A Series Of Study Based On MRI And Pathology To Distinguish Brucellar Spondylitis From Tuberculous Spondylitis

BackgroundBrucellar spondylitis(BS)and Tuberculous spondylitis(TS)are infectious spondylitis caused by two different bacteria invading the spine,and the spine is the most common among bone and joint infections.Both of them have slow onset,long course of disease,many complications,high disability rate,and easy recurrence.The incidence of patients with atypical and insidious clinical symptoms and serious illness has increased significantly,which has attracted more and more attention.Strengthening early diagnosis and treatment of BS and TS can not only avoid patients’ disability in pastoral areas and ethnic minority areas,but also reduce the economy burden of patients,and the society,which plays an important role in improving the quality of patients’ life.Therefore,the early diagnosis and differential diagnosis of BS and TS patients are of great significance for patients and society.This study is mainly divided into three parts:The first part is the study based on MR T2 mapping quantitative in BS patients and TS patients,to explore the diagnostic efficacy and differential diagnosis value.The second part is the study based on ANGPTL-4 and histopathology in BS patients and TS patients,to explore the differential diagnosis ability.The third part is the application research of MRI radiomics based on machine learning in BS patients and TS patients,to discusses and evaluates the differential diagnosis ability.Part I Differentiating brucellar spondylitis from tuberculous spondylitis by the conventional MRI and MR T2mappingObjective To investigate the diagnostic efficacy and differential diagnosis value in BS and TS patients by quantitative evaluation of MR T2 mapping.Materials and Methods We performed on 26 BS patients and 27 TS patients in conventional MRI and MR T2 mapping.We analyzed the features in conventional MRI and measured T2 mapping values of the lesion vertebrae(LV)and unaffected adjacent vertebrae(UAV)in BS and TS patients,respectively.Results There were no significant differences in gender,age,or ethnicity between BS and TS patients.It was significantly lower vertebral destruction,vertebral kyphosis,dead bone,and abscess scope in BS patients when compared to TS patients(P < 0.001,P = 0.048,P < 0.001,P < 0.001,respectively).The incidence of vertebral hyperplasia in BS patients was significantly higher than that in TS patients(P < 0.001).T2 mapping value of the LV with BS was significantly higher than that in the UAV with BS,in the LV with TS,and in the UAV with TS(P < 0.001,P < 0.037,P < 0.001,respectively).T2 mapping value of the LV with TS was significantly higher than that in the UAV with TS,and in the UAV with BS(P < 0.001,P < 0.001,respectively).There were no significant differences in the T2 mapping value of the UAV between BS and TS(P = 0.568).Conclusions The qualitative and quantitative evaluation may differentiate BS from TS.The conventional MRI helps to distinguish BS from TS by several distinctive features.MR T2 mapping has the additional potential to provide quantitative information between BS and TS.Part II Differentiating brucellar spondylitis from tuberculous spondylitis based on pathological multimodalityObjective To investigate the differential diagnosis ability in BS and TS patients by the immune protein ANGPTL-4 and histopathology.Materials and Methods In this study,53 patients with BS and TS were selected from the First Affiliated Hospital of Xinjiang Medical University from January 1,2016 to December 31,2018 according to the inclusion and exclusion criteria.The clinical data were analyzed retrospectively,and ANGPTL-4 was detected by histopathology and staining of frozen tissue sections.Result Among the 53 patients,BS had 26 patients,and TS had 27 patients.There was no significant difference between the baseline between the two groups(P = 0.682).The positive rate of ANGPTL-4 in TS patients was higher than that in BS patients(P = 0.042).The incidence of microangiopathy and fibrous connective tissue in BS patients with was distinctly higher than those in TS patients(P = 0.001,P = 0.008,respectively).The incidence of granuloma,caseous necrosis,and epithelioid reaction in TS patients was significantly higher than those in BS patients(P = 0.001,P = 0.003,P < 0.001,respectively).The values of interleukin-6 and C-reactive protein were significantly higher than those in BS patients(P < 0.001,P = 0.001,respectively).Conclusion The study provided novel insights into distinguishing BS from TS using the ANGPTL-4 combining with histopathology,which may become a new supporting evidence.Part III Distinguishing brucellar spondylitis from tuberculous spondylitis by machine learn-based MRI radiomicsObjective To investigate and evaluate the differential diagnosis ability of MRI radiomic based on machine learning in BS patients and TS patients.Materials and Methods MR images of BS and TS patients confirmed by diagnostic criteria were retrospectively analyzed.Patients were randomly divided into a training set(n=251)and a validation set(n=64)according to the ratio of 8:2.The region of interest(ROI)was delineated on the sagittal images of T2 WI,T2WI-FS and T1 WI + T2 WI + T2WI-FS.The Rad Cloud platform was used for feature extraction.The minimum absolute contraction and selection operator(LASSO)is applied for dimension reduction.Logistic regression model was constructed by selecting the most valuable image omics features.Results 27,36,and 24 features were obtained by T2 WI,T2WI-FS,and T1WI+ T2 WI + T2WI-FS sequences,respectively.In the diagnostic performance of three Logistic regression models,the AUC(sensitivity,specificity)for the training set were 0.858(0.80,0.91)for T2 WI,0.822(0.76,0.88)for T2WI-FS,0.883(0.83,0.94)for T1 WI + T2 WI + T2WI-FS,respectively.In three models,the AUC(sensitivity,specificity)for the validation set were 0.724(0.68,0.79)for T2 WI,0.698(0.64,0.76)for T2WI-FS,0.763(0.71,0.82)for T1 WI + T2WI+ T2WI-FS,respectively.Conclusion The MRI-based on machine learning is feasible to differentiate BS from TS,which may be applied to improve preoperative and differential diagnosis accuracy in clinical practice.