419 CT ARTHROGRAPHY OF THE KNEE TO MEASURE CARTILAGE QUALITY WITH LOW RADIATION EXPOSURE

Osteoarthritis and Cartilage 20 (2012) 678e685

CT arthrography of the human knee to measure cartilage qualitywith low radiation dose

J. van Tiel yz*a, M. Siebelt ya, J.H. Waarsing y, T.M. Piscaer y, M. van Straten z, R. Booij z,M.L. Dijkshoorn z, G.J. Kleinrensink x, J.A.N. Verhaar y, G.P. Krestin z, H. Weinans yk, E.H.G. Oei zyDepartment of Orthopedic Surgery, Erasmus Medical Center, Rotterdam, The NetherlandszDepartment of Radiology, Erasmus Medical Center, Rotterdam, The NetherlandsxDepartment of Neurosciences, Erasmus Medical Center, Rotterdam, The NetherlandskDepartment of Biomechanical Engineering, Delft University of Technology, Delft, The Netherlands

a r t i c l e i n f o

Article history:Received 28 November 2011Accepted 16 March 2012

Keywords:CT arthrographyRadiation doseSulphated glycosaminoglycan contentArticular cartilage qualityTranslational research

* Address correspondence and reprint requests toOrthopedic Surgery, Erasmus Medical Center, P.O. BoThe Netherlands. Tel: 31-10-7044627; Fax: 31-10-704

E-mail address: [email protected] (J. van Tiea Both authors contributed equally.

1063-4584/$ e see front matter � 2012 Osteoarthritidoi:10.1016/j.joca.2012.03.007

s u m m a r y

Objective: Recently, CT arthrography (CTa) was introduced as a non-destructive technique to quantita-tively measure cartilage quality in human knees. This study investigated whether this is also possibleusing lower radiation dose CT protocols. Furthermore, we studied the ability of (lower radiation) CTato distinguish between local sulphated glycosaminoglycan (sGAG) content differences.Design: Of ten human cadaveric knee joints, six CT scans using different radiation doses (81.33e8.13 mGy)were acquired after intra-articular ioxaglate injection. The capability of CTa to measure overall cartilagequality was determined in seven anatomical regions of interest (ROIs), using equilibrium partitioning of anionic contrast agent using (EPIC)-microCT (mCT) as reference standard for sGAG content. To test thecapability of CTa to spatially distinguish between local differences in sGAG content, we calculated thepercentage of pixels incorrectly predicted as having high or low sGAG content by the different CTaprotocols.Results: Low radiation dose CTa correlated well with EPIC-mCT in large ROIs (R ¼ 0.78; R2 ¼ 0.61;P < 0.0001). CTa can also distinguish between high and low sGAG content within a single slice. However,the percentage of incorrectly predicted quality pixels increases (from 35% to 41%) when less radiation isused. This makes is hard or even impossible to differentiate between spatial differences in sGAG contentin the lowest radiation scans.Conclusions: CTa acquired using low radiation exposure, comparable to a regular knee CT, is able tomeasure overall cartilage quality. Spatial sGAG distribution can also be determined using CTa, howeverfor this purpose a higher radiation dose is necessary. Nevertheless, radiation dose reduction makes CTasuitable for quantitative analysis of cartilage in clinical research.

� 2012 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.

Introduction

The current reference standard for grading the severity ofosteoarthritis (OA) in the knee is the radiography based Kellgrenand Lawrence score1. This technique is, however, not sensitiveenough to detect or follow OA at an early stage of the diseasebecause it only indirectly visualizes the cartilage and is not ableto (semi)quantitatively measure cartilage quality2. Therefore,

: J. van Tiel, Department ofx 2040, 3000 CA Rotterdam,4690.l).

s Research Society International. P

sophisticated magnetic resonance imaging (MRI) imaging tech-niques have been developed which can qualitatively measurecartilage quality in terms of the sulphated glycosaminoglycan(sGAG), collagen or sodium content of articular cartilage3e5.

