Mammogram Segmentation Based On Improved Template Matching

Breast cancer is a common cause of death.In recent years,the incidence of breast cancer has greatly increased which badly damages to women’s health.Clinical trials show that early detection,diagnosis and treatment are beneficial to the life and health of patients with breast cancer.Mammography is the most reliable and effective tool for early detection,diagnosis and treatment of breast cancer because of high resolution,good repeatability,high diagnostic accuracy.When screening breast pathology images,some factors should be taken into account,which are easy to cause the miscarriage of justice,such as the complex structure of the breast,the relatively close density of each part on the tissue and the control of imaging equipment.This problem promotes the rapid development of computer aided diagnosis technology based on breast imaging,and accelerates the research of breast imaging preprocessing and lesion segmentation algorithm.This paper proposed the nonlinear local transformation based mammographic image enhancement and template matching algorithm for breast image segmentation.The main research work are as follows:An enhancement algorithm based on local nonlinear transform is proposed.The nonlinear transformation is used to suppress the information of the gland and tissue in the non-mass region.And the local standard deviation is used to enhance the contrast of the image.Experimental results demonstrate that the proposed algorithm can improve the contrast effectively and enhance lesion information.Good enhancement effect is convenient for subsequent work.An improved template matching algorithm is proposed.It consists of four steps,constructing the new template,measuring the similarity with the mutual information,combining the results of the template measurement with different scales,and obtaining the suspicious area according to the morphological features.The novel template is composed of two-dimensional hyperbolic secant function and the maximal intensity region of the enhanced image,which makes full use of the intensity information of mammogram.Thus,it is more beneficial to imitate the ground truth of mass.The morphological features are applied in threshold segmentation to obtain suspicious regions.Finally,the gray features of region and edges are used to reduce the false positive region.Compared with other algorithms,the proposed algorithm has high sensitivity and specificity,and the number of false positive regions is well controlled.