Differential Diagnosis Of Colorectal Mucinous Adenocarcinoma And Non-mucinous Adenocarcinoma By Multimodal Imaging

PART1 Differential diagnosis of colorectal mucinous adenocarcinoma and non-mucinous adenocarcinoma by CT image features combined with radiomicsObjective: Based on CT examination,to explore the differential diagnostic value of image features,radiomics models for colorectal mucinous adenocarcinoma(MAC)and non-mucinous adenocarcinoma(non-MAC),and to evaluate whether the merge model of image features and radiomics has higher diagnostic value.Materials and Methods: The preoperative CT non-enhanced and enhanced images of 49 patients with colorectal MAC and 63 patients with non-MAC confirmed by surgery and pathology in the Affiliated Hospital of Guilin Medical College were collected retrospectively.CT image features of the lesions were evaluated by two physicians.ITK-SNAP is used to outline the region of interest(ROI)at the largest level of the lesion,and the image is imported into GE Artistic Intelligent Kit to extract radiomics features.The radiomics features with intraclass correlation coefficient(ICC)> 0.75 were included,and the radiomics features related to pathology were screened by variance threshold method and Lasso regression.The non-enhaned and enhanced radiomics models were constructed respectively,and the non-enhanced and enhanced radiomics score(Rad-score)were calculated.The merge model is established by combining image features,non-enhanced Rad-score and enhanced Rad-score.Draw receiver operating characteristic(ROC)of each model subject and calculate area under the curve(AUC).Results: There are differences between colorectal MAC and non-MAC on CT images.In the validation set,the AUC of image feature model for differential diagnosis is 0.85,and the sensitivity and specificity are 46.7% and 94.7%.The non-enhanced radiomics model consists of 10 features,and the AUC in the validation set is 0.82,sensitivity and specificity are 86.7% and 100%.The enhanced model consists of 9 features,and the AUC in the validation set is 0.92,and the sensitivity and specificity are 86.7% and 100%.In the merge model,only enhanced Rad-score was an independent risk factor(OR: 2.36,P=0.002),and the ROC curve of the merge model was the same as that of the enhanced radiomics model.The diagnostic efficiency of the enhanced radiomics model in the validation set(merge model)was better than that of the image features and non-enhanced radiomics model(P=0.048,0.047).Conclusions: Colorectal MAC is different from non-MAC in CT images,but the diagnostic model based on image features has some limitations.The enhanced radiomics model(merge model)has the best diagnostic performance,which is better than the image feature model and non-enhanced image radiomics.It is of great significance in differentiating colorectal MAC from non-MAC.PART2 Comparative study of synthetic MRI and DWI in differential diagnosis of rectal mucinous adenocarcinoma and non-mucinous adenocarcinomaObjective: To investigate the diagnostic value of synthetic MRI(syMRI)parameters and ADC in rectal MAC and non-MAC,and to compare the diagnostic efficacy of different parameters.Materials and Methods: The clinical and imaging data of patients with rectal MAC and non-MAC confirmed by pathology in the Affiliated Hospital of Guilin Medical College were collected(including MAC 10 cases and non-MAC 30 cases).The patient received routine pelvic MRI and syMRI examination within 2 weeks before operation.Two physicians manually delineated the ROI layer by layer along the edge of the lesion on the workstation to obtain the T1,T2,PD and ADC values of the lesion,and ICC was used to evaluate the inter-observer consistency.The differences in T1,T2,PD and ADC values of rectal MAC and non-MAC were compared,and the ROC curve was drawn to evaluate the diagnostic efficacy and calculate the corresponding AUC.Results: The T1,T2,PD and ADC values of rectal MAC were greater than those of non-MAC,and the diagnostic AUC were 0.860,0.955,0.766 and 0.899 respectively.The diagnostic efficacy of T2 value is better than PD value,and there is no difference among T1,T2 and ADC values.Conclusions: The T1,T2,PD values in syMRI parameters and the ADC value in DWI imaging can be used to differentiate between rectal MAC and nonMAC,and the diagnostic efficacy of T2 value in all parameters is relatively high.