Objective
We aimed to develop and validate a fully automated machine learning (ML) algorithm that predicts bone marrow edema (BME) on a quadrant level in sacroiliac (SI) joint magnetic resonance ...imaging (MRI).
Methods
A computer vision workflow automatically locates the SI joints, segments regions of interest (ilium and sacrum), performs objective quadrant extraction, and predicts presence of BME, suggestive of inflammatory lesions, on a quadrant level in semicoronal slices of T1/T2‐weighted MRI scans. Ground truth was determined by consensus among human readers. The inflammation classifier was trained using a ResNet18 backbone and five‐fold cross‐validated on scans of patients with spondyloarthritis (SpA) (n = 279), postpartum individuals (n = 71), and healthy subjects (n = 114). Independent SpA patient MRI scans (n = 243) served as test data set. Patient‐level predictions were derived from aggregating quadrant‐level predictions, ie, at least one positive quadrant.
Results
The algorithm automatically detects the SI joints with a precision of 98.4% and segments ilium/sacrum with an intersection over union of 85.6% and 67.9%, respectively. The inflammation classifier performed well in cross‐validation: area under the curve (AUC) 94.5%, balanced accuracy (B‐ACC) 80.5%, and F1 score 64.1%. In the test data set, AUC was 88.2%, B‐ACC 72.1%, and F1 score 50.8%. On a patient level, the model achieved a B‐ACC of 81.6% and 81.4% in the cross‐validation and test data set, respectively.
Conclusion
We propose a fully automated ML pipeline that enables objective and standardized evaluation of BME along the SI joints on MRI. This method has the potential to screen large numbers of patients with (suspected) SpA and is a step closer towards artificial intelligence–assisted diagnosis and follow‐up.
Abstract
Background/Aims
The combined use of Power Doppler and B-mode ultrasound (PDUS) allows visualisation of morphological and pathophysiological changes of the synovium. ULTIMATE (NCT02662985) ...was the first large, randomised, double-blind, placebo-controlled PDUS phase IIIb study in psoriatic arthritis (PsA) to demonstrate that the validated Global OMERACT EULAR Synovitis Score (GLOESS), an ultrasound score at patient level, could detect the early and continuous decrease in synovitis in a multicentre setting using different ultrasound devices and examiners. Since the ultrasound assessment for GLOESS is time-consuming owing to the number of joints assessed, we investigated the ability of various reduced joint sets to predict GLOESS.
Methods
ULTIMATE was a 52-week study of secukinumab with a 12-week, double-blind, placebo-controlled period followed by a 12-week, open-label treatment period and 6-month open-label extension period. GLOESS for 24 paired joints was calculated, with a resultant potential score ranging from 0 to 144. Based on their composite PDUS scores, highly correlated joint clusters were identified with a Spearman’s rank correlation matrix and a Clustered Image Map. Representative joints were then selected from each group by various approaches (best correlation, model optimisation, etc.) yielding several joint combinations. For these reduced joint sets, GLOESS was predicted with linear models based on data from 60% of randomly selected patients from ULTIMATE. The remaining 40% of patient data were used for model validation and diagnostics.
Results
Five models were established with reduced pairs of joint sets (9-13 pairs) for PDUS-detected synovitis. The joints included in each linear model are summarised in the Table. All five reduced joint set models demonstrated scores very close to full 48 joint GLOESS (R2 ∼0.95); models 1-3 (based on 9 joint pairs) were less accurate at predicting GLOESS than models 4 and 5 (based on 13 and 12 joint pairs, respectively) for the secukinumab cohort in the open-label part (data to be shown in poster).
Conclusion
All models of reduced joint sets for PDUS-detected synovitis predicted GLOESS well. The next steps will be to document responsiveness and ability to discriminate between active and placebo treatment.
