Research Presentation Session: Musculoskeletal
RPS 210 - Exploring knee imaging
February 28, 09:30 - 11:00 CET
7 min
Assessing lateral posterior meniscal root insertion variations of the knee on a clinical 7 Tesla MRI in an asymptomatic cohort
Roy P. Marcus, Zurich / Switzerland
- Author Block: R. P. Marcus, A. A. Marth, S. Zimmermann, R. Sutter; Zurich/CHPurpose: The aim of this study was to assess variations of posterolateral meniscal root insertions using a high-resolution sequence on a clinical 7 Tesla (T) MRI scanner.Methods or Background: Fifty-eight knees of healthy subjects were examined on a clinical 7 T MRI scanner (Magnetom Terra, Siemens Healthineers). Multiplanar reconstructions of a sagittal isotropic
- 24 mm double echo steady state 3D-sequence were used. The number of root insertion reins and insertion localisations was recorded for the lateral meniscus by two radiologists. Results or Findings: The majority of subjects presented with double root insertion reins (
- 7%), followed by single (31.6%) and triple insertion reins (1.8%). The majority of single posterior lateral rein roots inserted into the intercondylar area (77.8%) and the remaining 22.2% inserted into the posterior slope of the posteromedial tubercle of the tibia. The majority of double reins inserted with a major rein along the intertubercular area to the posteromedial insertion of the ACL and a minor rein to the lateral tubercle (78.9%). The inverted variant (major rein inserts into the lateral tubercle and the minor rein inserts next to the tibial ACL-insertion) was visible in 5.3%. The remaining double reins inserted into the posterolateral slope of the medial and lateral eminence. Interrater reliability for describing the meniscal root reins was perfect (κ = 1) and identifying the insertion sites was strong (κ = 0.808). Conclusion: Posterolateral meniscal roots consisted mostly of double root insertion reins, followed by single and triple reins. Five insertion locations were detected, with the majority inserting with a major rein next to the posteromedial insertion of the ACL and a minor rein to the lateral tubercle of the tibia.Limitations: Small sample size.Funding for this study: This was an in-house study and no funding was received for it.Has your study been approved by an ethics committee? YesEthics committee - additional information: This study was approved by the Local Ethics committee of the Canton of Zurich.
7 min
Stability of deep learning-based image quality improvement in MRI of the knee: correlation with arthroscopy
Seokhee Park, Seoul / Korea, Republic of
- Author Block: S. Park, S. H. Choi, H. Lee, S. Kim, J-Y. Jung; Seoul/KRPurpose: The study aimed to validate whether deep learning (DL)-based image-quality improvement changes the diagnostic performance and visibility of core features for meniscal, cartilage, and ligament knee lesions.Methods or Background: This retrospective study include 106 patients, who underwent knee MRI and arthroscopy within a three-month interval. Fat-suppressed 2D fast spin-echo (FSE) were processed with pre-trained DL. Two musculoskeletal radiologists independently reviewed both the original and DL-processed MRI for medial meniscus (MM), lateral meniscus (LM), articular cartilage (AC) and cruciate ligaments (CL). Sensitivity and specificity were compared with arthroscopic results used as the reference standard. Additionally, two radiologists compared the target lesions between arthroscopic images and original or DL-processed MRI with different denoising levels using pe-defined similarity score.Results or Findings: In reader 1, the sensitivity and specificity (%) of original vs. DL-IQI-processed MRI were
- 0 vs. 92.5 (P>0.99) and 92.3 vs. 97.4 (P=0.50) for MM, 88.9 vs. 93.3 (P=0.50) and 90.1 vs. 91.8 (P>0.99) for LM, 87.2 vs. 94.8 (*P<0.05) and 92.8 vs. 85.7 (P=0.625) for AC, and 91.3 vs. 91.3 (Non-Applicable, NA) and 98.8 vs. 98.8 (NA) for CL. In reader 2, the sensitivity and specificity of original vs. DL-IQI-processed MRI were 86.6 vs. 91.4 (P=0.25) and 97.4 vs. 97.4 (NA) for MM, 86.7 vs. 86.7 (P>0.99) and 88.5 vs. 88.5 (P>0.99) for LM, 80.8 vs. 84.6 (P=0.37) and 96.4 vs. 89.3 (P=0.5) for AC, and 78.3 vs. 78.3 (NA) and 98.8 vs. 98.8 (NA) for CL. The mean agreement scores between arthroscopy and MRI for target lesions were 7.18(O) and 7.33-7.36(DL) for meniscal tears (n=130), and 2.53(O) and 3.23-3.24(DL) for cartilage lesions (n=84). Conclusion: DL-processing of knee MRIs improved the identification of cartilage lesions, without affecting overall diagnostic performances as correlated with arthroscopic results.Limitations: This was a retrospective, single-centre study.Funding for this study: No funding was obtained for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: This study was approved by the Institutional review board in our hospital: 20230822-F-
7 min
M score as an alternative to T score in the evaluation of femoral osteoporosis?
