Extreme learning machine-based investigation on automated detection of architectural distortion in mammograms
by Elangeeran Malar; P. Deepan Chakravarthi
International Journal of Operational Research (IJOR), Vol. 41, No. 4, 2021

Abstract: Breast cancer, having its origin from the breast tissue is usually detected by mammographic screening. The early detection of breast cancer reduces the mortality rate. A subtle type of breast cancer that often leads to misinterpretation by radiologists is architectural distortion. Though the existing computer aided diagnosis systems efficiently and effectively detect the presence of micro-calcification and masses, the diagnosis of architectural distortion lacks a promising method. This project attempts to detect and classify the regions of mammograms having architectural distortion. MIAS and DDSM database images are enrolled in this research work. 350 region of interests (ROIs) of each architectural distortion and normal cases were extracted. They were subjected to a filtering process, followed by contrast enhancement. Application of Gabor filter to the images resulted in orientation differences between the normal and abnormal images. Statistical features extracted from the resulting images were classified using extreme learning machine classifier. The experimental results obtained from extreme learning machine in comparison with support vector machine had an accuracy of 98.49% and 87.21% for MIAS and DDSM respectively. The accuracy of combined database of which is 85.38%.

Online publication date: Mon, 16-Aug-2021

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
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 Operational Research (IJOR):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your 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