Dynamic contrast-enhanced magnetic resonance imaging of the synovium and synovial subregions in knee osteoarthritis: test-retest repeatability

Introduction

Knee osteoarthritis (OA) is a joint disorder significantly affecting the daily living activities of hundreds of millions of individuals worldwide, with an enormous impact on healthcare costs (1). Knee OA was originally thought to be the result of simple “wear and tear” of the cartilage. However, in recent years it has become clear that knee OA is a complex disease of the whole joint, involving low grade inflammation of other knee structures, such as the infrapatellar fat pad, the bone marrow, and the synovial membrane (2-4). Inflammation of the synovial membrane (synovitis) is especially common in knee OA, occurring in up to 90% of patients (5). The presence and extent of synovitis is strongly associated with both knee pain severity (6) and disease progression (7-10), and several studies even suggest that synovial inflammation could be an independent driver of OA progression (11). Therefore, synovitis is increasingly regarded as a specific target for disease-modifying treatments aimed at slowing down or halting OA progression (11). In the context of synovitis, there is increased synovial perfusion due to neovascularization accompanied with increased vessel permeability (12).

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is an advanced imaging technique that can potentially be used to assess synovial perfusion. DCE-MRI visualizes the accumulation and washout of gadolinium-based contrast agents in tissues of interest over time. The temporal enhancement patterns induced by these contrast agents are related to microcirculatory characteristics of the tissues. From the enhancement pattern, semi-quantitative and quantitative measures can be extracted, which are considered measures of perfusion and can be used as surrogate measures of inflammation. DCE-MRI has shown promise in accurately characterizing the extent of synovitis in patients with knee OA by quantifying synovial perfusion, which is increased in the context of inflammation (13). In patients with knee OA, DCE-MRI is more strongly correlated with pain than static contrast-enhanced (CE)-MRI (13). In addition, quantitative DCE-MRI has shown a stronger correlation with macroscopic and microscopic features of synovitis than qualitative or semi-quantitative assessments of synovitis on non-contrast MRI or static CE-MRI (14).

Over the years, DCE-MRI has become of particular interest for patient selection and evaluation of treatment response in clinical trials investigating novel disease-modifying OA treatments targeted at reducing synovitis. However, full adoption of DCE-MRI in this setting has not yet been achieved, at least partly due to a lack of standardization in both image acquisition and processing methods. In addition, thorough technical validation is needed to push the field forward, but test-retest data for DCE-MRI is scarce due to the need for contrast administration. To our knowledge, only two prior DCE-MRI repeatability study in patients with knee OA exists, showing good repeatability of DCE-MRI parameters (15,16). More repeatability studies are needed though, because DCE-MRI parameters are highly dependent on acquisition and post-processing methods (17). Moreover, methods that are capable of precise assessment of subregional synovial changes are needed, due to the rise of OA treatments, such as genicular artery embolization (GAE), that target specific anatomical subregions of the synovium (18).

The purpose of this study is to evaluate test-retest repeatability of DCE-MRI within the synovium of patients with knee OA, and develop a precise semi-automatic method for synovial subregional assessment. We present this article in accordance with the STROBE reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1800/rc).

Methods

Participants

This study is a secondary explorative analysis of patients with mild or moderate knee OA included in a prospective randomized controlled trial to test GAE who received sham GAE (19). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Erasmus University Medical Center institutional research review board (No. MEC-2018-081) and informed consent was provided by all individual participants. The sham GAE procedure involved a mock procedure and no actual treatment, and no other disease modifying treatments were performed during the study period. Inclusion criteria were age >18 years, knee pain for >6 months, mild or moderate radiographic knee OA (Kellgren and Lawrence grade 1–3) (20), and insufficient response to conservative treatment for at least 6 months as determined by an orthopedic surgeon. Exclusion criteria included contra-indications for MRI or angiography, previous surgical knee OA treatment, musculoskeletal co-morbidities, or renal insufficiency. Participants who used pain medication were included, but no changes to the dosage were made during the study period.

Image acquisition

Participants underwent DCE-MRI imaging of their self-assessed most symptomatic knee at baseline and at 1-month follow-up. Imaging was performed at 3 Tesla (SIGNA Premier, GE Healthcare, Waukesha, Wisconsin, USA) using a dedicated knee coil (18 channel Transmit/Receive Knee Coil, Quality Electrodynamics, Mayfield Village, Ohio, USA). Patients were in supine position with their knee fixated in the coil using padding to minimize motion during scanning. All imaging was performed on the same day of the week and around the same time of the day throughout the study.

