Evaluation Of The Bone Marrow Edema Activity Of Sacroiliac Joint In Ankylosing Spondylitis By Magnetic Resonance STIR And DWI Quantitative Parameters And Radiomics

BackgroundAnkylosing spondylitis(AS)is a common seronegative spondyloarthropathy.Although its etiology is not completely understood,AS is more common in young men.AS usually starts from sacroiliac joints before slowly extending upward to the central axis bones and surrounding secondary structures,eventually resulting into bone ankylosis and joint malformation.It is essential to early diagnose and delay joint distortion by therapeutic intervention.Evaluation of the activity of AS is key in guiding the clinical basic management and intensive therapy by drug combination.To date,there is no the gold standard of to evaluate the activity of AS.The Common methods in assessing AS include bath ankylosing spondylitis functional index(BASFI)or bath ankylosing spondylitis disease activity index(BASDAI)scoring,the HLA-B27,ESR,CRP of the laboratory serological indexes,and various imaging examinations including X-ray,US,CT,MRI,etc.ASDAS-CRP and ASDAS-ESR as two new comprehensive assessment methods were proposed by Assess Spondyloarthritis International Society(ASAS).However,these kinds of activity indexes are received widely by clinicians owing to complicating calculation formula.X-ray and CT could clearly show the changes of bone structure,However,there are some limitations in the early diagnosis or activity evaluation of AS.Ultrasound is difficult to show the whole picture of anatomical structure and pathological changes of sacroiliac joint because of bone interference.MRI is widely used in osteoarthropathy,and considered the most sensitive and specific imaging examination tool in the diagnosis and treatment of AS.MRI has great advantages in activity evaluation and therapeutic effect of AS,and MRI scoring system and quantitative parameters of multi-functional MRI are regarded as the main method to assess the illness state of AS.MRI scoring system include spondyloarthritis research consortium of canada(SPARCC),sacroiliac joint structural score(SSS),Berlin Sieperl/Rudwaleit score,Berlin Hermannl/Bollow score,Aarhus score,Lanss score etc.SPARCC score is a kind of inflammation score,mainly judging the early diagnosis and activity of AS.SSS score is a structural score,which mainly observe the structural changes of fat deposition,bone erosion and fibrous or bony ankylosis.Functional MRI sequences mainly include DWI,IVIM,DTI,T2*mapping and semi-quantitative and quantitative analysis of DCE-MRI.SPARCC score has high reliability and sensitivity in evaluating the activity of AS in MRI scoring system,and DWI was more sensitive to bone marrow edema of sacroiliac joint in functional MRI.Although these methods have been widely studied,the conclusions are not uniform.In recent years,Radiomics(Imageomics)transforms traditional image into high recognition rate spatial data.Texture classification is obtained by special computing method of machine to provide the valuable information for the prediction,diagnosis,treatment and prognosis of clinical diseases.Radiology has become a hot spot in clinical research,especially in the differential diagnosis of benign and malignant tumors according to large sums of documentation at home and abroad.However,the application of texture analysis in musculoskeletal diseases is still in the exploratory stage.This paper consists of two parts.In the first part,SPARCC scoring and ADC values were obtained by STIR and DWI sequences were quantitatively evaluated the activity of marrow edema in sacroiliac joint in early AS,And to further explore the correlation and efficacy evaluation with commonly used clinical activity indicators(BASDAI,ESR,CRP,ASDAS-ESR,ASDAS-CRP).In the second part,we discussed the application value of imaging texture of STIR sequenceanalys in the activity of bone marrow edema of sacroiliac joint in early stage of AS,and compared the efficacy with SPARCC score and ADC value,and analyzed the correlation with some clinical activity evaluation indexes(BASDAI,ESR,CRP,ASDAS-ESR,ASDAS-CRP).Part ?:Application of STIR and DWI Sequences in evaluating the Activity of Bone Marrow Edema of Sacroiliac Joint in Early Ankylosing SpondylitisObjective:To evaluate the value of SPARCC score and ADC value in evaluating the activity of sacroiliac joint bone marrow edema in early AS,and to explore the correlation between SPARCC score and clinical activity evaluation indexes,which could provide more reliable reference for clinical evaluation of AS activity.Materials and Methods:A total of 43 patients with early AS at hospital were included in our study.Out of the total patients,30 were male(69.77%)while 13 were female(30.23%),ranging in age from 14 to 51 years old,with a median age of(26.52 ± 3.28)years old.All the patients underwent MR examination and then the T1WI,T2WI,STIR and DWI sequences were performed.Evaluation index of clinical activity:the average value of BASDAI score was obtained by two clinicians who had been in rheumatology department for more than 5 years,and patients were divided into active group(n=28)or stable group(n=15)based on BASDAI scoring.Following the determination of HLA-B27,ESR and CRP values,we computed ASDAS-ESR and ASDAS-CRP values.MRI quantitative parameters:SPARCC score was obtained by two MRI diagnostic physicians more