Muhammet Batuhan Gökhan, Ankara / Turkey
- Author Block: M. B. Gökhan, N. Çay; Ankara/TRPurpose: This study aimed to contribute to the diagnosis of osteoporosis-osteopenia with MRI-based “M-scores” obtained from T1 weighted hip MRI.Methods or Background: Between February 2019 and May 2023, patients aged 18 years or older, who underwent DXA and also had hip MRI within six months before or after DXA, were included. A reference group consisting of 190 females with a normal BMI who had hip MRI and aged between 20 and 29 years was also included. Images were retrospectively evaluated by two radiologists, blinded to the DXA scores. Signal intensities of locations corresponding to DXA measurement sites, subcutaneous fat intensities, muscle intensities, and noise from artifact-free areas, were measured. Signal-to-noise(SNR), signal-to-fat (SFR) and signal-to-muscle ratios(SMR) were calculated. M scores for SNR, SMR and SFR were calculated based on the measurement values of the reference group, similar to the T score. Correlation and ROC analysis were performed for all M-scores .Results or Findings: Included 330 patients (mean age
- 61); 30(%9.1) osteoporotic for hip at DXA. There were a strong negative correlation between M-scores and T-scores (SNR-M-r=0.722, SFR-M-r=0.750, SMR-M-r=0.662) for neck and moderate negative correlation for total hip (SNR-M-r=0.599, SFR-M-r=0.625, SMR-M-r=0.568). Cut-off points for femoral neck osteoporosis detection were determined as follows: SNR-M-score at 2.23 (sensitivity 93%, specificity 71%), SFR-M-score at 3.052 (sensitivity 84%, specificity 85%), and SMR-score at 1.538 (sensitivity 100%, specificity 78%). And cut-off points for hip were determined as following: SNR-M-score at 1.22 (sensitivity 85%, specificity 68%), SFR-M-score at 1.313 (sensitivity 75%, specificity 82%), and SMR-M-score at 1.576 (sensitivity 67%, specificity 83%). Conclusion: Our study suggests the potential of MRI-based M scores for diagnosing hip osteoporosis, but further validation on a larger population is needed.Limitations: The limitations of the study is sample size.Funding for this study: No funding was received for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: The study was approved by Ankara Bilkent City Hospital Ethics Committee: E2-23-
7 min
Extension and costs of low-value MRI of the knee in Norway
Bjørn Morten Hofmann, Gjøvik / Norway
- Author Block: B. M. Hofmann1, V. Håvik2, I. Ø. Brandsæter1, E. R. Andersen1, E. Kjelle1; 1Gjøvik/NO, 2Oslo/NOPurpose: As low-value imaging is a challenge for the quality of care and patient safety, we need more evidence about the extension and costs of low-value imaging. Accordingly, the purpose of this study is to provide knowledge about the extension of knee MRIs that are of low value and to estimate their opportunity costs.Methods or Background: Contingent extraction of imaging and diagnostic data from the control and reimbursement of healthcare claims (Helfo) registry in Norway in 2021 analysed with descriptive statistics.Results or Findings: Between 60 and 80 per cent of knee MRIs in Norway could be of low value, as they yield no relevant diagnoses or actions prior to or after the MRI examination. This corresponds to between 24,000 and 35,000 MRIs, at a cost of USD 8 to 12 million annually. Fewer than 7% of the patients had a relevant diagnosis code or a prior imaging examination when having the MRI, and less than 15% got a knee-related diagnosis code from specialist care within six months after the MRI.Conclusion: Most MRIs of the knee in Norway could be of low value to the patient. Reducing low-value MRIs can free resources for high-value imaging, reduce wait times, increase the quality of care, in addition to improve patient safety, healthcare efficiency, and professional integrity.Limitations: Quality of registry data, inclusiveness of disease and treatment codes, investigative time span.Funding for this study: This study was funded by the Research Council of Norway: Project number Has your study been approved by an ethics committee? Not applicableEthics committee - additional information: This is a register study where no patient data are used outside data provider organization and so no ethical approval was sought for this study.
