The value of artificial intelligence-based CT radiomic model in classification and diagnosis of bacterial pneumonia
DING Chunyang
YIN Gongjian
Abstract：Objective To construct a multi-classification model of bacterial pneumonia based on radiomic CT images of artifi-cial intelligence,and to evaluate the diagnostic value of the model.Methods The clinical data of 389 patients with bacterial pneumonia were retrospectively analyzed,and 9 learning models were evaluated based on the radiomic features extracted from CT images.Three optimal sub-models were constructed and integrated together to form a multi-classification model.Receiver op-erating characteristic curve(ROC)and area under curve(AUC)were used to evaluate the classification diagnostic efficiency of the model.Results Three radiation sub-models were established with 5 characteristics:Gram-positive model,Gram-negative model and atypical model.The mean area under the curve(AUC)of the integrated radiomics model in the training set was 0.75(95%CI:0.65-0.83),the accuracy was 0.58,the sensitivity was 0.57,and the specificity was 0.78.The mean AUC in the test set was 0.73(95%CI:0.61-0.79),accuracy was 0.54,sensitivity was 0.52,and specificity was 0.75.Conclusion The multi-classification model of bacterial pneumonia based on artificial intelligence radiomic CT images has a certain value in the differen-tial diagnosis of gram-positive,gram-negative and atypical bacterial pneumonia,which is helpful to assist empirical antibiotic therapy in clinic.
Keywords：RadiomicsArtificial intelligenceTomographyX-ray computerBacterial pneumonia
Publication Date：2025-05-30
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：4( 59-62 )