CT图像自动识别系统诊断肝炎肝硬化的研究

目的探讨肝炎肝硬化早期诊断方法。方法基于肝脏CT平扫图像，采用计算机图像自动识别及测量技术．计算乙型肝炎患者肝叶比例，通过统计对照数据库中的历史数据，对早期肝硬化做出诊断。结果肝硬化组与乙型肝炎组比较，本系统L1／L2和R1／L1比值均有显著性差异（P〈0．05）；建立判别方程：D=1．650LI／L 2＋0．509R1／L1-5．019；确定判别的临界值为0．0795；D〉0．0795为肝硬化，D〈0．0795为乙型肝炎；判别效果分析（回代），结果诊断肝硬化的正确率为68．42％，诊断乙型肝炎或判断为无肝硬化的正确率为89．8％。结论可以利用肝脏各叶的比例变化来预测和早期诊断肝炎肝硬化。

The early diagnosis of liver cirrhosis based on CT imaging by auto-recognition technique

Objective An auto-recognition system for measurement of liver proportion in computerized tomography imaging was developed for early diagnosis of hepatitis B-related liver cirrhosis.Methods The right and left liver proportion in CT imaging in patient with hepatitis B was measured and an formula was established in the database.Results The difference of L1/L2 and R1/L1 ratio between patients with liver cirrhosis and with chronic hepatitis B was significant(p< 0.05);we got a formula of D=1.650 L1/L2 0.509 R1/L1-5.019 and the cut-off value was 0.0795;when D>0.0795,the liver cirrhosis was diagnosed,while D<0.0795,the chronic hepatitis B was considered.The accuracy of cirrhosis diagnosis was 68.42%,and of the exclusion was 89.8%.Conclusion We may use the auto-recognition system in CT-imaging to predict or diagnose the early liver cirrhosis.