Radiography, abdominal CT and MRI compared with sacroiliac joint CT in diagnosis of structural sacroiliitis

Deep learning methods in the diagnosis of sacroiliitis from plain pelvic radiographs

OBJECTIVES

The aim of this study is to develop a computer-aided diagnosis method to assist physicians in evaluating sacroiliac radiographs.

METHODS

Convolutional neural networks, a deep learning method, were used in this retrospective study. Transfer learning was implemented with pre-trained VGG-16, ResNet-101 and Inception-v3 networks. Normal pelvic radiographs (n = 290) and pelvic radiographs with sacroiliitis (n = 295) were used for the training of networks.

RESULTS

The training results were evaluated with the criteria of accuracy, sensitivity, specificity and precision calculated from the confusion matrix and AUC (area under the ROC curve) calculated from ROC (receiver operating characteristic) curve. Pre-trained VGG-16 model revealed accuracy, sensitivity, specificity, precision and AUC figures of 89.9%, 90.9%, 88.9%, 88.9% and 0.96 with test images, respectively. These results were 84.3%, 91.9%, 78.8%, 75.6 and 0.92 with pre-trained ResNet-101, and 82.0%, 79.6%, 85.0%, 86.7% and 0.90 with pre-trained inception-v3, respectively.

CONCLUSIONS

Successful results were obtained with all three models in this study where transfer learning was applied with pre-trained VGG-16, ResNet-101 and Inception-v3 networks. This method can assist clinicians in the diagnosis of sacroiliitis, provide them with a second objective interpretation and also reduce the need for advanced imaging methods such as magnetic resonance imaging.

Can Bone Erosion in Axial Spondyloarthropathy be Detected by Ultrashort Echo Time Imaging? A Comparison With Computed Tomography in the Sacroiliac Joint

BACKGROUND

Structural lesion evaluation in axial spondyloarthropathy (SpA) can improve accuracy of diagnosis. However, structural lesions (bone erosions) are difficult to be assessed using conventional MRI compared to computed tomography (CT).

PURPOSE

To evaluate the diagnostic performance of ultrashort echo time (UTE) for detecting bone erosion in axial SpA compared to T1WI and three-dimensional double-echo steady-state (3D DESS) imaging using CT as the reference standard.

STUDY TYPE

Retrospective.

POPULATION

Fourteen patients (eight females, 57.1%) and 14 healthy controls (seven females, 50.0%) who underwent sacroiliac (SI) joint MRI and CT.

FIELD STRENGTH/SEQUENCE

3 T; TSE T1WI, 3D DESS, 2D UTE.

ASSESSMENT

The bilateral SI joints were assessed for bone erosion. Three observers scored bone erosion for all three sequences of MRI. CT was used as the gold standard. Diagnostic confidence in axial SpA was measured based on a four-point confidence score.

STATISTICAL TESTS

Correlation of erosion scores between CT and MRI were evaluated using Spearman's correlation test. Sensitivity, specificity, and positive-negative predictive values were calculated. Confidence scores were compared using the Wilcoxon sum rank test. Statistical significance was set at P < 0.05.

RESULTS

Compared with erosion scores of CT, the correlation coefficients for each MRI sequence showed significant low-to-high positive correlations (0.39-0.72). UTE imaging showed the highest correlation coefficients for all observers (0.70, 0.72, and 0.67, respectively). The specificity of UTE imaging was equal or higher than those of T1WI and 3D DESS for all observers (0.86 vs. 0.71 vs. 0.57; 0.93 vs. 0.71 vs. 0.57; 0.79 vs. 0.79 vs. 0.43). All observers had the highest confidence in interpreting UTE imaging for detecting bone erosion among the three sequences (3.5, 3.4, and 3.3 for UTE; 3.1, 3.0, and 2.6 for T1WI; and 3.2, 2.7, and 2.4 for DESS).

DATA CONCLUSION

UTE imaging can detect bone erosions in patients with axial SpA and show higher specificity than conventional T1WI and 3D DESS.

EVIDENCE LEVEL

4 TECHNICAL EFFICACY: Stage 2.

Addition of CT to Improve the Diagnostic Confidence for the Detection of Sacroiliac Joint Erosions in Patients with Equivocal MRI Findings

PURPOSE

To determine if CT can improve the diagnostic confidence for the detection of sacroiliac joint (SIJ) erosions in patients with equivocal MRI findings.

