The Value Of CT Texture Analysis In The Study Of Regional Lymph Node Metastasis In Laryngeal Supracloglottic Squamous Cell Carcinoma

Research backgroundElectronic computed tomography(CT)is a common method for diagnosing solid focal lesions in the head and neck,Texture analysis(TA)can provide objective and quantitative image description features.Laryngeal supraglottic type squamous cell carcinoma(LSGSCC)is a common malignant tumor in otolaryngology head and neck science,regional lymph node metastasis is a key factor affecting the treatment and prognosis of laryngeal supracloglottic squamous cell carcinoma.ObjectiveTo explore the feasibility of preoperative prediction of regional lymph node metastasis in patients with supraglottic squamous cell carcinoma by extracting texture analysis features from primary tumor lesions CT images.Clinical dataA retrospective analysis of neck CT images of 38 laryngeal supracloglottic squamous cell carcinoma patients confirmed by surgery and postoperative pathology in our department from January 2010 to December 2018.All patients had complete clinical data,all selected patients underwent surgical treatment and neck nodal dissection(NND).After one year follow-up,no tumor recurrence was found.After the operation,pathological examination was performed and there were definite pTNM stages.According to the results of postoperative pathological examination,18 cases had no metastatic cervical lymph node and 20 cases had metastatic cervical lymph node involvement,pN0 18?pN1 12?pN2 8.Its further divided into nonmetastatic lymph nodes(NLN)namely pN0(n=18)and metastatic lymph node(MLN)namely pN+(n=20).Research methodsSelect the largest image of the tumor lesion on the enhanced CT image and use the MaZda4.6 software to manually draw the regions of interest(ROI)along the edge of the lesion on the CT enhanced image.The texture parameters are extracted using the software's own extraction method,those who appear twice or more are included in the study,and finally 5 texture parameters are extracted:Mean value,Skewness,Correlation,Entropy and WavEnHLS-2.Compare the difference in texture parameters between the venous and arterial phases in the pN0 group and pN+group.Draw receiver operator characteristic curve(ROC)for parameters with statistical differences,calculate area under the ROC curve(AUC),conduct diagnostic efficiency analysis,calculate sensitivity and specificity.The independent predictors were calculated by binary Logistic regression analysis to distinguish pN0 group from pN+group.SPSS22.0 software was used to calculate the joint predictor factor of the independent predictor,the ROC curve of the joint factor was plotted to analyze the diagnostic efficacy.Result1.Difference test:Venous phase and arterial phase Mean value,Skewness,wavenhls-2 in the pN+group were higher than that in the pN0 group,P>0.05 showed no statistical significance.Venous phase,Correlation(P=0.004)and Entropy(P=0.001)were higher in the pN+group than in the pN0 group,P<0.05 was statistically significant.Arterial phase Correlation(P=0.008)and Entropy(P<0.001)were higher in the pN+group than in the pN0 group,P<0.05 was statistically significant.2.3.Diagnostic efficiency:VEntropy,the AUC=0.875,the sensitivity and specificity were 85.0%and 77.8%;AEntropy,the AUC=0.872,the sensitivity and specificity were 80.0%and 83.3%.VCorrelation,AUC=0.764,the sensitivity and specificity were 90.0%,61.1%;ACorrelation,AUC=0.75,the sensitivity and specificity were 75.0%and 77.8%.3.Binary logistic regression analysis:VEntropy and AEntropy is an independent predictor to identify the pN0 group and the pN+group.The diagnostic efficacy of the combined factors of the two independent predictors was 0.942,the sensitivity and specificity were 85.0%and 94.4%.It could be seen that the diagnostic efficacy of the combined predictor factors was greater than that of the single independent predictors.Conclusion1.Correlation and Entropy of texture parameters can predict the lymph node metastasis in LSGSCC region before operation.Texture analysis may become a new clinical diagnosis and treatment technology.2.The entropy of venous phase and arterial phase is an independent predictor.The combined predictor of independent predictor can improve the diagnosis efficiency and has high sensitivity and specificity.