The Value Of CT Image Features And Texture Analysis In The Diagnosis And Differential Diagnosis Of Primary Pulmonary Lymphoma

PART 1 The CT image features of pulmonary lymphomaObjective: To explore the CT imaging features of primary pulmonary lymphoma((PPL))in order to improve the understanding and diagnostic accuracy of the disease.Methods: The chest CT findings and related clinical data of 34 patients with pulmonary lymphoma confirmed by pathology in the first affiliated Hospital of Southern Anhui Medical College from July 2012 to December 2020 were analyzed retrospectively.the main imaging signs included focus shape,location,lesion density and special signs.Results: Among the 34 cases of primary pulmonary lymphoma,there were 12 cases of nodular mass type,19 cases of non-nodular mass type and 3 cases of mixed type.At the same time,bilateral lungs were involved in 13 cases,solid density in 17 cases,mixed density in 17 cases,average CT value of 29 HU,vacuole sign in 15 cases,cavity sign in 6 cases,angiographic sign in 15 cases,air bronchial sign in 30 cases,bronchiectasis in 22 cases,bronchial stenosis or truncation in 8 cases,pleural effusion in 12 cases,enlarged lymph nodes in 15 cases and pleural metastasis in 2 cases.The final pathological results included 24 cases of mucosa-associated lymphoid tissue lymphoma((MALT)),9 cases of diffuse large B-cell lymphoma((DLBCL))and 1case of T-cell lymphoma.Conclusion:The(PPL)CT findings of primary pulmonary lymphoma are varied and have no obvious characteristics,which are related to pathological classification.Air bronchial signs,bronchiectasis,angiographic signs and other signs are of great significance for the diagnosis of PPL.PART 2 The Application of Imaging Radiomics in the Differentiation of Pulmonary Lymphoma and Organized PneumoniaObjective: To explore the value of radiomics in the differential diagnosis of pulmonary lymphoma and organizing pneumonia.Methods: The clinical and imaging data of 34 patients with pulmonary lymphoma and 51 patients with organizing pneumonia were retrospectively analyzed.These patients were confirmed by pathology at the First Affiliated Hospital of Wannan Medical College from July 2012 to April 2020,and randomly divided into training cohort and verification cohort.396 texture features were extracted by AK software and the redundant minimum correlation method(m RMR)was used to eliminate redundant and irrelevant features.Three models,namely,image feature model,image radiomics label and combined model,were established by LASSO-Logistic regression analysis.The Rad-score of each patient was calculated,and the combined model was represented by a nomogram.The ROC curve was used to evaluate the efficiency of each model in distinguishing pulmonary lymphoma and organizing pneumonia.The consistency between the nomogram and the observation value was evaluated by the calibration curve.The Hosmer-Lemeshow test was used to determine the goodness of fit among each model.Decision curve analysis(DCA)was used to judge the clinical utility value of each model.Results: There was no significant difference in sex and age between PPL group and organic pneumonia group(P > 0.05).The imaging feature model was established by preserving air bronchial sign,bronchiectasis and pleural thickening,and its AUC value was 0.93A(95%CI:0.86-0.99A).Quantitative texture features retain 5 best feature subsets after dimension reduction by m RMR and LASSO to establish imaging tag.The AUC value of imaging tag is 0.81(95%CI:0.70-0.92).The prediction efficiency of the line diagram of the joint model composed of the two is the best,and its AUC value is 0.96(95%CI:0.92-1.00).DCA results show that in a certain probability range,the combination of imaging tags and image features has higher clinical value in the differential diagnosis of the two diseases.Conclusion: The combined model nomogram composed of image feature model and imaging radiomics label can effectively identify pulmonary lymphoma and organizing pneumonia,and has good clinical utility value.