METHODS

A retrospective analysis of adult patients who had an SIJ MRI and a subsequent SIJ CT within 12 months was conducted. Using a 5-point Likert scale, two reviewers evaluated the de-identified MRI and CT images in randomized order and in separate sessions to answer the question: "Does the patient have SIJ erosions?". A Fisher's exact test was used to analyze the difference in diagnostic confidence, and intraclass correlation coefficient (ICC) was used to determine interrater reliability.

RESULTS

54 patients were included in the analysis (average age, 43.9 years). The average time interval between initial SIJ MRI and subsequent CT was 14.4 weeks (range, 5.6-50.3 weeks). CT resulted in significantly more cases with definitive diagnostic confidence than cases with probable or equivocal confidence compared to MRI (P < .001). Amongst cases with equivocal findings on MRI, 73.2% of cases had definitive diagnoses on CT. There was moderate interrater agreement for MRI, with an ICC of .490 [95% CI, .258-.669], and excellent agreement for CT, with an ICC of .832 [95% CI, .728-.899].

CONCLUSION

Overall, CT led to significantly increased diagnostic confidence and higher interrater reliability for the detection of SIJ erosions compared to MRI. Judicious use of CT may be useful in detecting SIJ erosions in patients with equivocal MRI findings.

Sacroiliac joint in sarcoidosis on computed tomography: a monocentric retrospective study (SISTER)

Sacroiliitis and spondyloarthritis (SpA) have been associated to sarcoidosis. Sarcoidosis bone involvement of the sacral or iliac bones has been reported to mimic SpA. We aimed to evaluate the prevalence of structural sacroiliitis and structural changes of the sacroiliac joints (SIJ) in patients with sarcoidosis by abdominal-pelvic computed tomography (AP-CT). In this monocentric retrospective study, three blinded readers evaluated AP-CT that had already been performed on patients with sarcoidosis and classified them as normal, degenerative, or inflammatory. A consensus was reached for the divergent cases. Erosion, ankylosis, and sclerosis, classically associated with sacroiliitis, were noted. SpA was defined according to the ASAS 2009 classification criteria. We identified 217 patients with proven sarcoidosis who underwent AP-CT. Only three patients had sacroiliitis by CT and four had SpA, representing 1.38% and 1.85% of the patients, respectively. Degenerative SIJs represented 28.1% of patients and were significantly associated with age, at least one pregnancy, rural lifestyle, ankylosis, diffuse idiopathic skeletal hyperostosis, sclerosis, and the presence of osteophytes. Four patients had axial bone sarcoidosis. Sacroiliitis, SpA, and degenerative changes of the SIJ have been highlighted by AP-CT in patients with sarcoidosis. Osteoarthritis of the SIJ in sarcoidosis was associated with age, pregnancy, and rural lifestyle. Further studies are needed to assess the link between SpA and sarcoidosis.

Disease progression associated with low bone mass in axial spondyloarthropathy patients

Through statistical analysis, we have found that inflammation and low femoral and lumbar spine BMD were strongly correlated with a high SIJ CT grade, and inflammation, low vitamin D levels, and a longer disease course within a certain time range influenced bone loss in axSpA.

PURPOSE

We investigated the relationship between bone mineral density (BMD), vitamin D, and computed tomography (CT)-based progression of disease grades of the sacroiliac joint (SIJ), and sought to identify parameters predicting low BMD in patients with axial spondyloarthropathy (axSpA).

METHODS

We collected the ankylosis spondylitis disease activity score (ASDAS), the course of the disease, HLA-B27 status, and vitamin D and C-reactive protein (CRP) levels of 98 axSpA patients. Lumbar spine and femoral BMD were assessed by dual-energy X-ray (DXA), and SIJ grade was determined by CT.

RESULTS

The axSpA patients (71 men, 27 women) with a mean age of 31.9 years (range 18-57 years) and body mass index 21.8 kg/m² (range 15.6-30.6 kg/m²), with disease duration 4.5 years (range 0.3-30 years) were included. A longer disease course, higher CRP level, and lower femoral and lumbar spine BMD were independently related to a higher CT grade. Older age, longer disease course, elevated CRP, and high SIJ CT grade were independently related to lower BMD (femur and/or lumbar spine L1-L4 T scores ≤ -1). Older age, elevated CRP, low vitamin D levels, and high CT grade were independently associated with low femur and lumbar spine BMD. However, a longer disease course was independently related to low femur BMD, but not low lumbar spine BMD.

CONCLUSIONS

Thus, inflammation and low femoral and lumbar spine BMD were strongly correlated with a high SIJ CT grade, and inflammation, low vitamin D levels, and a longer disease course within a certain time range influenced bone loss in axSpA.

