Differential Diagnosis Of Parotid Gland Malignant Tumor And Adenolymphoma Based On MRI Radiomics

Objective:To investigate the value of MRI-based imaging in the differentiation of parotid gland malignant tumor from adenolymphoma.Materials and Methods:A retrospective study analyzed 86 patients with parotid gland tumor medical pain admitted to our hospital from September 2016 to December 2022,including 39 patients with malignant tumors and 47 patients with adenolymphomas.Collect the age,gender and MRI imaging characteristics(number of lesions,existing form,boundary,whether the digital signal of lesions is average,whether there is cystic change area and ADC average value)of two major types of cancer patients with medical pain,and use IBM SPSS 26.0 to carry out statistical research and analysis of data information,and classify the application of change amount χ 2.Detection or Fisher’s accurate detection,continuous change of quantity shall be tested by independent standard sample t or Mann-Whitney U.The first step is to obtain the coordinate horizontal T2 WI FS and ADC images from the PACS system and automatically input them into the Shenrui medical and nursing science inquiry application operation platform,manually segment and delineate the largest level of tumor in the MRI images of all medical patients with pain,and quantitatively select the imaging characteristics through the region of interest(ROI),After that,the selected imageomics characteristics were analyzed and selected by using the minimum redundancy maximum correlation standard algorithm(m RMR)and the minimum absolute contraction function and the selection procedure operator standard algorithm(LASSO)regression analysis to build the imageomics standard model;Through the analysis of single influencing factor and multiple influencing factors,logical thinking regression research,analysis and analysis of clinical independent risk influencing factors,and parallel connection of group photo markers to create a collaborative standard model,and then draw and design a nomograph of imaging group,and apply the actual effective area under the curve(AUC),sensitivity The specificity and accuracy of the four standard models(clinical standard model,T2 WI FS imageomics standard model,ADC imageomics standard model and collaborative standard model)were evaluated for the effect of identification,diagnosis and treatment judgment,and the Delong test was used to evaluate the different differences of AUC between random two categories of standard models in each group.Results:According to T2WI-FS,the AUC of the imaging group standard model in the training group and the validation group was 0.816(95.0% CI: 0.686-0.946),0.759(95.0% CI:0.528-0.990);According to the standard model of imageomics created by ADC,the AUC of the training group and the validation group were 0.890(95.0% CI:0.798-0.982)and 0.830(95.0% CI: 0.657-0.904)respectively;The single influencing factor and multiple influencing factor analysis logical thinking regression analysis showed that the average value of ADC was an independent risk factor for predicting parotid gland malignant tumor,so the AUC of the clinical standard model in the training group and the validation group was 0.867(95.0% CI: 0.731-0.942),0.866(95.0% CI: 0.655-0.923).The AUC of the collaborative standard model obtained by combining the standard model of imageology and clinical standard model in the training group and the validation group were 0.951(95.0% CI: 0.888-1)and 0.884(95.0% CI: 0.743-1)respectively.The final result of De Long’s test showed that in the training group and the validation group,the P value of the collaborative standard model and the clinical standard model and the Rad score-T2 WI standard model were both lower than 0.05,and the difference had statistical analysis significance.The P value of the Rad score-ADC standard model and the clinical standard model wereboth lower than 0.05,and the difference had statistical analysis significance.The distribution function curve and strategy distribution function curve of proofreading and correction analysis show that the nomograph of imageology shows good proofreading and correction analysis ability and clinical net benefit.Conclusion:According to the standard imaging model of MRI and the clinical influence factors,the imageomics nomogram created has excellent diagnostic and therapeutic value in the identification of parotid gland malignant tumors and adenolymphoma,and has higher predictive effect compared with the clinical standard model and the standard imaging model created with T2WI-FS,and has statistical analysis effect.