Recently, it has been shown that CT arthrography of the knee(CTa) is able to measure overall cartilage quality in large anatomicalcartilage regions in human cadaveric knees6. Similar to microCT(mCT) arthrography in small animals7,8 and delayed gadoliniumenhanced MRI of cartilage (dGEMRIC) in humans9e11, this tech-nique uses the inversed relationship between a negatively chargedcontrast agent (ioxaglate) and the sGAG content of cartilage.

The reported CTa protocol has a CT-Dose Index (CTDIvol) of81.33 mGy per CTa scan, which poses a limitation on this tech-nique12. Therefore, the radiation dose must be reduced before CTacan be used in clinical research. The use of less radiation to acquire

ublished by Elsevier Ltd. All rights reserved.

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685 679

CT scans results however, in an increase of noise in the recon-structed CT images. This increase of noise may influence themeasured X-ray attenuation values and therefore interfere with thecapability of measuring quality of cartilage using CTa.

Therefore, we designed a cadaver study with the purpose toinvestigate the effect of radiation dose reduction of CTa on its abilityto measure articular cartilage quality in large cartilage regions. Wealso assessed the capability of CTa to distinguish between spatialhigh and low sGAG content of cartilage on a single slice and theinfluence of radiation dose reduction on this capability. The latter isof interest because it could enable the use of CTa as a tool todiagnose (focal) cartilage defects and follow the repair in thesedefects over time.

Methods

Cadaveric knee joints

For this study, we used ten randomly selected cadaveric lowerextremities from eight individuals who had donated their bodies toscience. All extremities were frozen at �20�C directly after death.Before the start of the experiment, the specimens were defrostedslowly in a cooled environment (7�C) for 5 days. All extremitieswere at room temperature during imaging procedures.

Acquisition and post-processing of CTa data

We injected 20 mL of 30% ioxaglate dilution (Hexabrix 320,Mallinckrodt, Hazelwood, MO, USA and saline) intra-articularly inall knee joints, using an 18 gauge needle. After the injection, weflexed (w120�) and extended (w0�) the knee joints for 5 min inorder to achieve optimal distribution of the contrast agentthroughout the joint cavity. Ten minutes after contrast injection,CTa scans of all knee joints were acquired using a second gener-ation dual source multidetector spiral CT scanner (SOMATOMDefinition Flash, Siemens Healthcare AG, Erlangen, Germany)with a tube voltage of 80 kV, an effective mAs value of 3,140 mAs,a pitch of 0.35 and a collimation of 32 � 0.6 mm, resulting in aCTDIvol of 81.33 mGy6. This protocol will be referred to asmaximum dose in this paper. Directly after the first scan, fiveadditional scans were acquired using the same tube voltage(80 kV), but with reduced radiation exposures: 1,570 mAs (50%),1,256 mAs (40%), 942 mAs (30%), 628 mAs (20%) and 314 mAs(10%) per scan. All knee joints were scanned in the axial planewith a scanning time of 30 s per scan. All CT datasets werereconstructed with an effective slice thickness of 0.75 mm anda sharp reconstruction kernel. Multiplanar reconstruction wasperformed (image pixel size 0.265 mm).

Fig. 1. 3D representation of the seven analyzed large cartilage ROIs per k

Using Skyscan analysis software (Skyscan, Kontich, Belgium), wesegmented all CT datasets into binary datasets using one fixedattenuation threshold of 500 Hounsfield units (HU) that wasselected because it resulted in the best segmentation of the carti-lage6. Next, we manually defined seven anatomical cartilageregions of interest (ROIs) in all CT datasets based on the nomen-clature and scheme as suggested by Eckstein et al.13. Each ROIconsisted of 40 consecutive slices covering the central weight-bearing area of the cartilage of both the medial and lateral femoralcondyles (wbMC and wbLC), the posterior non-weight bearingcartilage area of both femoral condyles (pMC and pLC), bothweight-bearingmedial and lateral tibial plateaus (wbMP andwbLP)and the mid-portion of the patellar cartilage (mpP) [Fig. 1(AeC)].After defining all ROIs, we calculated the mean X-ray attenuationper cartilage ROI on the CTa scans.

