The Applied Research Of Serum Protein Spectrum Of Esophageal Carcinoma Combined With Artificial Neural Network Diagnosis System For Esophageal Carcinoma Pathology Differentiation Degree

Abstract:Objective:Esophageal carcinoma is a malignant tumor of digestive system and it is has a high disease incidence in our country, the patients'5-year survival rate even after opration is still low, just about 8%-30%.In recent years,to chose individual treatment for the cancer patients attracts more attention because of the better recent and in the long term curative effect than the traditional treatment,but it depends on the right clinical diagnosis of pathology differentiation degree for the cancer patients.For the patients with esophageal cacinoma the correct clinical diagnosis about the cacinoma differentiation is very useful for chosing individual treatment. Protein spectrum technology is a wide range of disease markers analysis platform based on protein chip and mass spectrometry technology,it helps filter out associated molecular changes of esophageal cancer.Now,According to the serum protein spectrum specific expression to establishment the artificial neural network diagnostic system of esophageal cancer, and to research the significance at the pathology diagnosis of esophageal cancer. Methods:Collection of the serum of the experimental group (60 cases of esophageal cancer patients,23 cases of poorly differentiated patients,11 cases of moderately diffenrentiated patients,26 cases of well-diffenrented patients),the control group (30 cases of normal people),amount to 90 cases of the serum samples. All serum samples were detected using surface-enhanced laser desorption ionization time of flight mass spectrometry (SELDI-TOF-MS) combined with gold chips, using Ciphergen Proteinehip software and Biomarker Wizard 3.1 software to obtain mixed patterns of all serum for statistical analysis and calculate their differences in order to filter out the specific expression of protein; using specific expression of M/Z combined with artificial neural network software establish the diagnosis system of esophageal carcinoma pathology differentiation degree.Results:After experimental detected, we found four differences protein peaks (2332.2,4051.1,4260.0和4267.1, M/Z)(P<0.05)from 90 cases of the serum samples, and to establishment the artificial neural networks diagnostic system of esophageal carcinoma pathology differentiation degree. Then, using double-blind method to detected the artificial neural network diagnosis system of esophageal carcinoma pathology differentiation degree,to verify the diagnosis system of sensitivity and specificity. Statistical analysis showed that We establishment the diagnostic system of esophageal cancer in the cut-off of 0.3, the detection of esophageal carcinoma patients with a sensitivity of 88.7% and specificity was 81.3%, diagnose accordance rate 86.1%. Conclusion: 1.Four differences M/Z with specific expression in serum of esophageal carcinoma were found successfully; 2. Successfully established a artificial neural network diagnosis system of esophageal carcinoma pathology differentiation degree, and be verified with double-blind method, and obtain to satisfactory sensitivity and specificity from the objective indicators; 3. The value of diagnostic efficiency analysis of ANNs system tell us, when the cut-off value of 0.3, the detection of esophageal carcinoma patients with a sensitivity of 87%and specificity is 83.8%, and the positive predictive value is 76.9%,negative predictive value is 91.2%,diagnose accordance rate 85%. It could be applied to esophageal carcinoma patients with different differentiation degree.