一种基于高频超声图像的肝包膜提取算法研究

A Liver Capsule Extraction Algorithm Based on High Frequency Ultrasound Images

针对高频超声图像中严重断裂肝包膜提取率低的问题,提出一种腹水结合动态规划的肝包膜边界提取算法。文中利用腹水特征实现肝包膜的粗定位,结合动态规划算法实现肝包膜的准确提取。实验结果显示,正常组、轻度、中度和重度肝硬化组的提取准确率分别为100.00%、98.75%、93.75%和92.30%。实验结果表明,文中算法能够在自动提取肝包膜轮廓线的同时提高断裂肝包膜线的提取准确率,并为进一步应用计算机辅助诊断提供了符合医学超声视觉特征的腹水特征。

Aiming at the low extraction rate of severely ruptured hepatic capsule in high frequency ultrasound images, an accurate extraction algorithm of liver capsule boundary was proposed. In this paper, liver ascites was used to locate the liver capsule roughly, and dynamic programming algorithm was used to extract the liver capsule accurately. The results showed that the extraction accuracy of normal group, mild, moderate and severe cirrhosis group was 100.00%, 98.75%, 93.75% and 92.30%, respectively. The experimental results displayed that the proposed algorithm could automatically extract the contour of the liver capsule and improve the accuracy of extracting the ruptured liver capsule line. The obtained data also provided the ascites features that were in line with the medical ultrasound visual characteristics for further computer-aided diagnosis.