Image acquisition included a pre-contrast 3D spoiled gradient echo (SPGR) sequence with variable flip angle and two-point Dixon water-fat separation for T1 mapping, as well as B1 mapping with Bloch-Siegert shift. Dynamic imaging was acquired with a sagittal SPGR sequence [DIfferential Sub-sampling with Cartesian Ordering (DISCO)] with two-point Dixon water-fat separation. Pulse sequence details are provided in Table 1. Spatial resolution was optimized to allow for visualization of small vessels, with interpolated voxel size 0.78×0.78 mm² and slice thickness 0.5 mm. The field of view was set to 200 mm × 200 mm × 120 mm to ensure inclusion of the suprapatellar recess, which can extend cranially and is often distended in patients with knee OA due to effusion. After acquisition of the initial mask phase, 0.1 mL/kg Gadovist (Bayer, Leverkusen, Germany) was intravenously injected at 1.0 mL/s, followed by a 15 mL saline flush. Subsequently 34 phases were acquired, with a temporal resolution of 10.2 seconds per phase, providing a total imaging duration of 5 minutes 47 seconds after contrast injection. DCE imaging was followed by static post-contrast imaging using a T1-weighted water-excitation SPGR pulse sequence.

Outcome measures

All outcome measures were acquired at baseline and 1-month follow-up. Knee injury and Osteoarthritis Outcome Scores (KOOS) (21) pain subscale ranging from 0 to 100 was collected via a survey. A decrease in KOOS score means less pain. The semi-quantitative synovitis scores according to Guermazi et al. (22) were assessed on the static CE-MRI sequence. DCE-MRI image analysis and pharmacokinetic modeling are described in detail in the next paragraphs.

DCE-MRI image analysis

For all DCE-MRI images, motion compensation was performed with rigid groupwise registration (23), and follow-up images were registered to their corresponding baseline image with an affine transform using Elastix (v5.0.1) (24). Segmentation of the whole synovium was performed by a trained researcher (J.M.M.) under supervision of a musculoskeletal radiologist (R.A.v.d.H.) with >5 years of experience. Similar as in an earlier study (16), an initial rough segmentation was performed manually on the water-only image of the last phase of the DCE-MRI acquisition followed by automatic selection of enhanced voxels by the semi-automatic shuffle transform method (25). Enhanced voxels were selected by subtracting the initial pre-contrast image from the last dynamic phase, while correcting for residual patient motion. This motion correction was done by minimizing the absolute difference between an enhanced voxel and the corresponding voxels in a 3×3 voxel area in the pre-contrast image. The resulting shuffle-transformed images were then converted to binary synovial masks by selecting the 20% of voxels with the largest enhancement. Full details can be found in the original paper (25). For the 1-month follow-up scans, regions of interest (ROIs) were created by applying the rough baseline masks to the registered follow-up images, followed by a visual check whether the synovium was covered and repetition of the shuffle transform.

The whole synovium was divided in eight subregions according to the eight genicular arteries (Figure 1), wherein voxels were attributed to the closest genicular artery. First, on all baseline scans the midlines of the genicular arteries (18) were determined semi-automatically using an in-house developed module in MeVisLab (MeVis Medical Solutions AG, version 3.3). Vessels were manually located and the centerlines were extracted using a marker-controlled minimum cost path method. Then, subregional segmentations of the synovium were created by assigning each voxel in the image to its closest genicular artery using a distance transform, taking the Euclidean distance without taking anatomical boundaries into account, and then selecting the intersections of the whole-synovium ROI and the areas associated with each genicular artery (see Figure 2). The same method was applied to the baseline-registered follow-up images.

Figure 1 Schematic depiction of eight genicular arteries that were mapped for subregional assessment of synovial perfusion. Created in BioRender. Mostert et al. [2025] (https://BioRender.com/0bj69ny).

Figure 2 Stepwise depiction of our segmentation process and the resulting outputs. (A) Example DCE-MRI image in axial direction. Field of view is cropped for visualization purposes. (B) Initial manual segmentation of the synovium. (C) Whole synovium segmentation made by shuffle transform of (B). (D) Distance map after semi-automatic vessel mapping. Colors correspond to genicular arteries (see Figure 1). For example, pink represents the lateral inferior genicular artery and yellow represents the descending genicular artery. (E) Final synovial subregion segmentations through intersection of (C) and (D). (F) K^trans (min⁻¹) map of the synovium in sagittal direction. (G) v_e map. (H) v_p map. DCE, dynamic contrast-enhanced; MRI, magnetic resonance imaging; K^trans, transfer constant between blood plasma and extravascular extracellular space; v_e, the fractional volume of extravascular extracellular space; v_p, the fractional volume of blood plasma.