Equilibrium partitioning of an ionic contrast agent using (EPIC)-mCT

Mean X-ray attenuation values of EPIC-mCT have a good corre-lation with the sGAG content of articular cartilage measured witha dimethylmethylene blue essay or quantified with optical densitymeasurements14e16. Therefore, we selected the outcomes of EPIC-mCT in mean X-ray attenuation values as our reference test of sGAGcontent of articular cartilage.

After CTa, all knee joints were dissected into five parts: themedial and lateral femoral condyles, the medial and lateral tibialplateaus and the patella. Soft tissue was removed to a maximalextent, without harming the integrity of the cartilage. In order toachieve equilibrium between the contrast agent and the sGAGcontent of the cartilage, all dissected specimens were incubated inan ioxaglate dilution (Hexabrix 320, Mallinckrodt, Hazelwood, MO,USA and saline) for 24 h at room temperature17e19. We used a 20%dilution of ioxaglate, which resulted in the best cartilage segmen-tation at the air/cartilage and bone/cartilage interfaces6.

mCT scans were performed on a Skyscan 1076 in vivo mCT scanner(Skyscan, Kontich, Belgium). The following scan settings were used:isotropic voxel size of 35 mm; a voltage of 55 kV; a current of181mA; field of view 68mm; a 0.5 mm aluminium filter; 198� witha 0.4� rotation step6. Scanning time per specimen was 6e10 h,depending on the size of the specimen (patella, plateau orcondyle). A plastic foil was wrapped around the specimen toavoid dehydration during scanning. All scans were performedusing the same settings and all data were reconstructed identically.

Using Skyscan analysis software, we segmented the mCT data-sets using a fixed attenuation threshold between air and sub-chondral bone that was selected visually for the best segmentationresult in all datasets6. In all segmented mCT datasets, sevenanatomical ROIs of the cartilage corresponding with ROIs of the CTa

nee joint: (A) pMC/pLC; (B) wbMC/wbLC; (C) wbMP/wbLP and mpP.

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685680

were drawn manually and the mean X-ray attenuation per ROIwas calculated.

Spatial analysis of cartilage quality

Using commercially available software (Matlab version 7.1,MathWorks, Natick, MA, USA and Multimodality Image Registra-tion using Information Theory (MIRIT), Laboratory for MedicalImaging Research, Leuven, Belgium20), all CTa (50%, 40%, 30%, 20%and 10%) and EPIC-mCT datasets were registered using the datasetthat was acquired at the maximum dose as reference. Registrationof the datasets enabled comparison of corresponding cartilageregions (femoral condyles, tibial plateaus and patellar cartilage) inall CTa scans per knee.

To study the capability of CTa to analyze the spatial distributionof high and low sGAG content in cartilage and the influence ofradiation dose reduction on this capability, we used the EPIC-mCTas reference standard for spatial sGAG distribution in cartilage[Fig. 2(A)]14e16. Using Skyscan analysis software, we defined an areaof high and low sGAG content in the cartilage within a central slicethrough the medial and lateral tibiofemoral joint and on a centralslice of the mpP in all CTa datasets (maximum dose, 50%, 40%, etc.)[Fig. 2(BeD)]. To define these areas (which wewill refer to as masksfrom now on), we used 150 HU as cut-off point between high andlow sGAG content of cartilage. We used this number based on thepoint where the cumulative histogram of all cartilage ROIs used in

