A non-invasive detection method for diagnosing liver cancer based on electronic nose is introduced.The human body’s exhaled gases contain hundreds of ingredients. If some organs of the human body are diseased or abnormal

the exhaled gas may change. Exhaled gases from 34 patients with hepatocellular carcinoma and from 52 controls by electronic nose are collected. Features of signal are extracted from the collected data and a prediction model is constructed using linear discriminant analysis（ LDA） and support vector machine（ SVM）.Random and cross validation methods are used to judge classification precision. The results show that the classification precision of SVM can exceed 85 %

which can effectively distinguish the exhaled gases of patients with hepatocellular carcinoma from those of the control group. The research result shows that the electronic nose system is helpful for noninvasive detection of liver cancer. And it is more convenient

safe

and the cost is greatly reduced.