Detection and Classification of Breast Cancer using Variants of Discriminant Analysis

Abstract

Benign and Malignant are of two common classes of breast cancer that are found worldwide.
Appropriate detection and classification is a major necessity for patients as well as the physicians for
diagnosis. So that the necessary care can be taken to save the life.
The proposed approach is to detect and classify breast cancer using ten different derived features.
These features are utilized with two different Discriminant Analysis classifiers. These classifiers are
considered as linear and quadratic type. The obtained result is validated using different evaluation
metrics like accuracy, sensitivity, specificity. The result shows the effectiveness of the proposed
method in terms of detecting the cancer cell accurately and identifying the type of cancer as
malignant or benign. Overall the nonlinearity of Quadratic Discriminant Analysis is found more
effective for detection and classification in comparison with Linear Discriminant Analysis. It is found
clear detection of Benign and Malignant type of cysts.