Fig. 2. EPIC-mCT datasets are used as reference for the spatial sGAG distribution of cartilagmaximum radiation dose shown), a mask for high and low sGAG content was created (C, D).mCT images, the number of pixels correctly and incorrectly predicted as having a high and low

the spatial analysis of cartilage reaches 50% (supplementary Fig. 1).Next, both masks for sGAG distribution were used as an overlayfor cartilage on the registered corresponding EPIC-mCT images[Fig. 2(E, F)]. Within the masked EPIC-mCT images, we calculatedthe number of pixels defined as having high or low sGAG content byCTa, using a threshold of 70 gray values for EPIC-mCT. This was againbased on the cumulative histogram of all cartilage ROIs on theEPIC-mCT images (supplementary Fig. 1). Finally, we calculated thenumber of pixels which were incorrectly defined as high or lowquality by CTa by adding the number of incorrectly defined pixels inboth masks, dividing them by the total number of pixels in bothmasks together and then multiplying them by 100 to obtain thepercentage of incorrectly defined pixels [Fig. 2(G, H)].

Statistical analysis

In this study we used ten knees from eight individuals. The useof two knees from one individual could potentially lead to anoverestimation of the correlation between mCT and CTa measure-ments21,22. Exclusion of either one of the knees in the two patientsthat were scanned bilaterally did not influence the results of ourstudy. Therefore, we decided to exclude the bilaterally scannedknees from the analysis.

The correlation between the mean X-ray attenuation values ofCTa and the mean X-ray attenuation values of EPIC-mCT wascalculated per radiation dose for all cartilage ROIs pooled. Because

e (A). Using a fixed X-ray attenuation threshold of 150 HU in all CTa datasets (B, onlyThe masks were used as an overlay of EPIC-mCT cartilage (E, F). Within the masked EPIC-sGAG content by CTa was calculated (G, H). #: high sGAG content. *: low sGAG content.

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685 681

of the fact that we analyzed seven cartilage ROIs per knee joint andthe potential correlation which might already exist within all kneejoints itself, we used a linear mixed model to analyze if the corre-lation coefficients between, CTa outcomes and EPIC-mCT outcomeswere statistically significant.

All analyses were performed using GraphPad (Graphpad Soft-ware Inc., San Diego, USA) and SPSS version 17.0 (SPSS Inc., Chicago,USA). All P-values <0.05 were considered to be statisticallysignificant.

Results

Cadaveric knee joints

After CT scanning, four knees were excluded from the study dueto clearly visible calcifications in the cartilage anddue to the fact thatfrom two individuals two knees were scanned. Thus, a total of sixcadaveric knee joints from six individuals were included in theanalysis (three female, three male; mean age at death 72 years; age

Fig. 3. Correlation plots of mean attenuation from EPIC-mCT and CTa acquired using six difradiation dose (n ¼ 33); C: 40% of the maximum radiation dose (n ¼ 33); D: 30% of the maximthe maximum radiation dose (n ¼ 33). The dashed lines indicate the 95% confidence interv

range at death 30e94 years). Furthermore,12 cartilage ROIswere notincluded in our data analysis because of motion artifacts during mCTscanning and segmentation errors due to severe cartilage loss8.

Correlation of CTa with sGAG content in large anatomical ROIs

Mean X-ray attenuation values of the CTa scans acquired withmaximum radiation correlated strongly with the sGAG content ofcartilage expressed by EPIC-mCT attenuation values (n ¼ 33;R ¼ 0.81; R2 ¼ 0.66; P < 0.0001) [Fig. 3(A)]. In the analysis of theadditional CTa scans with reduced radiation dose, this correlationremained strong when radiation dose was reduced; 50% of themaximum radiation dose (n ¼ 33; R ¼ 0.78; R2 ¼ 0.60; P < 0.0001),40% of the maximum radiation dose (n ¼ 33; R ¼ 0.76; R2 ¼ 0.58;P < 0.0001), 30% of the maximum radiation dose (n ¼ 33; R ¼ 0.76;R2¼ 0.59; P< 0.0001), 20% of themaximum radiation dose (n¼ 33;R¼ 0.77; R2¼ 0.59; P< 0.0001), and 10% of the maximum radiationdose (n ¼ 33; R ¼ 0.78; R2 ¼ 0.61; P < 0.0001) radiation dose perscan was used [Fig. 3(BeF)].

ferent radiation doses. A: maximum radiation dose (n ¼ 33); B: 50% of the maximumum radiation dose (n ¼ 33); E: 20% of the maximum radiation dose (n ¼ 33); F: 10% ofal of the best fit regression line.

