Automated methodology for breast segmentation and mammographic density classification using co-occurrence and statistical and SURF descriptors Online publication date: Mon, 18-Jul-2022
by Roberto Pavusa Junior; João C.L. Fernandes; Alessandro P. Da Silva; Marcia A.S. Bissaco; Silvia R.M.S. Boschi; Terigi A. Scardovelli; Silvia C. Martini
International Journal of Biomedical Engineering and Technology (IJBET), Vol. 39, No. 3, 2022
Abstract: This paper presents a fully automated methodology for segmenting and classifying mammographic images. We developed a new set of descriptors for determination of breast density based in the standard MIAS database. We performed the image pre-processing, image laterality detection and background and artefacts removal as well as pectoral muscle identification and segmentation. From the segments named breast and pectoral muscle, descriptors from histogram, co-occurrence matrix, and points of interest analysis, were extracted. The descriptors were Spearman correlation analysis, principal component analysis and linear discriminant analysis. The image classification was performed through the classifiers k nearest neighbours (kNNs) and support vector machine (SVM). An accuracy of 72.05% was achieved with the SVM classifier, and of 91.30% with the kNN. The proposed methodology is promising, as well as the descriptors used for the breast density classification, which surpassed most of the studies found in the literature that used all images from the MIAS database.
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Biomedical Engineering and Technology (IJBET):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.
If you still need assistance, please email subs@inderscience.com