Medical Ultrasound Image Segmentation Study Of New Methods

Medical ultrasound imaging, computerized tomography, magnetic resonance imaging and nuclear imaging are the four important medical imaging techniques in modern society. Among them, medical ultrasound imaging is widely used in medical diagnosis and treatment due to its unique characteristic of real-time, noninvasiveness, cheapness, easy repetition, and high sensitivity. It has high potential in quantitive analysis, real-time monitoring and treatment scheduling, etc. When ultrasound images are utilized to do medical analysis and diagnosis, it's usual to extract the specific organs or areas to do further analysis better, so image segmentation is essential. Although there are lots of image segmentation methods up to now, ultrasound image segmentation remains a worldwide difficult problem because of its intrinsic speckle noises and the tissue-related textures. Based on the above facts, this paper does some researches on ultrasound image segmentation to improve current algorithms or propose new methods. It's also a meaningful exploration for the segmentation of low SNR images.The paper first gives an overview of the ultrasound image segmentation methods. Then several classic segmentation algorithms and some algorithms that have been used in medical images are analyzed and the experimental results based on them are given to show that they are not applicable for ultrasound images according to the clinic requirements. The paper also discusses their problems while used on ultrasound images briefly. Then, a new biologically simulated neural network, i.e., pulse coupled neural network is analyzed in depth. For the intrinsic problems of PCNN, the paper proposes a new corresponding algorithm and some good experimental results are acquired. Furthermore, the paper brings pulse coupled neural network into ultrasound image enhancement for the first time and rather good experimental results are got. The reason that the proposed enhancement algorithm can also be viewed as a multivalue segmentation method is given in the paper. The above experiments also show that PCNN is a tool of high potential in ultrasound image processing and analysis.