Disclosure
P.G. Conaghan: Consultancies; P.G.C. has received consultancy fees from AbbVie, AstraZeneca, BMS, Eli Lilly, Galapagos, Janssen, Merck, Novartis and Pfizer, Stryker and UCB. Member of speakers’ bureau; P.G.C. has received speakers bureau fees from AbbVie, Novartis and Pfizer. M. D'agostino: Consultancies; M.A.D'A. has received consultancy fees from AbbVie, BMS, Celgene, Eli Lilly, Janssen, Novartis, Roche, Sanofi, and UCB. Member of speakers’ bureau; M.A.D'A. has received speakers bureau fees from AbbVie, BMS, Celgene, Eli Lilly, Janssen, Novartis, Roche, Sanofi and UCB. M. Boers: Consultancies; M.B. has received consultancy fees from Novartis. E. Naredo: Honoraria; E.N. has received honoraria for clinical trials from AbbVie, BMS and Novartis. Member of speakers’ bureau; E.N. has received speakers bureau fees from AbbVie, BMS, Celgene GmbH, Janssen, Lilly, Novartis, Pfizer, Roche and UCB. Grants/research support; E.N. has received research grants from Eli Lilly. P. Mandl: Member of speakers’ bureau; P.M. has received speakers bureau fees from AbbVie, BMS, Celgene, Janssen, Lilly, MSD, Novartis, Roche and UCB. Grants/research support; P.M. has received grant/research support from AbbVie, BMS, Celgene, Janssen, Lilly, MSD, Novartis, Roche and UCB. P. Carron: Consultancies; P.C. has received consultancy fees from AbbVie, Bristol Myers Squibb, Celgene Corporation, Eli Lilly, Gilead, Merck Sharp Dohme, Novartis, Pfizer and UCB. Member of speakers’ bureau; P.C. has received speakers bureau fees from AbbVie, Bristol Myers Squibb, Celgene Corporation, Eli Lilly, Gilead, Merck Sharp Dohme, Novartis, Pfizer and UCB. Grants/research support; P.C. has received grant/research support from Merck Sharp Dohme, Pfizer and UCB. M. Backhaus: Consultancies; M.B. has received consultancy fees from BMS, Galapagos, Jonsson, MSD, Novartis, Pfizer, Roche and UCB. Member of speakers’ bureau; M.B. has received speakers bureau fees from BMS, Galapagos, Jonsson, MSD, Novartis, Pfizer, Roche and UCB. Grants/research support; M.B. has received grant/research support from BMS, Galapagos, Jonsson, MSD, Novartis, Pfizer, Roche and UCB. A. López-Rodríguez: Consultancies; A.L.R. has received consultancy fees from Eli Lilly, GSK, Janssen, Novartis, Roche and UCB. Member of speakers’ bureau; A.L.R. has received speakers bureau fees from Eli Lilly, GSK, Janssen, Novartis, Roche and UCB. P. Hanova: Consultancies; P.H. has received consultancy fees from Novartis. P. Goyanka: Other; P.G. is an employee of Novartis. B.G. Sahoo: Other; B.G.S. is an employee of Novartis. C. Gaillez: Shareholder/stock ownership; C.G. is a shareholder of Novartis and BMS. Other; C.G. is an employee of Novartis. W. Bao: Shareholder/stock ownership; W.B. is a shareholder of Novartis. Other; W.B. is an employee of Novartis.
The Assessment of SpondyloArthritis International Society (ASAS) aimed to develop a set of quality standards (QS) to help improve the quality of healthcare provided to adult patients affected by ...axial spondyloarthritis (axSpA) worldwide.
An ASAS task force developed a set of QS using a stepwise approach. First, key areas for quality improvement were identified, discussed, rated and agreed on. Thereafter, areas were prioritised and statements for the most important key areas were phrased on consensus. Appropriate quality measures were defined to allow quantification of the QS at the community level.
The ASAS task force, consisting of 20 rheumatologists, two physiotherapists and two patients, selected and proposed 34 potential key areas for quality improvement which were then commented by 140 ASAS members and patients. Within that process three new key areas came up, which led to a re-evaluation of all 37 key areas by 120 ASAS members and patients. Five key areas were identified as most important to determine quality of care: referral including rapid access, rheumatology assessment, treatment, education/self-management and comorbidities. Finally, nine QS were agreed on and endorsed by the whole ASAS membership.
ASAS successfully developed the first set of QS to help improving healthcare for adult patients with axSpA. Even though it may currently not be realistic to achieve the QS in all healthcare systems, they provide high-quality of care framework for patients with axSpA that should be aimed for.
Objective
This study was undertaken to assess the inflammatory burden in peripheral spondyloarthritis (SpA) by magnetic resonance imaging (MRI) of the legs in an early remission–induction strategy ...study of tumor necrosis factor (TNF) blockade. Furthermore, we sought to determine the value of MRI to predict disease relapse versus sustained remission after treatment discontinuation.
Methods
Thirty‐two patients with early peripheral SpA with involvement of the legs determined on clinical examination and confirmed by ultrasonography (US) participated in a remission‐induction trial of a TNF inhibitor (TNFi). Patients underwent MRI of the joints and entheses of the legs at baseline and at clinical remission, after which TNFi treatment was withdrawn. Images were evaluated for joint effusion, joint osteitis, entheseal soft tissue inflammation, and entheseal osteitis.