Improved detection of subchondral erosions in the sacroiliac joints with T1-weighted fat-suppressed MRI

OBJECTIVE

To compare the diagnostic performance of two different sets of magnetic resonance imaging (MRI) for the detection of subchondral erosions in the sacroiliac joints regarding the application of fat-water separation techniques when acquiring T1-weighted (T1w) images, using multi-detector computed tomography (MDCT) as the reference standard.

METHODS

We retrospectively included 31 consecutive patients having or being suspected for axial spondyloarthritis (SpA) assessed using both MRI and MDCT. Three sets of images were independently assessed for the presence of erosions by two musculoskeletal radiologists (R1, R2): (1) MRI with standard T1w without fat suppression, (2) MRI with both T1w with and without fat suppression, and (3) MDCT. The diagnostic performance of both sets of MRIs was assessed using MDCT as the referent.

RESULTS

The assessment of T1w images with fat suppression substantially increased sensitivity (76% vs. 63% R1; 70% vs. 60% R2), specificity (97% vs. 84% R1; 96% vs. 81% R2), positive predictive value (85% vs. 45% R1; 81% vs. 40% R2), and overall accuracy (94% vs. 80% R1; 92% vs. 77% R2) in the detection of erosions when compared to the assessment using T1w images without fat suppression.

CONCLUSION

The assessment of T1w images with fat suppression substantially improves the diagnostic performance of MRI in the detection of erosions in the sacroiliac joints.

KEY POINTS

• The presence of erosions in the sacroiliac joints may influence the decision on the diagnosis of axial spondyloarthritis. • T1w fat-suppressed MR imaging relatively increases the contrast between the joint space (high signal) and the adjacent subchondral bone (low signal), potentially improving the detection of erosions in the sacroiliac joints. • T1w fat-suppressed images improve the diagnostic performance of MRI in the detection of erosions in the sacroiliac joints compared to T1w without fat suppression, using MDCT as the reference.

Update of imaging in the diagnosis and management of axial spondyloarthritis

Imaging of the spine and sacroiliac joints has acquired a central role in the diagnosis and classification of axial spondyloarthritis (axSpA) in the earliest phases of the disease. New definitions of specific imaging lesions, particularly in magnetic resonance imaging (MRI), have been recently updated and revised by the ASAS MRI working group to reach a standardized understanding and diagnosis of axSpA among rheumatologists. Recognizing the misleading pitfalls of MRI lesions and differential diagnosis also represents an essential issue in clinical practice to avoid false-positive findings and establish the diagnosis of axSpA with careful regard to the clinical context, clinical signs, and biological tests. This review summarizes the current evidence on the different imaging modalities of the sacroiliac joints and the spine with their application in the clinical setting of SpA and their main pitfalls; it also highlights the newest emerging imaging techniques.

Low Back Pain and Sacroiliitis on Cross-Sectional Abdominal Imaging for Axial Spondyloarthritis Diagnosis in Inflammatory Bowel Diseases

Background

Joint damage is the most frequent extraintestinal manifestation in inflammatory bowel disease (IBD).

Aims

The aim of the study was to assess the value of low back pain (LBP) associated with sacroiliitis on abdominal imaging for the diagnosis of spondyloarthritis (SpA) in IBD.

Methods

We used a questionnaire assessing rheumatological symptoms for all patients with abdominal computed tomography (CT) and magnetic resonance enterography (MRE). Sacroiliitis was assessed on available CT and MRE. Patients were classified as axial SpA according to the Assessment of SpondyloArthritis International Society criteria.

Results

Fifty-one patients completed the questionnaire and performed both exams. LBP was present in 27 patients (52.9%), and 10 (19.6%) had an inflammatory component. Sacroiliitis was reported in 12 patients (23.5%), and 6 of them suffered from LBP. Among the 20 patients referred to the rheumatologist, 11 patients suffered from LBP. One patient was HLA-B27 positive and presented sacroiliitis. For the last 10 patients, none of them had a sacroiliitis, and 2 patients were negative for HLA-B27.

Conclusion

An axial SpA has been diagnosed in 11.8% of IBD patients undergoing cross-sectional imaging, whereas one-fifth had inflammatory LBP, and sacroiliitis was observed in one-quarter of them. To optimize the diagnosis of axial SpA, HLA-B27 testing might be required for patients with both IBD and LBP, but this will require further investigation before its implementation in routine practice.

