Diagnosis of Liver Diseases from P31 MRS Data Based on Feature Selection Using Genetic Algorithm
P31 MRS technique is important either in diagnosis or in treatment of many hepatic diseases for it can provides non-invasive information about the chemical content of the energy metabolism in cellular level. The data samples from P31 MRS are classified into three types of hepatocellular carcinoma, hepatic cirrhosis and normal hepatic tissue using computational intelligence methods. A genetic algorithm is used as main feature selection method and the Gaussian model is selected in the mutation operation. Two classification algorithms are used which consist of fisher linear discriminant analysis and quadratic discriminant analysis. Experiments show that the application of genetic algorithm and fisher linear classifier offers more reliable information for diagnostic prediction of liver cancer in vivo. And when the cross-validation method is 10-fold model, this algorithm can improve the average recognition correction rate of three types to 94.28％.
Genetic algorighm fisher linear classifier hepatic cirrhosis hepatocellular carcinoma P31 MRS
Jinyong Cheng Yihui Liu Jun Sang Qiang Liu Shaoqing Wang
School of Computer Science and Information Technology,Shandong Institute of Light Industry,Jinan, Sh Department of Magnetic Resonance Imaging,Shandong Medical Imaging Research Institute Jinan, Shandong
国际会议
无锡
英文
122-130
2010-09-17（万方平台首次上网日期，不代表论文的发表时间）