Study On Liver Ultrasound Image Lesion Segmentation

The morbidity rate of liver disease is high in our country, and the early diagnosis of liver disease has attracted many researchers in the past ten years.Ultrasound images can clearly show the sectional image between organs and surrounding tissue, close to the real structure, so the ultrasound image can be used for early diagnosis. With the development of computer-aided diagnosis technology, the automatic analysis for liver ultrasound images is possible. However, due to the complexity and uniqueness of liver ultrasound images, as well as the inevitable speckle noise and texture, the automatic analysis for liver ultrasound image having a certain degree of difficulty. By studying segmentation technologies of liver lesion in ultrasound image, achieve effective segmentation of the lesion area, then extract its contour accurately, and establish the foundation for subsequent diagnostic analysis. The main work is as follows:(1) This paper presents a segmentation of liver lesion for ultrasound image through combining Wellner's thresholding algorithm with particle swarm optimization (PSO). The proposed method obtains an optimal parameter, which expressed as a percentage or fixed amount of dark objects against a white background in a gray image, of Wellner's thresholding algorithm by PSO method. In the experiments, we analyzed the effect of particle swarm optimization method parameters, and then assess TPVF, FNVF and FPVF index for compared algorithms. Experimental results show that the improved algorithm Wellner-PSO is reliable on segmenting the contour of liver lesion.(2) This paper presents a segmentation method of liver lesion in the ultrasonic image through combining graph-based method with particle swarm optimization (PSO). The proposed method utilized the PSO and Otsu method to obtain an optimal parameter, which is a fixed value in graph-based segmentation method. Then four indicators, which included Hausdorff distance (HD), mean absolute distance (MD), Dice's relative coefficient, were estimated to verify the proposed method. Experimental results have shown that the proposed technique can successfully and accurately extract the contour of liver lesion.