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685682

Spatial analysis of cartilage quality

The number of pixels that were incorrectly defined as havinghigh or low sGAG content by CTa was lowest in the CTa scanacquired using the maximum radiation dose (35% � 9%) (Fig. 4).When less radiation was used to obtain CTa, the number of pixelswhich were incorrectly defined as high and low quality cartilageincreased (50% radiation: 37% � 9%, 40% radiation: 38% � 9%, 30%radiation: 38% � 9%, 20% radiation: 39% � 9%, 10% radiation:40% � 9%) (Fig. 4). The effect of this increase in incorrectly definedpixels on the capability of CTa to distinguish between the spatialdistribution of high and low sGAG content of cartilage withina single slice is clearly visible in Fig. 5.

Discussion

Recently, CTa was introduced as a non-destructive method tomeasure cartilage quality in human cadaveric knees6. The main aimof the present cadaver study was to assess whether radiation dosereduction influences the ability of CTa to measure cartilage quality.Lowering the ionizing radiation dose of the acquisition protocol isnecessary to make CTa suitable and acceptable for use in clinicalresearch in humans. The results of this study demonstrate thatmean attenuation values in large anatomical ROIs in CTa acquiredwith different radiation doses are strongly correlated with thesGAG content of articular cartilage measured with EPIC-mCT. Thiscorrelation was similar to previous reported results in cadavers6

and also similar to previously published in vitro results15,18,23.When the radiation dose used to acquire CT scans was decreased,the correlation between CTa X-ray attenuation values and the sGAGcontent of cartilage only slightly decreased, but remained good,even if the radiation dose was reduced to approximately 10% of theoriginal dose. The correlation between X-ray attenuation and thereference test for sGAG content remains relatively good becauseof the fact that the noise in the CT images averages out whencalculating the mean X-ray attenuation values in relatively largecartilage ROIs.

The second aim of this study was to assess the capability of CTato detect local differences in the sGAG content of articular cartilageand to study the effect of radiation dose reduction on this differ-entiation of cartilage quality within a single slice. The ability todetect local differences in cartilage sGAG content could make CTaapplicable as a diagnostic tool for focal cartilage damage instead ofa diagnostic arthroscopy. Additionally, it would enable the use ofCTa as an imaging tool to measure the effect of cartilage repair

Fig. 4. Bar graphs showing the percentage of pixels incorrectly predicted as high andlow sGAG content by the different radiation doses (maximum, 50%, 40%, 30%, 20% and10% of the maximum dose) used in this study. Whiskers show the 95% confidenceinterval of the mean.

therapies (e.g., microfracturing and autologous chondrocyteimplantation24,25) similar to MRI based techniques like dGEM-RIC26,27. Our results demonstrate that, using CTa acquired using themaximum radiation dose, high and low sGAG distribution can beclearly distinguished. An important remark is that the choice of theused thresholds for defining high and low sGAG content within thecartilage based on the pooled cumulative histograms has an arbi-trary component. This might introduce an over or underestimationof the capability of CTa to determine local sGAG differences. Theincrease of noise in the CT image obtained using lowered radiationdoses, however, causes an increased percentage of incorrectlydefined pixels with high and low sGAG content. In the lowestradiation dose used to obtain CTa, the increased noise evenmakes itimpossible to distinguish differences in sGAG distribution fromnoise in the CT images.

