Diagnosis Of Benign And Malignant Lung Masses Using CT Image Texture Analysis Based On Statistical Analysis

Objective: To compare the gray histogram and gray co-occurrence matrix texture features of benign and malignant lung masses,and to study the diagnostic value of gray histogram and gray co-occurrence matrix texture feature parameters in benign and malignant lung mass.Method:A retrospective analysis of 110 patients with lung masses confirmed by our hospital’s surgical pathology from April 2019 to October 2019 in Inner Mongolia Autonomous Region People’s Hospital,of which 50 were benign and the pathological types were chronic inflammation of the bronchial mucosa of the lung;60 were malignant,Including 35 cases of squamous cell carcinoma and 25 cases of adenocarcinoma.Scanned by German SIEMENS dazzling-speed dual-source CT and GE64-row spiral CT machine,uploaded to the GE post-processing workstation after the scan,and observed the image with a mediastinal window(window width: 400 HU,window level: 50 HU),and the largest display nodule Diameter axial image,transfer the central layer of the lesion into Ma Zda[Post-processing software(Version 4.6)] for texture analysis,and manually outline the region of interest(ROI)along the contour of the lesion.The software will automatically generate the grayscale histogram and grayscale co-occurrence matrix parameters,including the mean,kurtosis,variance,skewness,and percentiles of the grayscale histogram(1%,10%,50%,90%,99%)And other parameters and the anglesecond-order moments,homogeneity,contrast,correlation and other parameters of the gray level co-occurrence matrix,using SPSS 19.0 statistical analysis software.The measurement data are described statistically with(?X ± S).The measurement data that conforms to the normal distribution are compared using two independent sample t tests.For measurement data that do not meet the normal distribution,the Wilcoxon rank sum test is used to compare the two groups.P <0.05 was considered statistically significant.Establish receiver operating characterist(ROC)curve and obtain area under curve(AUC),determine the optimal critical value of each parameter,calculate the sensitivity and specificity,and finally compare its diagnostic efficacy analysis.Results:1.The mean,Perc1%,Perc10%,Perc50%,and Perc90% of the texture parameters in the gray histogram of benign and malignant lung lesions were statistically significant in the benign and malignant two groups(P <0.05),and the other texture parameters were not Statistical significance,further analyze the ROC curve of meaningful texture parameters;when analyzing the benign and malignant lung lesions using gray histogram texture features,the AUC of each parameter is less than 0.700,representing a certain diagnostic efficacy,but Less accurate.2.The second-order moments,homogeneity,contrast,and mean of the angles in the texture of the gray-scale symbiotic matrix of benign and malignant lung lesions were statistically significant in the benign and malignant groups(P <0.05),and the other texture parameters were not statistically significant.Significance;Perform ROC curve analysis on 4 parameters with statistically different texture parameters,and calculate the Area Under Curve(AUC)to compare the diagnostic performance of each texture parameter.Significant improvements have been made.Different combinations of the above-mentioned statistically significant texture parameters and the establishment of ROC curves have found that the different combinations are more effective in differentiating and diagnosing benign and malignant lung masses(all AUC> 0.800),and the accuracy of diagnosis is all Higher.Conclusions:1.Gray histogram texture features have certain diagnostic power and low accuracy in the differential diagnosis of benign and malignant lung masses.2.Gray-scale symbiosis matrix texture featuresand the combination of various texture features have higher diagnostic effectiveness in the differential diagnosis of benign and malignant lung masses.Among them,the contrast has a large difference between benign and malignant lung masses.You can study contrast Prospective study of the differences in benign and malignant masses.