Efficacy of clinical factors-based MRI balanced steady-state free-forward sequence imaging histology for diagnosing extramural vascular invasion in rectal cancer

ZHANG Jun; ZHANG Haiqing; YU Bing; SHEN Qianli; ZHAO Hui; LI Junjun

doi:10.12122/j.issn.1674-4500.2023.06.14

Abstract：Objective To evaluate the effectiveness of combining MRI balanced steady-state free precession(b SSFP)sequence imaging histology based on clinical factors for diagnosing extra-mural vascular invasion(EMVI)in rectal cancer.Methods A total of 135 patients were included in the study whose diagnosed with rectal cancer and treated at the Lujiang Branch of Dongfang Hospital,affiliated with Tongji University.71 cases were EMVI-positive,and 64 cases were EMVI-negative among the patients,all of which were confirmed by postoperative pathologic macroscopic specimens.Firstly,the imaging histological features were extracted from the conventional T2WI sequences and b SSFP sequences.The dataset was randomly divided into a training set(n=94)and a validation set(n=41)at a ratio of 7∶3.The imaging histological features and clinical features were analyzed,and the imaging histological model were constructed,by the t-test/Wilcoxon rank sum test,and least absolute shrinkage,and selection operator algorithms were utilized for analysis,and Spearman correlation analysis and logistic regression model.Subsequently,ROC curves were plotted,and the area under the ROC curve(AUC),specificity,and sensitivity were calculated.In order to assess and compare the diagnostic efficacy of the two sequential imaging histologies,it was imperative to quantify the discriminatory performance of the model for both the training and validation groups.Subsequently,the clinical factors were integrated to construct a joint clinical factor and b SSFP sequential imaging histology model for the assessment and validation of diagnostic efficacy.Results The imaging histology models displayed strong discriminatory ability in both the training and validation sets.Specifically,the training set of T2WI sequences demonstrated a diagnostic AUC of 0.751,a specificity of 0.746,and a sensitivity of 0.825,while the b SSFP sequences yielded an AUC of 0.829,a specificity of 0.778,and a sensitivity of 0.939.The validation set T2WI sequence imaging group demonstrated a pre-diagnostic AUC of 0.676,a specificity of 0.650,and a sensitivity of 0.706.In contrast,the b SSFP sequence exhibited an AUC of 0.792,a specificity of 0.716,and a sensitivity of 0.909.Comparison between the two indicates the latter's improved diagnostic efficacy and tangible clinical applicability.The combined clinical factors with b SSFP sequence imaging histology joint model exhibited an AUC of 0.954,specificity of 0.891,and sensitivity of 0.930 for the training set.For the validation set,the AUC was 0.915,specificity was 0.803,and sensitivity was 0.936.Both results were deemed statistically significant(P<0.05).Conclusion The combination of clinical factors and b SSFP sequence imaging histology demonstrates strong diagnostic efficacy for rectal cancer EMVI.This facilitates imaging physicians and clinicians in making accurate differential diagnoses and formulating appropriate treatment strategies.