7 min
Investigation of the anatomic risk factors in acute anterior cruciate ligament rupture coexisted with ramp lesions of the medial meniscus by quantitative MRI
Ziyi Tang, Chongqing / China
- Author Block: Z. Tang, Y. Luo, D. Liu, S. Zhou, H. Yang; Chongqing, China/CNPurpose: This study aimed to investigate the predictive anatomic risk factors of knees in patients with acute non-contact anterior cruciate ligament (aACL) rupture that coexisted with ramp lesions of medial meniscus.Methods or Background: A total of 202 subjects were retrospectively divided into three groups: (1) an aACL rupture with ramp lesions group (n=76); (2) an isolated ACL rupture group (n=56) and (3) a normal control group (n=70). Quantitative morphological parameters on MRI were measured: the diameter of medial femoral condyle (MFC), the anterior-posterior length and depth of medial tibial plateau (MTP AP length and MTP depth), the lateral posterior tibial slope (LPTS) and medial posterior tibial slope (MTPS), the asymmetry of LPTS and MPTS (LMPTS), the lateral meniscal slope (LMS), and medial meniscal slope (MMS).Results or Findings: The MTP AP length, MTP AP length/MFC diameter ratio, MTP depth, LPTS and the asymmetry of LMPTS showed significant differences among the three groups (p<
- 001). The risk factors associated with the ramp lesions including a longer MTP AP length (OR 1.17, 95% CI 1.00-1.44, p=0.044), deeper MTP depth (OR 1.91, 95% CI 1.22-3.00, p=0.005) and larger MTP AP length/MFC diameter ratio (OR 1.11, 95% CI 1.01-1.22, p = 0.036). The highest AUC was the MTP AP length/MFC diameter ratio (0.74; 95% CI, 0.66-0.82), with a sensitivity of 73% and specificity of 68% to predict ramp lesions. Conclusion: Several occult anatomic characteristics of the knee, especially the morphology of the medial tibia plateau, can be additional risk factors for aACL rupture that coexisted with ramp lesions.Limitations: Not all cases were diagnosed with ramp lesions under arthroscopy. However, the ramp lesions exhibited the typical MRI characteristics with specificity as high as 98% in previous studies.Funding for this study: No funding was received for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: The study was approved by The First Affiliated Hospital of Chongqing Medical University (No.2021-203).
7 min
Leveraging subtraction STIR-T1 MAVRIC sequences for precise discrimination of artefacts vs peri-implant infections in total knee replacement prostheses
Sriram Rajan, New Delhi / India
- Author Block: S. Rajan, V. K. Venugopal, H. Mahajan; New Delhi/INPurpose: This study aimed to explore the utility of a new subtraction technique which involved subtracting STIR from T1 images to unveil the persisting infective signals on STIR images while reducing the T1 artefacts.Methods or Background: Our study encompassed 80 patients suspected of peri-implant infections surrounding total knee replacement prostheses, all of whom underwent MRI scans with the standard T1 and STIR MAVRIC protocol. Subsequently, the anonymised and randomised images were evaluated by two radiologists, possessing 8 and 21 years of experience, who arrived at a consensus regarding the presence or absence of peri-implant infections or marked them as indeterminate cases. Further, subtraction images, derived from STIR minus T1 weighted images, were meticulously reconstructed and colour-coded (hot-iron) before being fused with conventional T1 images. These fused images underwent a second evaluation by consensus to ascertain the presence or absence of peri-implant infections or indeterminate cases.Results or Findings: The initial assessment revealed 38 cases detected with infection, 25 cases absent of infection, and 17 cases marked as indeterminate. However, upon implementing the subtraction technique, the numbers shifted to 34 cases with infection, 32 cases without infection, and 14 indeterminate cases. To ascertain the statistical significance of these results, a chi-square goodness-of-fit test was conducted, revealing a Chi-square value of
- 66. This value signifies a significant difference between the observed and expected frequencies, further affirming the utility of the subtraction technique. Conclusion: Our study underscores the substantial advantages of employing subtraction STIR-T1 MAVRIC sequences in distinguishing between infections and artefacts, outperforming conventional sequences.Limitations: This study was limited by the small size of the datasetFunding for this study: No funding was received for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: This study was IRB approved.