DCE pharmacokinetic modeling

Native T1 mapping and pharmacokinetic DCE modelling was performed using open-source software (MADYM, v4.21.1) (26) using the B1 map for field inhomogeneity correction. For pharmacokinetic modeling, the extended Tofts pharmacokinetic model in combination with a population-derived arterial input function (AIF) was applied to the DCE water-only images (27). Hematocrit correction was set at 0.42 for all patients. The extended Tofts model incorporates the parameters K^trans representing the transfer constant between blood plasma and extravascular extracellular space (EES), v_e representing the fractional volume of EES, and v_p representing the fractional volume of blood plasma, as well as a delay factor τ for the bolus arrival time. Voxel-wise perfusion parameters K^trans, v_e, and v_p were calculated. In addition, the semi-quantitative parameter quantifying the initial area under the contrast concentration curve at 60 seconds (IAUC₆₀) was also calculated. For each parameter, median values were taken across the entire enhancing synovium and for each synovial subregion separately, and these subject-wise median values were used for all further statistical analyses.

Statistics

Test-retest repeatability was evaluated in accordance with Quantitative Imaging Biomarkers Alliance (QIBA) standards (28,29) by estimating within-subject variance (σ_w²) and between-subject variance (σ_b²) with a two-way random effects model. Differences between baseline and one-month values for the whole synovium were visualized with limits of agreements and their 95% confidence intervals (CIs) using Bland-Altman plots. Intraclass correlation coefficients (ICCs) were calculated from these variances for all DCE parameters, and evaluation of ICC values was done according to thresholds proposed by Koo et al. (30). Within-subject standard deviation (wSD) was calculated for parameters of which the variability was approximately invariant to its magnitude. When the variability of a parameter varied with its magnitude, its within-subject coefficients of variation (wCV) was calculated on the log-transform of the parameter [following equation 12 of QIBA standards (28)]. Correspondingly, the repeatability coefficient (RC) was calculated from either the wSD or wCV. The RC, which is also known as the smallest detectable change, represents the smallest difference that can be considered real and unbiased by measurement noise with 95% difference (29). All statistics were calculated for the whole synovium and each subregion individually. Participant characteristics were described using mean and standard deviation (SD). Statistical analyses were performed using open-source Python modules (Pingouin 0.5.5 and Scipy 1.15.2 in python 3.12.9).

Results

Study population

A total of 31 participants were included, among which were two cross-over participants who were allocated to the intervention group but received no GAE treatment. One participant was excluded from analysis due to failed DCE-MRI acquisition. Participant characteristics are presented in Table 2. Mean participant age at baseline was 57.9 years (SD 8.0 years) with a mean duration of knee pain symptoms of 7.8 years (SD 8.1 years). Mean difference in KOOS pain subscore between baseline and 1-month follow-up was −19.7 (95% CI −25.3 to −14.1; P<0.001), while there was no significant difference in the semi-quantitative synovitis score (−0.03; 95% CI −0.76 to 0.69; P=0.93).

Test-retest repeatability of the whole synovium

Repeatability measures for all perfusion parameters extracted from the whole synovium are presented in Table 3. Median K^trans showed the highest repeatability across baseline and 1-month follow-ups, with an ICC of 0.84 and a RC of 0.039 (95% CI: 0.030–0.049). Semi-quantitative parameter IAUC₆₀ showed a similar ICC of 0.85 and a slightly higher RC of 0.090 (95% CI: 0.068–0.113). Bland-Altman plots showing agreement between baseline and 1-month measurements for all quantitative perfusion parameters are shown in Figure 3. An example case showing K^trans maps at baseline and 1-month follow-up is presented in Figure 4.