Results
Joint effusion and enthesitis on clinical examination and US correlated well with MRI abnormalities. In addition, a substantial amount of subclinical involvement was seen on MRI, mainly in the ankle joints and heel entheses. Inflammation scores were markedly lower in the subclinically involved joints and entheses versus those that were clinically involved (P values ranged from 0.01 to <0.001). Inflammatory load on MRI decreased significantly upon TNFi treatment (P < 0.001). Whereas 80% of the joints that were clinically involved at baseline showed no effusion on remission MRI, 2 of 3 entheses involved at baseline showed residual inflammation. In addition, patients who experienced a relapse after treatment discontinuation displayed more entheseal soft tissue inflammation on remission MRI compared to those who maintained drug‐free remission (P = 0.028).
Conclusion
Our findings delineate a differential response of synovitis and enthesitis, with enthesitis on MRI being less responsive to TNFi treatment. Furthermore, residual entheseal inflammation might be indicative of the need for continuous therapy.
Objective
To determine the prevalence of sacroiliac joint variants in patients with axial spondyloarthritis (axSpA) using MRI-based synthetic CT images and to evaluate their relationships with the ...presence of bone marrow edema, as this may potentially complicate diagnosing active sacroiliitis on MRI in patients with suspected axSpA.
Methods
172 patients were retrospectively included. All patients underwent MRI because of clinical suspicion of sacroiliitis. The diagnosis of axSpA was made by a tertiary hospital rheumatologist. Two readers independently determined the presence of bone marrow edema and the presence of one or more of the nine known sacroiliac joint (SIJ) variants.
Results
SIJ variants were common in axSpA patients (82.9%) and the non-SpA group (85.4%); there were no significant differences in prevalence. Bone marrow edema was frequently found in axSpA (86.8%) and non-SpA patients (34%). AxSpA patients with SIJ variants (except for accessory joint) demonstrated 4 to 10 times higher odds for bone marrow edema, however not statistically significant. The more variants were present in this group, the higher the chance of bone marrow edema. However, some multicollinearity cannot be excluded, since bone marrow edema is very frequent in the axSpA group by definition.
Conclusion
SIJ variants are common in axSpA and non-SpA patients. SIJ variants were associated with higher prevalence of bone marrow edema in axSpA patients, potentially due to altered biomechanics, except for accessory joint which may act as a stabilizer.
To determine the link between extraarticular manifestations (EAMs) and baseline characteristics in patients with axial spondyloarthritis (SpA), and to define their potentially differential prognostic ...value in 2 large, independent Belgian axial SpA cohorts with distinct recruitment periods.
Information on demographic and clinical characteristics and extraarticular manifestations (EAMs) was obtained from patients with axial SpA originating from the (Be)Giant (Belgian Inflammatory Arthritis and Spondylitis) cohort, which includes consecutive axial SpA patients whose data have been collected since 2010, and from the ASPECT (Ankylosing Spondylitis Patients Epidemiological Cross-sectional Trial) cohort, a Belgian registry of cross-sectional data collected between February 2004 and February 2005 from consecutive patients with ankylosing spondylitis (AS) or probable AS.
Among the 1,250 Belgian patients studied, disease duration was associated with risk of developing inflammatory bowel disease (IBD), with an increase in risk by 20% per 10 years of disease duration (relative risk RR 1.2, P = 0.026), and associated with risk of developing acute anterior uveitis, with an increase in risk by 30% per 10 years of disease duration (RR 1.3, P < 0.001). In the subgroup of 171 newly diagnosed patients with prospective follow-up data, higher mean C-reactive protein levels over time were demonstrated in those with acute anterior uveitis or IBD compared to those without EAMs or those with psoriasis alone (each P = 0.01).
The risk of developing acute anterior uveitis or IBD, but not psoriasis, in patients with axial SpA seems to increase with disease duration and appears to be linked to a higher cumulative exposure to inflammation, thus providing a possible explanation for the differential structural progression observed in those with axial SpA.
To study the reliability and construct validity of ultrasound in interphalangeal finger joints affected by erosive osteoarthritis (EOA) and non-EOA with MRI as the reference method.