Based on the results of this study, we suggest using a CTaprotocol with a low radiation dose if overall cartilage quality is ofinterest in clinical research. The lowest radiation dose we used(CTDIvol of 8.13 mGy per scan) is comparable to the dose ofa regular CT scan of the knee (CTDIvol of approximately 8 mGy12).In addition to cartilage quality, morphological abnormalities canalso be diagnosed using CTa with accuracy comparable toconventional MRI sequences. This was demonstrated in previousresearch by De Filippo et al. and Vande Berg et al.28,29, however, wedid not investigate this in the present study. If the spatial distri-bution of sGAG on a single slice is of interest, we recommendusing a higher radiation dose than for overall cartilage qualitymeasurements since decrease in contrast to noise ratio increasesthe number of incorrectly predicted quality pixels and makes ithard or impossible to differentiate high from low quality cartilageat low radiation dose scans.

Despite the promising results, a limitation of CTawill remain theuse of ionizing radiation, because of the risk of predisposingpatients to the development of certain cancers by using (repetitive)CT scans12. Therefore, MRI based techniques which quantitativelymeasure cartilage quality (e.g., dGEMRIC, Na23 mapping, T2mapping, and T1rho4,5) remain favourable in a clinical researchsetting in large cohorts in humans. However, we think that by usinga relatively low radiation dose protocol, subgroups of patients inwhich CTa is favourable of MRI can be identified (e.g., patients withcontra-indications to undergo MRI). In addition, CT has also someadvantages over MRI (e.g., relative short acquisition time and lowcosts). Therefore, we expect that low radiation dose CTa canbecome a complementary technique to MRI based techniques toquantitatively measure cartilage quality in clinical research. Inaddition to ionizing radiation, other potential limitations of CTawhen applied in humans are: the risk of infection and pain due tothe intra-articular injection with contrast agent, and the risk of an(allergic) reaction to the contrast agent.

Future research using CTa should focus on implementing andvalidating CTa in a clinical research setting in humans in vivo usinga low radiation dose protocol. Filtering the CT data using a low-passimage processing filter will diminish the amount of noise in CTimages and might enable the use of even less radiation than sug-gested in our study. A drawback of using such a filter is, however,the decrease in spatial resolution of the CT images. Another methodto lower the radiation dose is the use of an iterative reconstructionalgorithm30,31 instead of the standard filtered back projectionimage reconstruction algorithm as used in this study. Because ofthe high in plane resolution of CT images acquired with multi-detector CT scanners, future research could also focus on investi-gating the potential of CTa to detect subchondral bone changes andchanges in cartilage quality simultaneously. Recently, the feasibilityof contrast-enhanced peripheral quantitative CT to analyze carti-lage and subchondral bone status on a single scan in vitro was

Fig. 5. Registered images of both EPIC-mCT and CTa acquired using different radiation doses (maximum radiation dose, 50%, 40%, 30%, 20% and 10% of the maximum radiation dose)per scan. The attenuation of cartilage regions is visualized in colour and representative for the sGAG content of the cartilage. High attenuation values represent low sGAG contentand low attenuation values represent high sGAG content.

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685 683

described32 and therefore it is of interest to test this as well asin vivo using CTa.

In conclusion, CTa acquired using a low radiation dose is able tomeasure overall articular cartilage quality throughout the wholehuman knee with a radiation dose comparable to a regular CT scan

of the knee. Spatial sGAG distribution assessment is also possibleusing CTa, however for this purpose a higher radiation dose isnecessary. Nevertheless, due to the reduction in radiation dose, CTamight be implemented as a non-destructive tool to quantitativelymeasure articular cartilage in clinical research.

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685684

Contributions

All authors have substantially contributed to the conception anddesign of the study, acquisition of data, or analysis and interpre-tation of data. All authors have participated in the writing processand approved the final version of the manuscript.

Conflict of interestsAll authors indicate that they have no conflicts of interest.

Role of funding sourceSources of funding are Dutch Arthritis Association (LLP 11) and theSmartMix Program of the Netherlands Ministry of Economic Affairsand the NetherlandsMinistry of Education, culture and Science. Thefunding sources had no role in the study design, collection, analysisor interpretation of data; in the writing of the manuscript or in thedecision to submit the manuscript for publication.

Acknowledgements

We acknowledge the Dutch Arthritis Association (LLP11) and theSmartMix Program of the Netherlands Ministry of Economic Affairsand the Netherlands Ministry of Education, Culture and Science fortheir financial support.