Figure 3 Bland-Altman plots for DCE-MRI parameters for the whole synovium in patients with knee osteoarthritis. Each point represents the median value across the whole synovium, showing the difference as function of the mean of the baseline and 1-month follow-up measurements of K^trans, v_e, v_p, and IAUC₆₀. Dotted lines indicate the limits of agreement (mean ±1.96 standard deviation of the differences) and shaded areas indicate 95% confidence intervals. DCE, dynamic contrast-enhanced; IAUC₆₀, initial area under the contrast concentration curve at 60 seconds; MRI, magnetic resonance imaging; K^trans, transfer constant between blood plasma and extravascular extracellular space; v_e, the fractional volume of extravascular extracellular space; v_p, the fractional volume of blood plasma.

Figure 4 Example DCE-MRI images in axial direction with K^trans (min⁻¹) maps overlayed. Left: baseline. Right: 1-month follow-up, registered to the baseline image. The left knee is imaged, and an area of high K^trans-values is seen on the lateral side. DCE, dynamic contrast-enhanced; K^trans, transfer constant between blood plasma and extravascular extracellular space; MRI, magnetic resonance imaging.

Segmentation and test-retest repeatability for synovial subregions

Segmentation of synovial subregions through semi-automatic vessel mapping yielded subregional ROIs for each participant. In some participants one or more subregions corresponding to the anterior tibial recurrent artery (ATRA, n=7) or descending genicular artery (DGA, n=1) did not contain any voxels because there were no voxels in the whole-synovium ROI for which these arteries were the closest.

For the synovial subregions, repeatability measures for K^trans are provided in Table 4. Within-region ICCs ranged between 0.70 and 0.89 while RCs ranged between 0.028 and 0.099. Repeatability metrics for v_e, v_p, and IAUC₆₀ are provided in the supplementary material (Tables S1-S3).

Discussion

In this study we aimed to evaluate test-retest repeatability of DCE-MRI within the whole synovium of patients with knee OA, and develop a precise semi-automatic method for synovial subregion assessment. Our results indicate that DCE-MRI parameters have good test-retest repeatability, both in the whole synovium and in synovial subregions. K^trans has shown the best repeatability out of the quantitative perfusion parameters.

Previous studies have reported on the use of DCE-MRI in knee OA, evaluating inflammation in several anatomical structures including the infrapatellar fat pad (2,31), peri-articular muscles (32), subchondral bone (3), and bone marrow lesions (4). Application of quantitative DCE-MRI in synovial tissues has also been reported in several studies (13,33-35), showing that quantitative DCE-MRI of the synovium in patients with knee OA is more strongly correlated to KOOS pain scores than synovial volume measurement and static CE-MRI. In addition, inter- and intra-observer variability for quantitative DCE-MRI parameters were excellent, and lower than for the semi-quantitative parameter IAUC (13).

To the best of our knowledge, only one other trial reported on test-retest repeatability of DCE-MRI in the knee synovium. MacKay et al. prospectively evaluated quantitative perfusion parameters in fourteen patients with mild to moderate knee OA and six age-matched healthy volunteers (16). In this study, ICC for K^trans was 0.90, comparable to our findings. However, our repeatability for K^trans was substantially lower with a RC of 0.039 compared to 0.013. Interestingly, our repeatability for the other quantitative parameters v_e and v_p was substantially better compared to that reported by MacKay et al. These differences in findings may be attributable to differences in image acquisition protocols. Voxel size in our trial was much smaller, which was needed to visualize small vessels for subregional assessment, but consequently results in significantly lower signal-to-noise ratio and lower temporal resolution. We also used a different gadolinium-based contrast agent at a slightly lower injection rate and, more importantly, with a lower dose, which also resulted in lower signal-to-noise ratio. Important to note is that our dose of 0.1 mL/kg is the standard dose used in clinical care. Another potential explanation could be the difference in study populations, since MacKay et al. included a small number of patients and also healthy volunteers, the latter supposedly have even more stable perfusion parameters in the absence of pain. Opposite to Mackay et al. we chose to apply a general hematocrit value across the study population. Patient-specific hematocrit correction is not often done in literature and its benefit is still subject of debate. This is further complicated by confounding factors such as the Fåhræus effect, indicating that hematocrit values decrease with vessel size and thus lower hematocrit values are expected in the capillaries where the exchange actually occurs (36). In cancer patients undergoing chemotherapy it may be important to apply patient specific hematocrit correction (37), but in our population we do not expect any relevant changes in hematocrit values and therefore no benefits of patient-specific corrections are expected. Finally, we used a slightly different segmentation method with a simple threshold while MacKay et al. used Otsu thresholding. However, ad hoc analysis on our population with Otsu’s method for ROI selection showed slightly worse repeatability for all perfusion parameters compared to simple thresholding (Table S4).

