Application Of PCNN In The Ultrasound Image Segmentation Of Cholelithiasis

Gallstone is a kind of common gallbladder illness. Gallstone illness ranks only after appendicitis in the acute abdomen disease, which is mainly in the northwest of China. Ultrasound imaging is often used for a preliminary diagnosis of gallstone disease. Ultrasound image has the characteristics of real time and lossless performance, which is suitable for soft tissue diagnosis. However, ultrasound images are various, low contrast and have too much noise, thus the qualities of them are lower than computed tomography (CT), magnetic resonance imaging (MRI) and other medical images. In addition, the clinical diagnosis of ultrasound images is mainly affected by the doctor’s subjective opinions, which can lead to the missed diagnosis and misdiagnosis. Therefore, it is necessary to do some work in the image segmentation of gallstones. Gallstones region segmentation and especially gallstones edge segmentation is very helpfull to improve the accuracy of the clinical diagnosis.Pulse Coupled Neural Network (PCNN), which has the biological background, is a new generation artificial Neural Network. It derived from the research of the nerve cells of the cat’s visual cortex, and is the precise imitation of visual system of mammals. So it is closer to the image processing process by visual system. Pulse coupled neural network is a single layer neural network model which has the characteristic of pulse burst. It is very suitable for image segmentation application in biomedical image processing.This paper studies PCNN based image processing methods, and segments ultrasonic gallstone image according to the characteristics of ultrasound images of gallstones. The major work is as follows:1. The paper lists the typical kinds of pathological changes of ultrasound gallstone images and analyzs the characteristics of gallstones.2. PCNN model is analyzed and its parameters are experimentally set.3. This paper proposes a PCNN based ultrasound gallstone image segmentation method. The ultrasound gallstone image segmentation results are discussed by comparing to the manual segmentation results.The simulation results show that the gallstones lesions of ultrasound images can be accurately segmented by the proposed PCNN based segmentation method.