腹部MR画像における肝硬変の自動識別法の開発

概要

論文の詳細を見る
Cirrhosis of liver is a late stage of progressive liver disease defined as structural distortion of entire liver by fibrosis and parenchymal nodules. As the liver regenerates, fibrous connective tissue forms that may cause gross and microscopic distortion of normal hepatic morphology. In MR images, shape and texture analysis is regarded as an important and useful tool to differentiate cirrhosis from normal liver. In this paper, we propose a method to calculate the shape features from the segmented liver regions on MR image. Meanwhile, the texture features are quantified by using gray-level difference method (GLDM) within the small ROIs (regions of interest) selected in the liver region. The degree of liver cirrhosis is derived from integrating the shape and texture features of liver into a three-layer feed-forward artificial neural network (ANN). A liver is finally regarded as cirrhosis if the percentage of the ROIs with the degree over 0.5 is greater than 50%. The initial result showed that the ANN based method classified liver cirrhosis with a training accuracy of 100% on the 100 ROIs in the training set and that 82% liver cirrhosis and 100% normal cases were correctly differentiated from 18 test cases, which demonstrates the effectiveness of our proposed method.
医用画像情報学会の論文
2004-05-01