18F-deoxyglucose(FDG)PET/CT has been widely used in the diagnosis,efficacy evaluation,and prognosis prediction of lung tumor.In recent years,artificial intelligence has shown increasing capabilities in the analysis and processing of medical images and has influenced various aspects of molecular imaging,at the same time,radiomics analysis based on artificial intelligence has received great attention in recent years.This article reviews the latest research progress of artificial intelligence in optimizing 18F-FDG PET/CT lung tumor images,explores the new developments and challenges of artificial intelligence in image acquisition,reconstruction,segmentation,denoise,radiomics,etc.,and looks forward to its future applications in lung tumor.

Purpose To develop and validate an integrated model based on preoperative CT radiomic habitat analysis for predicting overall survival(OS)and progression-free survival(PFS)in patients with stage Ⅰ-ⅢA resectable non-small cell lung cancer(NSCLC).Materials and Methods A retrospective analysis was conducted on 341 NSCLC patients from the Second Affiliated Hospital of Naval Medical University from January 2016 to February 2023.Patients were randomly divided into training(n=238)and validation(n=103)cohorts.Tumor volumes of interest were manually delineated using 3D Slicer software.Habitat subregions were segmented via K-means clustering,and radiomic features were extracted using Pyradiomics.Key features were selected using LASSO-Cox regression.Integrated Cox proportional hazards models combining clinical factors and radiomic habitat features were constructed to predict OS and PFS.Model performance was evaluated using the concordance index,receiver operating characteristic curve and Kaplan-Meier analysis.Interpretability was explored via SHAP analysis.Results Habitat subregion analysis demonstrated superior prognostic predictive performance compared to conventional whole-tumor radiomic analysis.In the validation cohort,the integrated OS model and PFS model achieved the highest predictive accuracy,with concordance index 0.878 and 0.795,respectively.Receiver operating characteristic curve analysis showed that the integrated OS model predicted 1-,3-,and 5-year survival with areas under the curve of 0.781,0.878 and 0.881,respectively;while the integrated PFS model achieved areas under the curve of 0.786,0.725 and 0.715.Kaplan-Meier analysis confirmed significant survival differences between high-risk group and low-risk group(P<0.05).Conclusion The integrated model based on CT radiomic habitat analysis non-invasively and accurately predicts OS and PFS in stage Ⅰ-ⅢA NSCLC patients,providing critical insights for personalized treatment decision-making.

18F-deoxyglucose(FDG)PET/CT has been widely used in the diagnosis,efficacy evaluation,and prognosis prediction of lung tumor.In recent years,artificial intelligence has shown increasing capabilities in the analysis and processing of medical images and has influenced various aspects of molecular imaging,at the same time,radiomics analysis based on artificial intelligence has received great attention in recent years.This article reviews the latest research progress of artificial intelligence in optimizing 18F-FDG PET/CT lung tumor images,explores the new developments and challenges of artificial intelligence in image acquisition,reconstruction,segmentation,denoise,radiomics,etc.,and looks forward to its future applications in lung tumor.

Purpose To develop and validate an integrated model based on preoperative CT radiomic habitat analysis for predicting overall survival(OS)and progression-free survival(PFS)in patients with stage Ⅰ-ⅢA resectable non-small cell lung cancer(NSCLC).Materials and Methods A retrospective analysis was conducted on 341 NSCLC patients from the Second Affiliated Hospital of Naval Medical University from January 2016 to February 2023.Patients were randomly divided into training(n=238)and validation(n=103)cohorts.Tumor volumes of interest were manually delineated using 3D Slicer software.Habitat subregions were segmented via K-means clustering,and radiomic features were extracted using Pyradiomics.Key features were selected using LASSO-Cox regression.Integrated Cox proportional hazards models combining clinical factors and radiomic habitat features were constructed to predict OS and PFS.Model performance was evaluated using the concordance index,receiver operating characteristic curve and Kaplan-Meier analysis.Interpretability was explored via SHAP analysis.Results Habitat subregion analysis demonstrated superior prognostic predictive performance compared to conventional whole-tumor radiomic analysis.In the validation cohort,the integrated OS model and PFS model achieved the highest predictive accuracy,with concordance index 0.878 and 0.795,respectively.Receiver operating characteristic curve analysis showed that the integrated OS model predicted 1-,3-,and 5-year survival with areas under the curve of 0.781,0.878 and 0.881,respectively;while the integrated PFS model achieved areas under the curve of 0.786,0.725 and 0.715.Kaplan-Meier analysis confirmed significant survival differences between high-risk group and low-risk group(P<0.05).Conclusion The integrated model based on CT radiomic habitat analysis non-invasively and accurately predicts OS and PFS in stage Ⅰ-ⅢA NSCLC patients,providing critical insights for personalized treatment decision-making.