Ultrasonic Thyroid Nodule Image Segmentation Algorithms

In recent years,the incidence of thyroid nodules has showed a year-on-year growth trend,which has become one of the important diseases that endanger human health.Ultrasound medical image is widely used in clinical diagnosis because of its low cost and no radiation.The use of image processing technology can accurately segment the nodular region and provide important auxiliary information for the doctor's diagnosis to guide the clinical treatment,which is of great value.The resolution and contrast of ultrasonic thyroid images is low and the spot noise is serious.The size,location and shape of nodules of the different groups are of great difference,which greatly affect the accuracy of nodular segmentation.In this paper,two segmentation algorithms of ultrasound thyroid images are proposed based on active contour level set method.Firstly,the level set segmentation model is proposed based on the global and local combination.The global directivity of the area energy terms are added to the traditional local LBF model with to solve the shortcomings that LBF model is sensitive to the initial contour sensitive and easy to fall into the local minimum.Using the adjustment of the LBF model to adjust the length and smoothness of the level set in this paper,the energy level set equation is obtained by minimizing the energy functional.By comparing with the traditional CV model and LBF model,the experimental results show that the proposed algorithm is more accurate.The new segmentation algorithm which combines fuzzy kernel clustering and improved distance regularized level set model is also proposed in this paper.The algorithm solves the problem that the distance regularized level set is sensitive to initial contour,the evolution parameters need to be manually set,and the poor ability for weak edges segmentation.Firstly,the fuzzy clustering algorithm is used to segment the nodule region coarsely which is regarded as the initial evolution contour after binaryzation.Secondly,the evolution parameters of the level set are calculated automatically using the region surrounded by the initial evolution outline.Finally the course of evolution by Gaussian regularization rule is used to segment the region of thyroid nodules.The experimental results are compared with the traditional CV model,the LBF model and the DRLSE model.The experimental results show that the proposed algorithm can complete the segmentation more accurately.