Comparison between radiography and magnetic resonance imaging for the detection of sacroiliitis in the initial diagnosis of axial spondyloarthritis: a cost-effectiveness study

OBJECTIVE

The purpose of our study was to determine the cost-effectiveness of radiography and MRI-based imaging strategies for the initial diagnosis of sacroiliitis in a hypothetical population with suspected axial spondyloarthritis.

MATERIALS AND METHODS

A decision analytic model from the health care system perspective for patients with inflammatory back pain suggestive of axial spondyloarthritis was used to evaluate the incremental cost-effectiveness of 3 imaging strategies for the sacroiliac joints over a 3-year horizon: radiography, MRI, and radiography followed by MRI. Comprehensive literature search and expert opinion provided input data on cost, probability, and utility estimates. The primary effectiveness outcome was quality-adjusted life-years (QALYs), with a willingness-to-pay threshold set to $100,000/QALY gained (2018 American dollars).

RESULTS

Radiography was the least costly strategy ($46,220). Radiography followed by MRI was the most effective strategy over a 3-year course (2.64 QALYs). Radiography was the most cost-effective strategy. MRI-based and radiography followed by MRI-based strategies were not found to be cost-effective imaging options for this patient population. Radiography remained the most cost-effective strategy over all willingness-to-pay thresholds up to $100,000.

CONCLUSION

Radiography is the most cost-effective imaging strategy for the initial diagnosis of sacroiliitis in patients with inflammatory back pain suspicious for axial spondyloarthritis.

Clinical practice recommendations for the use of imaging in the diagnosis and management of axial spondyloarthritis in Taiwan

OBJECTIVE

Axial spondyloarthritis (axSpA) is a chronic inflammatory disease bearing challenges in early diagnosis. To improve clinical diagnosis and management of axSpA, recommendations were developed with current axSpA classification criteria and recent advances in medical imaging applications.

METHODS

A systematic literature review was conducted by 10 rheumatologists and radiologists in Taiwan to retrieve research evidence on the utilization of imaging modalities, including conventional radiography (CR), magnetic resonance imaging (MRI), computed tomography (CT), ultrasound (US), quantitative sacroiliac scintigraphy (QSS), and dual-energy X-ray absorptiometry (DXA). The panel of experts proposed six key issues on the role of imaging in early diagnosis of axSpA, monitoring of disease activity and structural changes, predicting treatment effects, and assessing complications such as osteoporosis and spinal fracture. The consensus was established on the basis of research evidence, clinical experiences and expert opinions. For each recommendation statement, the level of evidence was evaluated, the strength of recommendation was graded and the final level of agreement was determined through voting.

RESULTS

In total, four overarching principles and 13 recommendations were formulated. These recommendations outlined different imaging approaches in the diagnosis and management of axSpA disease progression. Considering CT is easy to perform when MRI is less available in Taiwan, the expert panel proposed a concise and practical diagnostic scheme to strengthen the valuable role of MRI and CT in the diagnostic evaluation of axSpA without evident radiographic features.

CONCLUSION

These modified recommendations provide guidance for rheumatologists, radiologists and healthcare professionals on timely diagnosis of axSpA and disease management with appropriate imaging modalities.

Radiographic Prevalence of Sacroiliac Joint Abnormalities and Clinical Outcomes in Patients With Femoroacetabular Impingement Syndrome

PURPOSE

To quantify the prevalence of sacroiliac joint (SIJ) abnormalities in patients undergoing hip arthroscopy for femoroacetabular impingement syndrome (FAIS) by use of various imaging modalities and to compare outcomes based on SIJ abnormalities.

METHODS

Plain radiographs, computed tomography (CT) scans, and magnetic resonance imaging (MRI) scans of patients who underwent primary hip arthroscopy for FAIS from January 2012 to January 2016 were identified. The exclusion criteria included patients undergoing bilateral or revision surgery, those with a history of dysplasia, and those with less than 2 years' follow-up. On radiographs, the SIJs were graded using modified New York criteria for spondyloarthropathy. CT and MRI scans were reviewed for joint surface erosion, subchondral sclerosis, joint space narrowing, pseudo-widening, bone marrow edema, and ankylosis. Patients with SIJ abnormalities were matched to patients without SIJ abnormalities in a 1:2 ratio by age and body mass index. Outcomes included the Hip Outcome Score-Activities of Daily Living (HOS-ADL), Hip Outcome Score-Sports Subscale (HOS-SS), modified Harris Hip Score (mHHS), visual analog scale (VAS) for pain, and VAS for satisfaction.

