Prediction Of Occult Lymph Node Metastasis Using Texture Analysis Of ¹⁸F-FDG PET Image In Early Non-small Cell Lung Cancer

Objective:Because occult lymph node metastasis has a high incidence in the treatment of patients with non-small cell lung cancer?NSCLC?and the diagnostic efficiency ofnon-invasive diagnostic methods is limited,it has become an important factor affecting the efficacy and prognosis of patients.The study found that the baseline ¹⁸F-FDG PET/CTprimary tumor metabolic parameters,volume parameters have a certain value in the prediction of occult lymph node metastasis in early NSCLC,and the emerging FDG PET image texture analysis technology in addition to providing the metabolic information,but also Providesinformation on tumor metabolism heterogeneity.The purpose of this study was to usebaseline FDG PET/CT image texture analysis techniques to construct an optimal risk model for early occult lymph node metastasis in NSCLC.Methods:Baseline ¹⁸F-FDG PET/CT images of NSCLC patients from Shandong cancer Hospital Affiliated to Shandong University were retrospectively collected.The ¹⁸F-FDG PET DICOM data before treatment was introduced into the Matlab platform,and the SUV=2.5threshold was used to outline the region of interest of the primary tumor,A total of 60parameters of the tumor primary tumor were extracted.SUVs(SUV_max,SUVmean,SUV_min,SUVpeak,etc.)volume parameters?MTV,TLG?and texture parameters?5 overall;26 local;26 regional?.SPSS21.0 and MedCale 12.7 software was performed for the following statistical analysis.P<0.05 has a significant statistical difference.1.First,normality?Kolmogorov-Smimov test?and homogeneity test?Levene test?are used to extract all parameters.2.The normality and variance homogeneity parameters use two independent sample T-tests,and the remaining parameters use M-W U nonparametric test to compare thedifferences between groups.3.Pearson chi-square test was used to analyze the differences between groups.4.Receiver operating characteristic curve?ROC curve?analysis and screening parameters to predict the threshold and test efficiency of occult lymph node metastasis.5.Multivariate risk regression model was used to predict the risk factors of occult lymph node metastasis.Results:A total of 75 patients met the inclusion criteria.17.33%?13/75?had occult lymph node involvement.10.67%?8/13?were pN1 and 6.67%?5/13?were pN2.Theperitubular area and the 7th area are common occult lymph node metastatic sites.In terms of univariate analysis,except for tumor location?P=0.004?,there was nostatistical difference between the other clinicopathological factors.such as tumor size,pleural involvement,and solid nodules;there were no statistics differences between SUV-relatedparameters and volume parameters.Local texture parameters Homogeneity,Inversedifference moment,and Code similarity were significantly higher than those without lymph node metastasis.Tumor location?P=0.012,OR=7.615?and local texture parameters?Homogeneity,P=0.012,OR=7.615;IDM,P=0.038,OR=4.321;Code similarity,P=0.001,OR=0.560?as occult lymph nodes Risk factors for metastasis.In multivariate analysis:tumor location?P=0.003,OR=9.369?and Homogeneity?P=0.008,OR=10.688?were independent risk factors for occult lymph node metastasis in early NSCLC.In terms of predictive efficacy,the accuracy of local texture parameters in predicting occult lymph node metastasis in NSCLC patients was only 57.33%,54.67%,and 65.33%.86.67%,86.67%,and 81.33%of tumor location and local texture parameters were combined;accuracy was 80.00%,80.00%,and 57.14%in 61 patients with lung adenocarcinoma;patients with solid nodules with lung adenocarcinoma of 52 The accuracy is 66.67%,80.00%,and57.14%.Conclusion:Local texture parameters on baseline ¹⁸F-FDG PET/CT were significantpredictors for occult lymph node metastasis in T_1-2N₀M₀ NSCLC patients.Therefore,in order to reduce the risk of hidden residual lymph node metastases in patients,appropriatepreoperative evaluation and treatment methods can be selected according to tumor location and texture parameters.