252 joints were ...examined by ultrasound, conventional radiography and clinical examination. Ultrasound was performed using a high-frequency linear transducer (12 × 18 MHz). On the same day, magnetic resonance images of 112 joints were obtained on a 3.0 T magnetic resonance unit. The ultrasound and MRI images were re-read independently by other readers unaware of the diagnosis, clinical and other imaging findings. Interobserver reliability was calculated by the percentage of exact agreement obtained and κ statistics. With MRI as the reference method, the sensitivity and specificity of ultrasound in detecting structural (bone erosions and osteophytes) and soft tissue (effusion and grey-scale synovitis) changes in EOA were calculated.
Ultrasound and MRI were found to be more sensitive in detecting erosions than conventional radiography in EOA. A high agreement between ultrasound and MRI in the assessment of bone erosions (77.7%), osteophytes (75.9%) and synovitis (86.5%) was present. A high percentage of inflammatory changes was found in EOA, and in smaller amount in non-EOA, both confirmed by MRI. Good interobserver reliability of ultrasound was obtained for all variables (all median κ > 0.8).
Grey-scale ultrasound proved to be a reliable and valid imaging technique to assess erosions and soft tissue changes, compared with MRI as a reference method in EOA.
Objectives
To evaluate the feasibility and diagnostic accuracy of a deep learning network for detection of structural lesions of sacroiliitis on multicentre pelvic CT scans.
Methods
Pelvic CT scans ...of 145 patients (81 female, 121 Ghent University/24 Alberta University, 18–87 years old, mean 40 ± 13 years, 2005–2021) with a clinical suspicion of sacroiliitis were retrospectively included. After manual sacroiliac joint (SIJ) segmentation and structural lesion annotation, a U-Net for SIJ segmentation and two separate convolutional neural networks (CNN) for erosion and ankylosis detection were trained. In-training validation and tenfold validation testing (U-Net—
n
= 10 × 58; CNN—
n
= 10 × 29) on a test dataset were performed to assess performance on a slice-by-slice and patient level (dice coefficient/accuracy/sensitivity/specificity/positive and negative predictive value/ROC AUC). Patient-level optimisation was applied to increase the performance regarding predefined statistical metrics. Gradient-weighted class activation mapping (Grad-CAM++) heatmap explainability analysis highlighted image parts with statistically important regions for algorithmic decisions.
Results
Regarding SIJ segmentation, a dice coefficient of 0.75 was obtained in the test dataset. For slice-by-slice structural lesion detection, a sensitivity/specificity/ROC AUC of 95%/89%/0.92 and 93%/91%/0.91 were obtained in the test dataset for erosion and ankylosis detection, respectively. For patient-level lesion detection after pipeline optimisation for predefined statistical metrics, a sensitivity/specificity of 95%/85% and 82%/97% were obtained for erosion and ankylosis detection, respectively. Grad-CAM++ explainability analysis highlighted cortical edges as focus for pipeline decisions.
Conclusions
An optimised deep learning pipeline, including an explainability analysis, detects structural lesions of sacroiliitis on pelvic CT scans with excellent statistical performance on a slice-by-slice and patient level.
Clinical relevance statement
An optimised deep learning pipeline, including a robust explainability analysis, detects structural lesions of sacroiliitis on pelvic CT scans with excellent statistical metrics on a slice-by-slice and patient level.
Key Points
•
Structural lesions of sacroiliitis can be detected automatically in pelvic CT scans.
•
Both automatic segmentation and disease detection yield excellent statistical outcome metrics.
•
The algorithm takes decisions based on cortical edges, rendering an explainable solution.
We compared embryogenic capacities of integument explants excised from three sources of the Hevea brasiliensis (Müll. Arg.) mature genotype PB 260. The three sources were 17-year-old (BT 86) and ...7-year-old (BT 96) budded trees and 7-year-old emblings (EM 96). The highest proportions of embryogenic calluses obtained from the total number of integument explants initially used were from trees of EM 96 origin, followed by BT 96 trees, with explants from BT 86 trees producing the lowest number of embryogenic calluses. Further initiation of embryogenic callus lines from the primary somatic embryos derived from the three sources was successful only for EM 96. Somatic embryo cultures from BT 86 and BT 96 sources produced only friable calluses that could not be further amplified. Overall, somatic embryo explants derived from EM 96 responded over a wider range of 3,4-dichlorophenoxyacetic acid and kinetin concentrations than the somatic embryo explants from BT 86 and BT 96 origins. The effects of chronologic, ontogenetic and physiologic aging on explant capacity for somatic embryogenesis and on the overall efficiency of the process in H. brasiliensis are discussed.
PDF