than 5 years on STIR sequence,then averaged the results.The ADC value was measured separately on the original DWI image by two physician,then averaged the results.The clinical data were analyzed by ?2 test.All indicators were expressed by median ± quartile interval,because these values did not conform to normal distribution after normal analysis.The clinical activity evaluation indexes(ESR,CRP,ASDAS-ESR,ASDAS-CRP,BASDAI)and MRI quantitative parameters(SPARCC score,ADC value)were tested by non-parametric independent sample test(Mann-Whitney U test)between the active group and the stable group.The ADC values of bilateral sacroiliac joints were compared by non-parametric paired test(Wilcoxon Signed-Rank test).Spearman correlation test was used to test the correlation between MRI quantitative parameters and clinical activity indexes.In addition,we used ROC curves to evaluate the effect of MRI parameters and clinical inflammatory indexes.Results:1.Clinical general data and evaluation index of clinical activityAmong those 43 patients,there were 28 cases in the active group(20 males and 8 females)and 15 cases in the stable group(10 males and 5 females).The positive rate of HLA-B27 in the active group was 92.9%,while that in the stable group was 93.3%,with no statistical difference.The values of ESR,CRP,ASDAS-ESR,ASDAS-CRP and BASDAI scores were 21.00±16.25mm/h,12.60±15.92mg/L,2.73±1.25,3.13±1.64 and 5.60±1.10 score in the active group,and 14.00±11.00mm/h,3.03±2.46mg/L,1.27±0.57,1.41±0.77 and 4.90±0.80 score in the stable group.These values were significantly different between the two groups(P<0.05).2.MRI quantitative parametersThe SPARCC score in the active group was 35.50±20.00,whereas that in the stable group was 17.00 ± 8.00(P<0.05),which were significantly different between the two groups(P<0.05).In the activity group,the ADC values of right and left iliac/sacral of the sacroiliac joint were(8.81 ±4.64)/(6.24±2.57)× 10-4mm2/s,(7.36±4.60)/(6.13±2.81)× 10-4mm2/s,respectively.In the stable group,the ADC values of right and left iliac/sacral of the sacroiliac joint were(5.33±2.26)/(4.61±2.70)× 10-4mm2/s,(5.16±1.11)/(4.46±1.71)× 10-4 mm2/s,respectively.The ADC values of iliac and sacral sides in left and right sacroiliac joints in the active group were higher than those in the stable group(P<0.05).Besides,each group showed higher ADC values of the iliac side compared to those of sacral side(P<0.05).3.Comparison of ADC values of bilateral sacroiliac jointsADC values were compared between ilium and ilium,ilium and sacrum of bilateral sacroiliac joint of active group and the stable group,respectively.The results show that there was no significant difference in ADC values between the left and right the ilium and sacrum(P>0.05),but the ADC values were statistically significant left and right between ilium and sacrum(P<0.05).Therefore,in the following statistical analysis,the ilium ADC values is the average of the left and right ilium ADC values of sacroiliac joint,the sacrum ADC value is the average of the left and right sacrum ADC values of sacroiliac joint.4.Correlation analysis between MRI parameters and clinical activity indexesThere was a positive correlation between SPARCC score and clinical activity index(P<0.05)with the activity group higher than the stable group.The correlation between SPARCC score and ASDAS-CRP was the highest in the active group and stable group(r=0.806/0.621,P<0.05).The ADC value was positively correlated with clinical activity indexes(P<0.05).In the active group,the correlation between ADC value and ASDAS-CRP was the moderate correlation,which the iliac/sacral correlation coefficient was 0.687/0.608(P<0.05).In the stable group,the correlation between ADC value and ASDAS-CRP was low or moderate correlation,which the iliac/sacral correlation coefficient was 0.364/0.575(P<0.05).However,there was no significant correlation between ADC value and the value of ESR,CRP,ASDAS-ESR,BASDAI(P>0.05).5.ROC curve analysisASDAS-CRP was the most effective index in clinical activity indicators.The optimal cut-off value was 2.33 to distinguish active group from stable group in bone marrow edema of sacroiliac arthritis,in which area under the curve,sensitivity and specificity were 0.933,78.6%and 100%,respectively.SPARCC score was shown to be the most effective among the MRI quantitative parameters.The optimal cut-off value was 19.5 score to distinguish active group from stable group in bone marrow edema of sacroiliac arthritis,in which area under the curve,sensitivity and specificity were 0.887,89.3%,73.3%,respectively.Conclusion:1.MRI quantitative parameters(SPARCC score and ADC value)and clinical activity indexes(ESR,CRP,ASDAS-ESR,ASDAS-CRP,BASDAI)play an important role in differential diagnosis of the activity of bone marrow edema of sacroiliac joint in early AS.2.SPARCC score has the highest efficacy and sensitivity,but lower specificity in MRI quantitative parameters.ASDAS-CRP has the strongest efficiency and specificity,but the low sensitivity in the clinical evaluation indexes.3.There is the positive correlation between MRI quantitative parameters and clinical activity evaluation indexes,and the correlation of activity group was higher than that of stable group,which SPARCC score has the highest correlation with ASDAS-CRP.Part II:Study of Radiomics Texture Analysis in evaluating the Activity of Bone Marrow