The Research Application Of Differential Equations In Breast Cancer DCE-MRI Analysis

Breast DCE-MRI examination technology has the advantages of no radiation damage, good soft-tissue imaging, high resolution and high contrast, playing an important role in breast cancer pre-inspection and post-efficacy analysis. But due to the particularity of breast location and structure, the limbs moving, parts trembling, heartbeat, respiration and other factors have contributed to the images to appear blurred, overlap region, speckle noise, motion artifacts and other phenomenon in the image process, then the preprocessing and analysis for the medical images is necessary. Differential equation image processing technology has mature mathematical theory foundation and perfect mathematical model, thus using differential equations to analyze and deal with breast DCE-MRI images has become one of the important ways to assist doctors to forecast lesions installments and cancer diagnosis. This paper mainly studies the applications of differential equation on breast DCE-MRI images’analysis and processing, specific work as follows.(1) Breast cancer DCE-MRI image restoration. This paper analyzes several major image restoration model based on partial differential equations:total variation (TV) model, BSCB model, curvature driven (CDD) model, and get the numerical solution through spatial dispersion. Based on the CDD model, this paper combines Perona-Malik functional model and the idea of repairing join proliferation, coming out the PMP-CDD repairing method. The method repairing and proliferation carry out alternately and repair information spread along the image isophote line direction, maintaining the edge of damaged area while smoothing noise. The experiments show while repairing breast cancer DCE-MRI images of the same damage rate, the results of PMP-CDD model has a higher PSNR(Peak Signal to Noise Ratio) value and better repairing effect, and meet the criteria of image edge connectivity.(2) Breast cancer DCE-MRI image segmentation. Study the geodesic active contour model and CV model of partial differential equation image segmentation method. Against the disadvantage of traditional C-V model only using the region information to conduct image segmentation, this paper proposes a mixed C-V image segmentation model witch based on gradient and curvature energy and punish energy. Gradient and curvature energy term combines the scale transformation stop function based on gradient information and the thought of mean curvature flow equation, making the topology of evolution curve changes while maintaining a smooth. Punish energy term is an adjustment formula of level set function, and it makes the evolution curve to develop without having to re-initialize the level set function to a signed distance function (SDF), accelerating the image segmentation speed. Experiments show that mixed C-V model having a better segmentation result for breast cancer DCE-MRI images and the segmentation time is shorter, having a certain practicality.(3) Initial building a system for breast DCE-MRI image restoring and segmentation. Combine PMP-CDD repairing algorithm and mixed C-V image segmentation model to achieve a simple operator interface system with Matlab’s Guide controls. Breast cancer tumor region segmentation based on restoration, reducing the affect of noise and artifact on image segmentation, improving the accuracy of the tumor region segmentation.The breast cancer DCE-MRI data of this article’s experiments are all provided by Fujian Provincial Tumor Hospital.