Supplementary material

Supplementary data related to this article can be found online atdoi:10.1016/j.joca.2012.03.007.

References

1. Kellgren JH, Lawrence JS. Radiological assessment of osteo-arthrosis. Ann Rheum Dis 1957;16:494e502.

2. Guermazi A, Roemer FW, Burstein D, Hayashi D. Why radiog-raphy should no longer be considered a surrogate outcomemeasure for longitudinal assessment of cartilage in kneeosteoarthritis. Arthritis Res Ther 2011;13:247.

3. Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG,Jordan JM, et al. Osteoarthritis: new insights. Part 1: thedisease and its risk factors. Ann Intern Med 2000;133:635e46.

4. Crema MD, Roemer FW, Marra MD, Burstein D, Gold GE,Eckstein F, et al. Articular cartilage in the knee: current MRimaging techniques and applications in clinical practice andresearch. Radiographics 2011;31:37e61.

5. Roemer FW, Crema MD, Trattnig S, Guermazi A. Advances inimaging of osteoarthritis and cartilage. Radiology 2011;260:332e54.

6. Siebelt M, van Tiel J, Waarsing JH, Piscaer TM, van Straten M,Booij R, et al. Clinically applied CT arthrography to measure thesulphated glycosaminoglycan content of cartilage. Osteoar-thritis Cartilage 2011;19:1183e9.

7. Piscaer TM, Waarsing JH, Kops N, Pavljasevic P, Verhaar JA, vanOsch GJ, et al. In vivo imaging of cartilage degeneration usingmicroCT-arthrography. Osteoarthritis Cartilage 2008;16:1011e7.

8. Siebelt M, Waarsing JH, Kops N, Piscaer TM, Verhaar JA, Oei EH,et al. Quantifying osteoarthritic cartilage changes accuratelyusing in vivo microCT arthrography in three etiologicallydistinct rat models. J Orthop Res 2011.

9. Bashir A, Gray ML, Hartke J, Burstein D. Nondestructiveimaging of human cartilage glycosaminoglycan concentrationby MRI. Magn Reson Med 1999;41:857e65.

10. Tiderius CJ, Olsson LE, Leander P, Ekberg O, Dahlberg L.Delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) inearly knee osteoarthritis. Magn Reson Med 2003;49:488e92.

11. Gray ML, Burstein D, Kim YJ, Maroudas A, 2007 ElizabethWinston Lanier Award Winner. Magnetic resonance imagingof cartilage glycosaminoglycan: basic principles, imagingtechnique, and clinical applications. J Orthop Res 2008;26:281e91.

12. Biswas D, Bible JE, Bohan M, Simpson AK, Whang PG,Grauer JN. Radiation exposure from musculoskeletal comput-erized tomographic scans. J Bone Joint Surg Am2009;91:1882e9.

13. Eckstein F, Ateshian G, Burgkart R, Burstein D, Cicuttini F,Dardzinski B, et al. Proposal for a nomenclature for magneticresonance imaging based measures of articular cartilage inosteoarthritis. Osteoarthritis Cartilage 2006;14:974e83.

14. Kallioniemi AS, Jurvelin JS, Nieminen MT, Lammi MJ, Toyras J.Contrast agent enhanced pQCT of articular cartilage. Phys MedBiol 2007;52:1209e19.

15. Palmer AW, Guldberg RE, Levenston ME. Analysis of cartilagematrix fixed charge density and three-dimensionalmorphology via contrast-enhanced microcomputed tomog-raphy. Proc Natl Acad Sci U S A 2006;103:19255e60.

16. Xie L, Lin AS, Guldberg RE, Levenston ME. Nondestructiveassessment of sGAG content and distribution in normal anddegraded rat articular cartilage via EPIC-microCT. Osteoar-thritis Cartilage 2010;18:65e72.