Repeatability of quantitative DCE-MRI has been reported in other inflammatory joint diseases such as rheumatoid arthritis. For example, Waterton et al. evaluated DCE-MRI repeatability in the synovium of the hand and wrist and reported a wCV of 30.0% for K^trans_, 53.4% for v_p and 31.4% for IAUC₆₀. Although these findings cannot directly be compared to our results, notably K^trans and IAUC₆₀ had comparable repeatability and were preferred over other quantitative DCE-MRI parameters (38).

Our findings indicate that quantitative DCE-MRI can be used to evaluate synovitis in knee OA patients and assess changes over time, for example in clinical trials investigating the effect of disease modifying interventions. With our method, subject changes in K^trans of 0.039 or larger can be considered an effect of the intervention with 95% certainty, suggesting that K^trans can be applied as an objective imaging biomarker for synovitis in knee OA patients. However, smaller changes in DCE-MRI parameters cannot be adequately differentiated, possibly limiting the use of DCE-MRI for applications that require quantification of subtle changes in synovial perfusion. We additionally evaluated the sensitivity in subregional analyses; this is relevant when effect sizes are expected to be dependent on the feeding artery, as is the case with GAE, which has been proposed as a novel treatment for knee OA. Our findings indicate overall good test-retest repeatability in synovial subregions, but potentially lower repeatability values in synovial subregions corresponding to the median genicular artery (MGA), DGA, and ATRA. Changes in DCE-MRI parameters within these regions are, therefore, potentially less detectable, but it should be noted that our findings for the ATRA and, to lesser extent, DGA are based on only a subset of participants since for some patients no voxels within the synovium were assigned to these regions. Moreover, subregional segmentations were created by mapping the distance to each genicular artery. This strategy assumes that any part of the synovium is supplied with blood by the artery with the shortest distance to that part, which is likely not always the case. As such, subregional perfusion outcomes should be interpreted with caution.

Test-retest DCE-MRI data is scarce and therefore this work fills an important gap in the clinical translation of quantitative DCE-MRI derived imaging biomarkers. A strength is the relatively large number of participants and the standardized technical performance evaluation according to the recommendations defined by the QIBA (28,29). Another strength is the application of the semi-automatic segmentation method, which reduces the time burden to the user and ensures the delineation of the synovial areas with active inflammation better reflecting disease activity. Another strength and novelty is the addition of the subregional segmentations, which enables assessment of perfusion at a subregional level. This is important to assess response to targeted treatment, like GAE.

Some limitations of this trial need to be taken into account. First, the test-retest interval was relatively long (1 month), and clinical fluctuations in synovitis could have occurred. The fact that there was no significant difference in the synovitis score between baseline and one-month follow-up is reassuring, but small fluctuations could still have taken place. Therefore, the measured variability in DCE parameters likely includes both methodological and biological variability, and the true methodological variability is likely lower than our reported variability. Additionally, our subjects were randomized to the control group of a clinical trial, and as such underwent a sham (‘placebo’) intervention. Although no effect on synovial perfusion is expected in these subjects, participants did report a significant reduction in pain symptoms after the sham intervention, similarly to the participants who did receive GAE (19). DCE-MRI biomarkers show no clear trend in any direction and thus we believe our data allows for a reliable evaluation of test-retest repeatability. Another potential limitation is that we used a population-derived AIF, which is known to be robust but does not capture the inter-subject variations in the arterial input. The use of subject-specific AIFs may improve the characterization of subject-specific physiology and improve reproducibility. However, subject-specific AIF can also introduce additional variability (27). In our study, capturing the subject-specific AIF was difficult because of T2* induced arterial intensity reduction at the peak of the bolus. Finally, this is a single center study, and a multicenter study is needed to provide clear benchmarks. However, this study does replicate the findings from the prior study, demonstrating the potential of DCE-MRI.

Conclusions

Semi-automatic segmentation of the synovium and synovial subregions on DCE-MRI in knee OA is feasible. Quantitative DCE-MRI biomarkers have good test-retest repeatability on both the whole synovium and synovial subregions, with K^trans showing the best performance.

