Breast lesions detection using FADHECAL and Multilevel Otsu Thresholding Segmentation in digital mammograms

• 主要な著者: Kamarul Amin, Abdullah@Abu Bakar, Saifullah Harith, Suradi, Nor Ashidi, Mat Isa
• フォーマット: Conference or Workshop Item
• 言語: English; English; English
• 出版事項: 2021
• 主題: R Medicine (General); RC0254 Neoplasms. Tumors. Oncology (including Cancer)
• オンライン･アクセス: http://eprints.unisza.edu.my/4591/1/FH03-FSK-21-55109.pdf; http://eprints.unisza.edu.my/4591/2/FH03-FSK-21-52802.png; http://eprints.unisza.edu.my/4591/3/FH03-FSK-21-55109.pdf; http://eprints.unisza.edu.my/4591/

Breast cancer is the most common cause of mortality among women. Early detection plays an important role to improve survival rates. Digital mammograms can be used to detect breast lesions within the breast tissue. However, digital mammograms have a limitation of low contrast images due to the low exposure factors used. As a result, the extraction of breast lesions using the region of interest (ROI) tool will be difficult and, thus, lead to misclassification. This paper presents a novel technique to detect breast lesions in digital mammograms, known as Fuzzy Anisotropic Diffusion Histogram Equalization Contrast Adaptive Limited (FADHECAL) incorporated with Multilevel Otsu Thresholding Segmentation. FADHECAL will enhance the breast lesions by reducing the image noise while preserving the details. Multilevel Otsu Thresholding Segmentation detects the breast lesions using the ROI tool at different intensity levels. The performance of FADHECAL incorporated with Multilevel Otsu Thresholding Segmentation has been tested on 115 digital mammograms from the Mammographic Image Analysis Society (MIAS) database with the abnormal conditions. The efficiency of the proposed technique is 94.8%, and the error rate is 5.2%. In conclusion, FADHECAL incorporated with the Multilevel Otsu Thresholding Segmentation has provided sufficient detection of breast lesions with the appropriate quality of the digital mammograms.