Breast Cancer Image Analysis on Distribution Features of Texture in Reduced Dimension Space

In this paper an Improved Kernel Linear Discriminant Analysis algorithm is proposed to study the distribution difference between breast cancer images and breast fibroids images in the reduced dimensional space. By extracting lacunarity as feature we observe that the cancer images are clustered far away from the fibroids images in the corresponding reduced dimension space. A scheme is developed to discriminate benign images from the malignant images. The experimental results confirm the validity of the proposed approach and its potential to support a computer-aided diagnosis system.