Vancouver

Hiwa H, Abdulla AA. Early Breast Cancer Detection for Mammogram Images using U-Net Deep Learning Model. Zanin Journal of Science and Engineering. 2025. doi:10.64362/zjse.182

BibTeX
@article{Hiwa2025,
  author = {Hiwa, Hardi and Abdulla, Alan Anwer},
  title = {Early Breast Cancer Detection for Mammogram Images using U-Net Deep Learning Model},
  journal = {Zanin Journal of Science and Engineering},
  year = {2025},
  doi = {10.64362/zjse.182}
}
Kostenlos starten →