Acknowledgments

Some of the data have been presented in a more preliminary form at the annual meetings of the International Society for Magnetic Resonance in Medicine and the European Society for Magnetic Resonance in Medicine and Biology.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1800/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1800/dss

Funding: This work was supported by Cook Medical, Erasmus Medisch Centrum, Stichting Coolsingel and Boston Scientific.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2025-1800/coif). E.H.G.O. serves as an unpaid editorial board member of Quantitative Imaging in Medicine and Surgery. T.A.v.Z. and E.H.G.O. have received research grant support from Cook Medical, Stichting Coolsingel, Boston scientific and Erasmus University Medical Center. D.H.J.P. and E.H.G.O. have received research grant support from GE Healthcare. The other authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Erasmus University Medical Center institutional research review board (No. MEC-2018-081) and informed consent was provided by all individual participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Cui A, Li H, Wang D, Zhong J, Chen Y, Lu H. Global, regional prevalence, incidence and risk factors of knee osteoarthritis in population-based studies. EClinicalMedicine 2020;29-30:100587. [Crossref] [PubMed]
2. de Vries BA, van der Heijden RA, Poot DHJ, van Middelkoop M, Meuffels DE, Krestin GP, Oei EHG. Quantitative DCE-MRI demonstrates increased blood perfusion in Hoffa's fat pad signal abnormalities in knee osteoarthritis, but not in patellofemoral pain. Eur Radiol 2020;30:3401-8. [Crossref] [PubMed]
3. de Vries BA, van der Heijden RA, Verschueren J, Bos PK, Poot DHJ, van Tiel J, Kotek G, Krestin GP, Oei EHG. Quantitative subchondral bone perfusion imaging in knee osteoarthritis using dynamic contrast enhanced MRI. Semin Arthritis Rheum 2020;50:177-82. [Crossref] [PubMed]
4. Daugaard CL, Riis RG, Bandak E, Gudbergsen H, Henriksen M, Bliddal H, Boesen M. Perfusion in bone marrow lesions assessed on DCE-MRI and its association with pain in knee osteoarthritis: a cross-sectional study. Skeletal Radiol 2020;49:757-64. [Crossref] [PubMed]
5. Roemer FW, Kassim Javaid M, Guermazi A, Thomas M, Kiran A, Keen R, King L, Arden NK. Anatomical distribution of synovitis in knee osteoarthritis and its association with joint effusion assessed on non-enhanced and contrast-enhanced MRI. Osteoarthritis Cartilage 2010;18:1269-74. [Crossref] [PubMed]
6. Baker K, Grainger A, Niu J, Clancy M, Guermazi A, Crema M, Hughes L, Buckwalter J, Wooley A, Nevitt M, Felson DT. Relation of synovitis to knee pain using contrast-enhanced MRIs. Ann Rheum Dis 2010;69:1779-83. [Crossref] [PubMed]
7. Hill CL, Hunter DJ, Niu J, Clancy M, Guermazi A, Genant H, Gale D, Grainger A, Conaghan P, Felson DT. Synovitis detected on magnetic resonance imaging and its relation to pain and cartilage loss in knee osteoarthritis. Ann Rheum Dis 2007;66:1599-603. [Crossref] [PubMed]
8. MacFarlane LA, Yang H, Collins JE, Jarraya M, Guermazi A, Mandl LA, Martin SD, Wright J, Losina E, Katz JN. MeTeOR Investigator Group. Association of Changes in Effusion-Synovitis With Progression of Cartilage Damage Over Eighteen Months in Patients With Osteoarthritis and Meniscal Tear. Arthritis Rheumatol 2019;71:73-81. [Crossref] [PubMed]
9. Roemer FW, Kwoh CK, Hannon MJ, Hunter DJ, Eckstein F, Fujii T, Boudreau RM, Guermazi A. What comes first? Multitissue involvement leading to radiographic osteoarthritis: magnetic resonance imaging-based trajectory analysis over four years in the osteoarthritis initiative. Arthritis Rheumatol 2015;67:2085-96. [Crossref] [PubMed]
