USG Image Segmentation for Breast Cancer Detection Using the Active Contour Method

Abstrak

Breast cancer is one of the leading causes of cancer-related mortality among women. In Indonesia, it accounts for 16% of all cancer cases and 22.9% of invasive cancers in women. Ultrasonography (USG) is commonly used for breast imaging; however, its effectiveness is often limited by speckle noise, low contrast, and intensity inhomogeneity, which hinder accurate tumor segmentation. Therefore, this study aims to propose an effective segmentation framework for breast ultrasound images by integrating image enhancement and segmentation techniques. The proposed method combines Contrast Limited Adaptive Histogram Equalization (CLAHE) and contrast stretching for image enhancement, followed by Active Contour segmentation using the Chan–Vese model to accurately delineate tumor boundaries. CLAHE improves local contrast while controlling noise amplification, and contrast stretching enhances global intensity distribution to increase lesion–background separability. Experiments were conducted on 20 breast ultrasound images consisting of benign and malignant cases. The segmentation performance was evaluated using Receiver Operating Characteristic (ROC) analysis. The results show that for benign cases, the method achieved an accuracy of 92.35%, sensitivity of 80.5%, and specificity of 93.41%, whereas for malignant cases, it achieved an accuracy of 73.9%, sensitivity of 57.83%, and specificity of 73.49%. These findings indicate that the proposed approach is effective in improving segmentation performance and boundary detection in breast ultrasound images, although performance on malignant cases remains relatively lower.

Identifier

• DOI: https://doi.org/10.48084/etasr.18899
• OpenAlex: https://openalex.org/W7202272965