Research On Medical Image Segmentation Model Based On Bias Field And Atlas Information

In recent years,with the rapid development of modern medical image technology,the medical image processing technology is becoming more important.In particular,the accurate segmentation of medical images is significant for doctors to diagnose and analyze the etiology.However,the false contours appearing in medical images due to fuzzy image boundary,intensity inhomogeneity and random noise,may lead to the inaccurate segmentation results.Therefore,in this paper,we propose three medical image segmentation models to solve the problem of medical image segmentation.Due to the magnetic resonance images have severe intensity inhomogeneity,which poses a great challenge to accurately segment magnetic resonance images.And magnetic resonance images usually contain multi-tissues,so it is particularly important to establish an image segmentation model that can segment multi-targets simultaneously.Therefore,we present an improved active contour model by combining the bias field,membership function and split Bregman method.A slowly varying bias field is the key to correct inhomogeneous images.By estimating the bias fields,not only can we accurately segment magnetic resonance image,but also a homogeneous image after correction is provided.The membership function enables our model segment the magnetic resonance image into multi-regions simultaneously.And the split Bregman method accelerates the minimization process of our model by reducing the computation time and iterative times.Then we apply our model to segment a large amount of magnetic resonance images.Experimental results shows that our model can obtain satisfactory segmentation and correction results for those images,and is superior to other models.Medical images not only have intensity inhomogeneity,but also may have severe noise.So it is a major problem in medical image segmentation due to the existence of noise causes false contour in medical images.Therefore,we propose an improved active contour model based on global image information and bias field correction,which can accurately segment images disturbed by intensity inhomogeneities and serious noise.We give the two-phase energy functional and multi-phase energy functional of our model,respectively,and minization the energy functional by the iteration of the split Bregman method.By computer simulation,we apply our model to segment and correct images,ultrasound images,X ray images and synthetic images.Experimental results and comparisons with other models have shown that our model has the advantages of higher accuracy,higher efficiency and robustness in dealing with the intensity inhomogeneity and serious noise in image segmentation.And the intensity distribution of the bias field corrected images is more homogeneity.It is difficult to segment medical images because of the overlap of multi-tissues and the complexity of background.Nowadays,atlases or labels,which can be seen as prior segmentation results,have been widely applied in image segmentation models for providing prior knowledge in segmentation process.In this paper we present a new model that combining the region-scalable fitting energy model idea and the advanced atlas concept.The proposed energy functional consists of a smooth length term,a target image data term and a transcendental constraint term.And the split Bregman method is then applied to minimize the energy functional.We apply our model to segment a large amount medical images and natural images.Experimental results shows that the proposed model is robust to initial conditions and can accurately segment images with serious intensity inhomogeneity or complicated boundaries.And the introduced transcendental constraint term plays a key role in the improvement and good performance of the proposed model.