A Study Of Mammograms Recognition Based On A Neural Network For Multiple Features

Breast cancer is one of the most prevalent tumor diseases among women. There is an increasing trend of its incidence in recent years. Early detection and diagnosis is the key to reduce breast cancer mortality. Mammography is the most effective method for the early detection at present. With the rapid progress of imaging and computer technology, Computer-aided diagnosis (CAD) techniques provide potential necessaries for the early detection and diagnosis of breast cancers. It can help to reduce the burden in reading of physicians who will concentrate their limited energy mainly in suspicious lesions areas. There is import clinical significance to improve the accuracy of breast cancer diagnosis.Based on analyzing the characteristics of mammograms, an algorithm based on a neural network for multiple features is proposed to segment and recognize mammograms. Firstly, it preprocesses original images, which help to clear noise. After that, an enhancement algorithm is use to enhance images. And then, a model combined multiple features extraction and a neural network is set up. It is used to extract the feature information of R, G, B, gray scale, entropy, mean and standard deviation in receptive fields of the input neurons in the network. A characteristic vector which combined the features organically is used to train the neural network. After training, the network is used to segment a breast cancer X-ray image into regions of normal and regions of interest (ROI). Finally, an edge detection based on spiking neural network is proposed to detect the edge of ROI. After that, a method of clustering analysis based on the Euclidean distance is proposed to locate masses and calcifications in ROI area.The experimental results show that the algorithm proposed in this paper is able to extract the ROI area in an image efficiently. Also, it is able to locate masses and calcifications accurately. The proposed approach with good detection performance and practical can be applied in automatic diagnosis systems of breast cancers.