Design And Implementation Of Texture Feature Extraction In B-mode Ultrasound Scanner

B-mode ultrasound examination is the first choice for body inspection, with no side effects, convenience and low cost, and it has been widely used in the hospitals. Texture feature of B-mode ultrasound image is the most important foundation for diagnosis. The function of texture feature extraction of B-mode ultrasound images was appended in B-mode ultrasound scanner. It provides doctors with quantitative aspects foundation for increasing the rate of accuracy and objectivity of diagnosis compared with estimate by eyes.Firstly,the structure of hardware was analyzed, the system which was appended with function of texture quantification mainly consists of a FPGA(Field Programmable Gate Array) and a SCM(Single Chip Machine). The peak value of SCM's speed is 50MIPS. SCM has 8KB memory cells. Because of the other part of system software used many memory cells, in fact about 2KB memory cells can be used for the module of texture quantification. Then, texture feature extraction were designed and implemented possible arrangement of memory and design of efficient calculate methods, by thinking of the SCM'feature.For texture feature extraction, The B-mode ultrasound scanner can extract feature from image like first order statistics, such as mean, standard deviation, max, min, skewness and kurtosis, and feature of bright specks, such as speck size and speck count. They can measure the distribution of B-mode ultrasound image. For extracting specks'feature, firstly, use a LoG (Laplacian of Gaussian) filter to make a convolution with image and segment the region of bright specks, secondly, calculate speck size using the method of region growing. The test result shows that the B-mode ultrasound scanner with function of texture feature extraction has a friendly interface and can provide doctor with kinds of texture feature from image.