Molybdenum-Ray Image Recognition of Breast Cancer Based on Fractal Texture Analysis

摘要

A kind of computer-aided diagnosis method for mammary gland molybdenum palladium X-ray image is put forward. Using fuzzy histogram equalization algorithm to improve the original image contrast in order to extract the interested region and strengthen regional edge. In order to accurately extract lesions the improved fuzzy clustering algorithm is put forward. Fractal box counting dimension is used to describe the texture characteristics of lesions which have been segmented. According to the shape and gray distribution of 10 forms of tumour and calcification spot, 6 characteristic parameters are extracted. A BP neural network of three layers is established. After training on 244 samples, 238 samples are used to validate the accuracy of the system, the result shows average accuracy rate is 96.59%.