RESULTS

Of 1,009 consecutive patients, 743 (73.6%) were included; 187 (25.2%) showed SIJ changes. Of these 187 patients, 164 (87.7%) had changes on plain radiographs, 88 (47.1%) had changes on CT, and 125 (66.8%) had changes on MRI. SIJ changes on any imaging modality were weakly correlated with pain to palpation of the SIJ (r = 0.11, P = .004) on physical examination. Pain to palpation of the SIJ on physical examination (odds ratio [OR], 1.12; P = .031) and a history of SIJ pain (OR, 1.93; P = .018) increased the odds of having an SIJ abnormality on any imaging modality. After matching, patients without SIJ abnormalities had a significantly greater HOS-ADL (95.4 vs 90.6, P = .001), HOS-SS (91.1 vs 77.5, P < .001), and mHHS (91.3 vs 84.5, P < .001) and a significantly lower VAS pain score (10.9 vs 25.7, P < .001) than patients with abnormalities at a mean follow-up of 34.1 ± 9.7 months (range, 24-54 months). Patients without SIJ abnormalities had greater odds of achieving the minimal clinically important difference for the HOS-ADL (OR, 2.91; P = .001) and for the HOS-SS (OR, 2.83; P < .001) but not for the mHHS (OR, 1.73; P = .081).

CONCLUSIONS

A high prevalence of SIJ abnormalities (25.2%) is seen on imaging in FAIS patients. These patients may show significantly inferior clinical outcomes and persistent postoperative pain after FAIS treatment. The results of this study may allow treating orthopaedic surgeons to better inform patients with SIJ abnormalities that they may not achieve clinically significant outcome improvement after hip arthroscopy.

LEVEL OF EVIDENCE

Level III, retrospective comparative study.

CT and gross pathology are comparable methods for detecting some degenerative sacroiliac joint lesions in dogs

Degenerative sacroiliac joint disease is a cause of lumbosacral pain in dogs; however, published information on cross-sectional imaging characteristics is limited. Objectives of this retrospective, secondary analysis, methods-comparison study were to test hypotheses that CT lesions reported in humans with degenerative sacroiliac joint disease are also present in dogs, and that CT is comparable to gross pathology for detecting these lesions. Matched CT and gross pathology slice images of 30 sacroiliac joints were retrieved from a previous prospective, canine cadaver study. A veterinary radiologist interpreted randomized CT images for each joint based on previously published CT characteristics of lesions in humans with degenerative sacroiliac joint disease. A veterinary pathologist independently interpreted randomized gross pathology images using the same criteria. All joints had at least one CT lesion consistent with degenerative sacroiliac joint disease. A new CT lesion was also identified and termed "subarticular cleft." The CT and gross pathology methods agreed for detecting joints with subchondral sclerosis, subchondral erosion, and intra-articular ankylosis lesions (P > .05, McNemar's test), but disagreed for detection of joints with subchondral cyst, para-articular ankylosis, and subarticular cleft lesions (P ≤ .05). Using gross pathology as the reference standard, CT had 100% sensitivity for detection of subarticular cleft and subchondral cyst lesions, with 56% and 22% specificity, respectively. Para-articular ankylosis lesions were detected by CT but not by gross pathology. Findings supported the hypothesis that CT lesions reported in humans with degenerative sacroiliac joint disease are also present in dogs, and partially supported the hypothesis that CT is comparable to gross pathology for detecting joints with these lesions.

Computed tomography in axial spondyloarthritis

PURPOSE OF REVIEW

Computed tomography (CT) is increasingly being used in ankylosing spondylitis (AS) for imaging the spine and sacroiliac joint (SIJ). We review new insights to diagnosis and evaluation revealed by the use of CT.

RECENT FINDINGS

Studies using low-dose CT in AS to detect syndesmophytes can image the entire spine, but semiquantitative scoring of the scans by human readers decreases the reliability and validity of this method. The thoracic spine is the segment most involved with syndesmophytes. Syndesmophytes are not randomly distributed around the vertebral rim but have preferred locations, which vary with the vertebral level and may be related to biomechanics. Examination of SIJ on abdominal CT scans has found structural changes of sacroiliitis in up to 35% of patients with inflammatory bowel disease. The significance of monosodium urate crystal deposition in the pelvis of axial spondyloarthritis patients without coexisting gout is uncertain.

SUMMARY

Low-dose CT is a promising tool in AS. Studies of biomarkers or medications and their relations with syndesmophyte progression should take the thoracic spine into account. Abdominal CT scans are useful for detecting changes related to sacroiliitis.