Edema of Sacroiliac Joint in Early Ankylosing Spondylitis based on STIR SequenceObjective:To explore the value of texture analysis of imageomics based on STIR sequence of magnetic resonance to evaluate the activity of bone marrow edema of sacroiliac joint in early ankylosing spondylitis.Materials and methods:A total of 43 patients with early AS were collected in our hospital,which included 30 males and 13 females.All patients were divided into active group(28 cases)and stable group(15cases)according to BASDAI activity evaluation criteria.The STIR images of all patients were imported into ITK-snap software in DICOM format,and the ROI of the lesions was outlined.We outlined the iliac and sacral sides of the left and right sacroiliac joints separately,including four ROI areas(right ilium,right sacrum,left ilium and left sacrum),and 172 ROI regions were outlined in 43 patients.The original STIR and ROI images were imported into software of GE.A.K.and texture feature parameters are automatically extracted by software.According to the ratio of 7:3,all the patients were randomly divided into training set(n=116)and verification set(n=56).The training set was mainly used for texture feature selection and model building,and the verification set was mainly used for model effectiveness evaluation.Optimal feature subset were obtained by Mann-Whitney U test,the minimum-Redundancy Maximum-Relevancy(mRMR),and then the least absolute shrinkage and selection operator(LASSO)from these texture feature parameters.The optimal feature subset was used to construct the final prediction model,and Radscore by summing the selected features weighted by their coefficients.Statistical methods:The count data were analyzed by using ?2 test.All Measurement data were expressed in median±quartile interval and detected by Mann-Whitney U test.The effectiveness of Radscore value,SPARCCscore and ADC value was evaluated by ROC curve.The difference of area under ROC curve of three methods was compared by Delong's test.The correlation between Radscore values and the values of SPARCC,ADC,BASDAI,ESR,CRP,ASDAS-ESR ASDAS-CRP was analyzed by Spearman correlation test.Results:1.Radiomics texture feature extraction and Iomics modeling based on STIR sequenceA total of 895 radiomics features were extracted and 8 texture feature subsets were selected by Mann Whitney U test,mRMR,LASSO regression model,which the texture features include:three gray level co-occurrence matrix features(correlation,maximum correlation coefficient,cluster shadow),two gray level dependence matrix features(large dependence high gray level focus,small dependence low gray level focus),one gray level run distance matrix feature(gray level nonuniformity),one gray level size area matrix feature(large range high gray level area focus),and one neighboring gray tone difference matrix features(strength).Radscore by summing the selected features weighted by their coefficients.2.ROC curve analysis of Radscore,SPARCC score and ADC valueIn the training set,the optimal cut-off values(AUC,accuracy,sensitivity,specificity,positive predictive value,negative predictive value)for differentiating early as sacroiliac joint bone marrow edema between active group and stable group were Radscore=51.46(0.81,76.72%,81.25%,66.67%,84.42%,61.5%),SPARCC=20 scores(0.91,89.66%,90.00%,88.89%,94.97%,80.00%),ADC=6.71×10-4mm2/s(0.78,62.93%,48.75%,94.44%,95.12%,45.33%).In the validation set,the optimal cut-off values(AUC,accuracy,sensitivity,specificity,positive predictive value,negative predictive value)for differentiating early as sacroiliac joint bone marrow edema between active group and stable group wereRadscore=51.46(0.87,82.14%,90.65%,70.83%,80.56%,85.00%),SPARCC=20sco re(0.89,78.57%,87.50%,66.67%,77.78%,80.00%),ADC=6.71 × 10-4mms(0.85,71.42%,53.13%,95.83%,94.44%,60.53%).In the training set and validation set,there was no significant difference in AUC between Radscore and SPARCC score,ADC value(P>0.05).There was significant difference in AUC the training between SPARCC score and ADC value(P<0.05),but there was no statistical significance in the validation set(P>0.05),3.Correlation of between Radscore and MRI quantitative parameters,clinical activity indexesThere were low correlations between Radscore values and SPARCC scores,ADC values,and the correlation coefficients were both 0.29(P<0.05).There were all low correlation between Radscore values and the values of the ESR,CRP,BASDAI,ASDAS-ESR,ASDAS-CRP,which correlation coefficients were 0.28,0.21,0.37,0.28 and 0.36,respectively(P<0.05).Conclusion:1.The radiomics prediction model can predict the activity of sacroiliac joint bone marrow edema in the training set and validation set,which meaningful texture features include:large dependence high gray level(LDHGL),small dependence low gray level(SDLGL),large zone high gray-Level emphasis(LZHGE),gray-level non-uniformity(GLN),neighboring gray tone difference matrix based features(Strength),gray-level co-occurrence matrix based features(correlation,maximal correlation coefficient,cluster shade).2.ROC curve efficiency:SPARCC score>Radscore>ADC value,and there was no statistical difference between the three evaluation methods in the validation set.3.There was the low correlation between Radscore and the values of SPARCC,ADC.There was the low correlation between Radscore and the values of BASDAI,ESR,CRP,ASDAS-ESR,ASDAS-CRP.