10. Atukorala I, Kwoh CK, Guermazi A, Roemer FW, Boudreau RM, Hannon MJ, Hunter DJ. Synovitis in knee osteoarthritis: a precursor of disease? Ann Rheum Dis 2016;75:390-5. [Crossref] [PubMed]
11. Mathiessen A, Conaghan PG. Synovitis in osteoarthritis: current understanding with therapeutic implications. Arthritis Res Ther 2017;19:18. [Crossref] [PubMed]
12. Mapp PI, Walsh DA. Mechanisms and targets of angiogenesis and nerve growth in osteoarthritis. Nat Rev Rheumatol 2012;8:390-8. [Crossref] [PubMed]
13. Riis RG, Gudbergsen H, Henriksen M, Ballegaard C, Bandak E, Röttger D, Bliddal H, Hansen BB, Hangaard S, Boesen M. Synovitis assessed on static and dynamic contrast-enhanced magnetic resonance imaging and its association with pain in knee osteoarthritis: A cross-sectional study. Eur J Radiol 2016;85:1099-108. [Crossref] [PubMed]
14. Shakoor D, Demehri S, Roemer FW, Loeuille D, Felson DT, Guermazi A. Are contrast-enhanced and non-contrast MRI findings reflecting synovial inflammation in knee osteoarthritis: a meta-analysis of observational studies. Osteoarthritis Cartilage 2020;28:126-36. [Crossref] [PubMed]
15. van der Leij C, Lavini C, van de Sande MG, de Hair MJ, Wijffels C, Maas M. Reproducibility of DCE-MRI time-intensity curve-shape analysis in patients with knee arthritis: A comparison with qualitative and pharmacokinetic analyses. J Magn Reson Imaging 2015;42:1497-506. [Crossref] [PubMed]
16. MacKay JW, Nezhad FS, Rifai T, Kaggie JD, Naish JH, Roberts C, Graves MJ, Waterton JC, Janiczek RL, Roberts AR, McCaskie A, Gilbert FJ, Parker GJM. Dynamic contrast-enhanced MRI of synovitis in knee osteoarthritis: repeatability, discrimination and sensitivity to change in a prospective experimental study. Eur Radiol 2021;31:5746-58. [Crossref] [PubMed]
17. Kim H. Variability in Quantitative DCE-MRI: Sources and Solutions. J Nat Sci 2018;4:e484.
18. Okuno Y, Korchi AM, Shinjo T, Kato S. Transcatheter arterial embolization as a treatment for medial knee pain in patients with mild to moderate osteoarthritis. Cardiovasc Intervent Radiol 2015;38:336-43. [Crossref] [PubMed]
19. van Zadelhoff TA, Bos PK, Moelker A, Bierma-Zeinstra SMA, van der Heijden RA, Oei EHG. Genicular artery embolisation versus sham embolisation for symptomatic osteoarthritis of the knee: a randomised controlled trial. BMJ Open 2024;14:e087047. [Crossref] [PubMed]
20. KELLGREN JH. LAWRENCE JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis 1957;16:494-502. [Crossref] [PubMed]
21. Roos EM, Roos HP, Lohmander LS, Ekdahl C, Beynnon BD. Knee Injury and Osteoarthritis Outcome Score (KOOS)--development of a self-administered outcome measure. J Orthop Sports Phys Ther 1998;28:88-96. [Crossref] [PubMed]
22. Guermazi A, Roemer FW, Hayashi D, Crema MD, Niu J, Zhang Y, Marra MD, Katur A, Lynch JA, El-Khoury GY, Baker K, Hughes LB, Nevitt MC, Felson DT. Assessment of synovitis with contrast-enhanced MRI using a whole-joint semiquantitative scoring system in people with, or at high risk of, knee osteoarthritis: the MOST study. Ann Rheum Dis 2011;70:805-11. [Crossref] [PubMed]
23. Huizinga W, Poot DH, Guyader JM, Klaassen R, Coolen BF, van Kranenburg M, van Geuns RJ, Uitterdijk A, Polfliet M, Vandemeulebroucke J, Leemans A, Niessen WJ, Klein S. PCA-based groupwise image registration for quantitative MRI. Med Image Anal 2016;29:65-78. [Crossref] [PubMed]
24. Klein S, Staring M, Murphy K, Viergever MA, Pluim JP. elastix: a toolbox for intensity-based medical image registration. IEEE Trans Med Imaging 2010;29:196-205. [Crossref] [PubMed]
25. Xanthopoulos E, Hutchinson CE, Adams JE, Bruce IN, Nash AF, Holmes AP, Taylor CJ, Waterton JC. Improved wrist pannus volume measurement from contrast-enhanced MRI in rheumatoid arthritis using shuffle transform. Magn Reson Imaging 2007;25:110-6. [Crossref] [PubMed]
