|
|
Bimonthly Since 1986 |
ISSN 1004-9037
|
|
|
|
|
Publication Details |
Edited by: Editorial Board of Journal of Data Acquisition and Processing
P.O. Box 2704, Beijing 100190, P.R. China
Sponsored by: Institute of Computing Technology, CAS & China Computer Federation
Undertaken by: Institute of Computing Technology, CAS
Published by: SCIENCE PRESS, BEIJING, CHINA
Distributed by:
China: All Local Post Offices
|
|
|
|
|
|
|
|
|
|
Abstract
Digital mammography is the method that has proven to be the most trustworthy and productive in terms of early and precise detection of breast cancer. Within the realm of medicine, the diagnosis and categorization of breast cancer are both helped along significantly by image processing. In this article, a system is developed to classify mammography pictures into three distinct categories: benign, malignant, and normal. These categories are discussed more below. Mammogram images go through a preliminary processing step, during which the segmented region's features are extracted. These features are input into a modified version of the SVM and a KNN classifier to be trained. In order to assist in the classification of the mammography pictures, the authors propose a hybrid approach that combines the modified SVM and KNN classifiers. The most recent method, which is an improvement on the SVM algorithm, is one that introduces multi class for the classification of breast cancer. Utilising the KNN method in accordance with the distribution of test images inside a feature space, it does this. In addition to that, the SVM and KNN classifiers' degrees of accuracy are evaluated here. The accuracy of the prognosis produced by the modified SVM and KNN hybrid algorithm is superior to that produced by either the KNN approach or the SVM methodology. Using 10 test photos and 20 taught ones, this approach was evaluated. When it comes to the classification of mammography pictures, our system obtains an overall mean accuracy of 99.3406%. Classification, KNN, MIAS, and Proposed KNN with SVM are some of the keywords here.
Keyword
#
PDF Download (click here)
|
|
|
|
|