17. Silvast TS, Jurvelin JS, Aula AS, Lammi MJ, Toyras J. Contrastagent-enhanced computed tomography of articular cartilage:association with tissue composition and properties. ActaRadiol 2009;50:78e85.

18. Silvast TS, Jurvelin JS, Lammi MJ, Toyras J. pQCT study ondiffusion and equilibrium distribution of iodinated anioniccontrast agent in human articular cartilageeassociations tomatrix composition and integrity. Osteoarthritis Cartilage2009;17:26e32.

19. Silvast TS, Kokkonen HT, Jurvelin JS, Quinn TM, Nieminen MT,Toyras J. Diffusion and near-equilibrium distribution of MRIand CT contrast agents in articular cartilage. Phys Med Biol2009;54:6823e36.

20. Maes F, Collignon A, Vandermeulen D, Marchal G, Suetens P.Multimodality image registration by maximization of mutualinformation. IEEE Trans Med Imaging 1997;16:187e98.

21. Bryant D, Havey TC, Roberts R, Guyatt G. How many patients?How many limbs? Analysis of patients or limbs in the ortho-paedic literature: a systematic review. J Bone Joint Surg Am2006;88:41e5.

22. Park MS, Kim SJ, Chung CY, Choi IH, Lee SH, Lee KM. Statisticalconsideration for bilateral cases in orthopaedic research.J Bone Joint Surg Am 2010;92:1732e7.

23. Bansal PN, Joshi NS, Entezari V, Grinstaff MW, Snyder BD.Contrast enhanced computed tomography can predict theglycosaminoglycan content and biomechanical propertiesof articular cartilage. Osteoarthritis Cartilage 2010;18:184e91.

24. Bedi A, Feeley BT, Williams 3rd RJ. Management of articularcartilage defects of the knee. J Bone Joint Surg Am2010;92:994e1009.

25. Bos PK, van Melle ML, van Osch GJ. Articular cartilage repairand the evolving role of regenerative medicine. Open AccessSurg 2010;3:109e22.

26. Trattnig S, Marlovits S, Gebetsroither S, Szomolanyi P,Welsch GH, Salomonowitz E, et al. Three-dimensional delayedgadolinium-enhanced MRI of cartilage (dGEMRIC) for in vivoevaluation of reparative cartilage after matrix-associated

J. van Tiel et al. / Osteoarthritis and Cartilage 20 (2012) 678e685 685

autologous chondrocyte transplantation at 3.0T: preliminaryresults. J Magn Reson Imaging 2007;26:974e82.

27. Vasiliadis HS, Danielson B, Ljungberg M, McKeon B, Lindahl A,Peterson L. Autologous chondrocyte implantation in cartilagelesions of the knee: long-term evaluation with magnetic reso-nance imaging and delayed gadolinium-enhanced magneticresonance imaging technique. Am J SportsMed 2010;38:943e9.

28. De Filippo M, Bertellini A, Pogliacomi F, Sverzellati N,Corradi D, Garlaschi G, et al. Multidetector computed tomog-raphy arthrography of the knee: diagnostic accuracy andindications. Eur J Radiol 2009;70:342e51.

29. Vande Berg BC, Lecouvet FE, Poilvache P, Maldague B,Malghem J. Spiral CT arthrography of the knee: technique and

value in the assessment of internal derangement of the knee.Eur Radiol 2002;12:1800e10.

30. Gervaise A, Osemont B, Lecocq S, Noel A, Micard E, Felblinger J,et al. CT image quality improvement using adaptive iterativedose reduction with wide-volume acquisition on 320-detectorCT. Eur Radiol 2011.

31. Tamm EP, Rong XJ, Cody DD, Ernst RD, Fitzgerald NE,Kundra V. Quality initiatives: CT radiation dose reduction: howto implement change without sacrificing diagnostic quality.Radiographics 2011.

32. Aula AS, Jurvelin JS, Toyras J. Simultaneous computed tomog-raphy of articular cartilage and subchondral bone. Osteoar-thritis Cartilage 2009;17:1583e8.