26. Berks M, Parker G, Little R, Cheung S. Madym: A C++ toolkit for quantitative DCE-MRI analysis. J Open Source Softw 2021;6:3523.
27. Parker GJ, Roberts C, Macdonald A, Buonaccorsi GA, Cheung S, Buckley DL, Jackson A, Watson Y, Davies K, Jayson GC. Experimentally-derived functional form for a population-averaged high-temporal-resolution arterial input function for dynamic contrast-enhanced MRI. Magn Reson Med 2006;56:993-1000. [Crossref] [PubMed]
28. Raunig DL, McShane LM, Pennello G, Gatsonis C, Carson PL, Voyvodic JT, Wahl RL, Kurland BF, Schwarz AJ, Gönen M, Zahlmann G, Kondratovich MV, O'Donnell K, Petrick N, Cole PE, Garra B, Sullivan DCQIBA Technical Performance Working Group. Quantitative imaging biomarkers: a review of statistical methods for technical performance assessment. Stat Methods Med Res 2015;24:27-67. [Crossref] [PubMed]
29. Shukla-Dave A, Obuchowski NA, Chenevert TL, Jambawalikar S, Schwartz LH, Malyarenko D, Huang W, Noworolski SM, Young RJ, Shiroishi MS, Kim H, Coolens C, Laue H, Chung C, Rosen M, Boss M, Jackson EF. Quantitative imaging biomarkers alliance (QIBA) recommendations for improved precision of DWI and DCE-MRI derived biomarkers in multicenter oncology trials. J Magn Reson Imaging 2019;49:e101-21. [Crossref] [PubMed]
30. Koo TK, Li MY. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J Chiropr Med 2016;15:155-63. [Crossref] [PubMed]
31. Daugaard CL, Henriksen M, Riis RGC, Bandak E, Nybing JD, Hangaard S, Bliddal H, Boesen M. The impact of a significant weight loss on inflammation assessed on DCE-MRI and static MRI in knee osteoarthritis: a prospective cohort study. Osteoarthritis Cartilage 2020;28:766-73. [Crossref] [PubMed]
32. Bandak E, Boesen M, Bliddal H, Riis RG, Gudbergsen H, Henriksen M. Associations between muscle perfusion and symptoms in knee osteoarthritis: a cross sectional study. Osteoarthritis Cartilage 2015;23:1721-7. [Crossref] [PubMed]
33. Riis RG, Gudbergsen H, Simonsen O, Henriksen M, Al-Mashkur N, Eld M, Petersen KK, Kubassova O, Bay Jensen AC, Damm J, Bliddal H, Arendt-Nielsen L, Boesen M. The association between histological, macroscopic and magnetic resonance imaging assessed synovitis in end-stage knee osteoarthritis: a cross-sectional study. Osteoarthritis Cartilage 2017;25:272-80. [Crossref] [PubMed]
34. Riis RGC, Henriksen M, Klokker L, Bartholdy C, Ellegaard K, Bandak E, Hansen BB, Bliddal H, Boesen M. The effects of intra-articular glucocorticoids and exercise on pain and synovitis assessed on static and dynamic magnetic resonance imaging in knee osteoarthritis: exploratory outcomes from a randomized controlled trial. Osteoarthritis Cartilage 2017;25:481-91. [Crossref] [PubMed]
35. Gait AD, Hodgson R, Parkes MJ, Hutchinson CE, O'Neill TW, Maricar N, Marjanovic EJ, Cootes TF, Felson DT. Synovial volume vs synovial measurements from dynamic contrast enhanced MRI as measures of response in osteoarthritis. Osteoarthritis Cartilage 2016;24:1392-8. [Crossref] [PubMed]
36. Secomb TW, Pries AR, Gaehtgens P. Hematocrit fluctuations within capillary tubes and estimation of Fåhraeus effect. Int J Microcirc Clin Exp 1987;5:335-45.
37. Sahoo P, Gupta PK, Awasthi A, Pandey CM, Patir R, Vaishya S, Saha I, Gupta RK. Comparison of actual with default hematocrit value in dynamic contrast enhanced MR perfusion quantification in grading of human glioma. Magn Reson Imaging 2016;34:1071-7. [Crossref] [PubMed]
38. Waterton JC, Ho M, Nordenmark LH, Jenkins M, DiCarlo J, Guillard G, Roberts C, Buonaccorsi G, Parker GJM, Bowes MA, Peterfy C, Kellner H, Taylor PC. Repeatability and response to therapy of dynamic contrast-enhanced magnetic resonance imaging biomarkers in rheumatoid arthritis in a large multicentre trial setting. Eur Radiol 2017;27:3662-8. [Crossref] [PubMed]