Breast cancer is a health concern that affects women greatly in the global world, and the only solution to improving the survival rates is early detection of this cancer. Mammograms are a procedure that is common among the physicians during screening. These images may be difficult to interpret because of the presence of blurred or obscured information with the effect of thick breast tissue. This research examines the application of an image processing tool known as an octagon structuring element to assist a computer-aided diagnosis (CADx) system to make accurate decisions whether a tumor is benign or malignant. The data used in the research were taken from the Mini Mammography Image Analysis Society (MIAS) database as the primary source of data. The mammogram images were pre-processed by using median filters and histogram stretching to remove noise and enhance contrast, respectively. This was followed by the segmentation of the images using a modified Otsu thresholding technique to detect the important sections of the breast. To capture local features in the images, BRISK keypoints were then employed to obtain features and identify them in the images. Data were then reduced in terms of dimension with Principal Component Analysis (PCA), before the classification phase. Support Vector Machine (SVM) was utilized in refining the data to be used in the final classification. The results of the research found that the most effective octagon size was at a radius of 48. Among the 9 radii with high accuracy performance, radius 48 was chosen as the optimal level because it offers the best balance between short training time and fast prediction speed. With this radius, the system managed to achieve an accuracy of 97.60% with PCA. Then, 99.96% sensitivity, 94.39% specificity, a Dice Coefficient of 0.9789, and a Jaccard Index of 0.9586. This proves that the use of the octagon shape is very helpful in improving the quality of medical images, as well as facilitating medical experts to make more accurate and faster diagnoses.
| Item Type: | Article |
|---|---|
| Creators: | Creators Email / ID Num. Yasiran, Siti Salmah salmahyasiran@uitm.edu.my Abd Halim, Suhaila suhaila889@uitm.edu.my Hazwani Zaidi, Nur Ainna 2023861884@student.uitm.edu.my |
| Subjects: | T Technology > TA Engineering. Civil engineering > Applied optics. Photonics > Optical data processing T Technology > TA Engineering. Civil engineering > Applied optics. Photonics > Optical data processing > Image processing |
| Divisions: | Universiti Teknologi MARA, Shah Alam > Faculty of Computer and Mathematical Sciences |
| Journal or Publication Title: | Mathematics Letters |
| UiTM Journal Collections: | Other UiTM Journals > Mathematics Letters |
| ISSN: | eISSN: 2948-3735 |
| Volume: | 5 |
| Number: | 2 |
| Page Range: | pp. 29-39 |
| Keywords: | Breast tissue classification, Mathematical morphology support vector machine, Structuring element |
| Date: | 31 October 2026 |
| URI: | https://ir.uitm.edu.my/id/eprint/149385 |
149385.pdf
