Establishment of a rapid differentiation model to distinguish malignancy from benignancy of intraop-erative lung adenocarcinoma tissues by probe electrospray ionization mass spectrometry combining with artificial intelligence
Objective:To develop and evaluate an intraoperative rapid diagnostic model for distinguishing ma-lignant lesions from benign ones by probe electrospray ionization mass spectrometry combining with artificial intelligence for unfixed lung surgical specimens.Methods:A total of 47 surgical specimens of lung adenocarci-noma and peripheral non-cancerous tissue were collected from Department of Pathology in China-Japan Friendship Hospital from July to December 2022.After weighing,extraction,centrifugation and other processing,the supernatant was directly scanned and analyzed using a rapid ambient mass spectrometer DPiMS QT.A self-developed program was adopted for data reading,preprocessing,multivariate analysis,feature screening,model training and optimization with Anaconda.Metaboanalyst® was used to analyze the metabolic pathway of characteristic metabolites found in lung adenocarcinoma.Results:Using the PLS regression method for di-mensionality reduction analysis,10 principal components were synthesized using 199 features for the determi-nation of malignant tumors and non-cancerous tissues.Ten models were developed with algorithms,such as support vector machine(SVM),random forest(RF)and multilayer perceptron(MLP),and it was found that the classification model based on MLP had the best accuracy for the 20-fold cross-validation(99％±2％).It was also found that glutamine was down-regulated and glutamic acid was up-regulated in lung adenocarcinoma,which was consistent with the regulation via TCA cycle in lung cancer,which would promote protein synthe-sis.Conclusion:Probe electrospray ionization mass spectrometry combined with artificial intelligence technology can be used to differentiate between benign and malignant lesions in the lungs within minutes,achieving rap-id and accurate classification of fresh lung tissue specimens and assisting in intraoperative rapid diagnosis of lung tissue.In addition,the differentially expressed substances found in lung adenocarcinoma can also be used for further tumor metabolism research.
probe electrospray ionization mass spectrometrylung neoplasmsadenocarcinomaartificial intelligence
NETL
NSTL
万方数据
