Application Of Tumor Markers Protein Biochip Combined With Artificial Neural Network In Aided Diagnosis Of Lung Cancer

ObjectiveLung cancer is one of the malignant tumors, which threats to people's health. Its incidence and mortality rates are high.1.2 million new cases are diagnosed as lung cancer every year. The most common symptoms of lung cancer are cough and dyspnea, so it's often confused with benign pulmonary diseases. There is no specific symptom at early stage of lung cancer, when lung cancer is diagnosed; it is at middle or end stage, so the patients lose the best opportunity of surgical therapy, thus there are high costs in the treatment and poor prognosis to the patients. Therefore early diagnosis and active treatment of lung cancer are the key points in improving survival rate. Detection of tumor markers in serum has the merits of high efficiency, convenient, small trauma and available. How to screen, identify and test tumor markers has been a hotspot in auxiliary diagnosis of lung cancer. Until now, non specific biomarker has been found, so a single biomarker is not enough. Multiple biomarkers testing can raise the rate of positive diagnosis notably and distinguish the pathological classification of lung cancer better. However, multiple biomarkers testing can cause multi-parameter which cannot be efficiently analyzed by general statistics analysis methods. Artificial neural network (ANN) is a kind of fast developing information processing systems especially for medical science of pattern recognition and classification.This study is to establish a lung cancer ANN model that makes use of ANN technology platform combined with tumor markers for diagnosis of lung cancer as a clinical assistant method.Materials and methods1. The clinical data and multiple tumor marker protein biochip detective system records of 102 lung patients (50 cases with lung cancer and 52 cases with benign pulmonary diseases) were retrospectively reviewed from May 2010 to October 2010. All patients were confirmed histopathologically and there was no significant difference between the two groups.2. All tumor markers were tested by multiple tumor marker protein biochip detective system, of those nine of complete data. There were CA199,NSE,CEA,CA242,SF,AFP,CA125,HGH and CA153. The nine indexes were used as the basic data for the study. The positive criterion were CEA>5μg/L, CA19-9>35U/ml, NSE>13μg/L, CA242>U/ml, CA153>35U/ml, CA125>35U/ml, AFP>2μg/L, SF>322μg/L(male), SF>219μg/L(female), HGH>7.5μg/L.3. The sample was randomly divided into train set (38 cases of lung cancer,39 cases of benign pulmonary diseases) and test set (12 cases of lung cancer,13 cases of benign pulmonary diseases) at the proportion of 3:1. ANN and Fisher discrimination were used to establish classification model by use of training data, then test set data were classified by the models and compared with the two models.4. The SPSS 12.0 and Matlab 7.1 software were used. Methods of representation and examination were based on the distribution of quantitative data. The statistic data were analyzed with Chi-square tests,α=0.05 was the level of test.Results1. The serum levels of AFP,CA125,CEA,NSE and SF in lung cancer patients were significantly higher than those in benign pulmonary diseases respectively (P<0.05). The expression positive rates of CA125,CEA and SF in lung cancer patients were significantly higher than those in benign pulmonary diseases respectively (P<0.05).2. The sensitivity, specificity, accuracy, positive prognostic value and negative prognostic value of fisher discrimination analysis were 58.3%,76.9%,68%,70% and 66.7%, respectively.3. The accuracy of ANN for training set was 90.9%, the sensitivity, specificity, accuracy, positive prognostic value and negative prognostic value of fisher discrimination analysis were 83.3%,92.3%,88%,90.9% and 85.7%, respectively.4. The RUC of ANN was significantly higher than that of fisher discrimination analysis (P<0.05).Conclusion1. The accuracy of ANN combined with nine tumor markers was excellent in distinguishing lung cancer from benign pulmonary diseases.2. The diagnosis and distinguish of lung cancer by ANN combined with nine tumor markers were better than those by fisher discriminatory analysis.