7 min
Deep learning-based MRI quantification of the thigh muscles in sarcopenia
Zaizhu Zhang, Beijing / China
- Author Block: Z. Zhang1, W. Yu1, F. Santini2, J. Zhu1, F. Wang1; 1Beijing/CN, 2Basel/CHPurpose: This study aimed to explore whether MRI markers of cross-sectional volume (CSV), fat fraction (FF) and T2 values are affected in sarcopenia and to examine associations between MRI markers and components of sarcopenia (muscle mass, handgrip strength, and gait speed) in Chinese subjects.Methods or Background: Postmenopausal women were prospectively recruited for this cross-sectional study. Muscle mass was measured by dual-energy X-ray absorptiometry (DXA). Jamar hydraulic hand dynamometer was used to assess muscle strength and gait speed was used to assess physical performance. A
- 0T MRI machine was used to scan bilateral thigh muscles. A DL-based automated segmentation software (Deep Anatomical Federated Network, DAFNE) was used for the segmentation of anatomical images. MRI measures were compared between sarcopenic and non-sarcopenic groups and correlations with other variables were assessed. Their optimal cut-off points (Youden index) for sarcopenia diagnosis were identified in relation to the AWGS2 criteria. Results or Findings: 68 participants (mean age,
- 6 years ± 5.7 [SD]) were studied, and 25 (36.8%) were diagnosed as sarcopenic. The sarcopenic group had lower CSV (737.8 cm3 vs 862.2 cm3; P<0.001) and elevated T2 values (45.8 ms vs 44.4 mol/L; P=0.018) than the non-sarcopenic group, although there was no evidence of a difference in FF. CSV was positively associated with DXA-derived muscle mass and handgrip strength (P<0.001), whereas T2 values were inversely associated with gait speed (P<0.05). The CSV and T2 value optimal cut-off points in discriminating sarcopenic patients were identified at 798.7 cm3 (AUC-ROC = 0.883) and 44.8 ms (AUC-ROC = 0.666) respectively. Conclusion: CSV and T2 values can differentiate sarcopenic versus non-sarcopenic subjects, representing an imaging biomarker of sarcopenia.Limitations: This preliminary study was limited in size, and therefore could not account for all confounders in the sarcopenic group.Funding for this study: No funding was received for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: This study was approved by the institutional review board of the Peking Union Medical College Hospital, Beijing, China.
7 min
Five-minute knee MRI: an AI-based super resolution reconstruction approach for compressed sensing. A validation study on healthy volunteers
Robert Angelo Terzis, Cologne / Germany
- Author Block: R. A. Terzis, T. M. Dratsch, R. Hahnfeldt, L. Basten, P. Rauen, K. Sonnabend, D. Maintz, A-I. Iuga, G. Bratke; Cologne/DEPurpose: This study aimed to investigate the feasibility of a complete five-minute 2D knee MRI protocol obtained by combining Compressed Sensing (CS) and a new AI-aided super-resolution reconstruction method.Methods or Background: 20 volunteers were examined on a 3T-MRI scanner (Ingenia Elition X, Philips) using a 2D knee protocol similar to clinical standards: a sagittal fat-saturated proton-density-, and a T1-weighted- sequence in coronal, sagittal and transversal orientation. Images were acquired in standard and low resolution and reconstructed with a traditional CS and a new, AI-aided technique (CS-SuperRes). Two experienced radiologists assessed subjective image quality using 8 criteria on a 5-point scale, and signal-to-noise ratio was measured as an objective parameter.Results or Findings: CS-SuperRes showed superior ratings over equivalent CS reconstructions, particularly for low-resolution images (overall image quality:
- 3±0.4 vs. 3.4±0.4, p<0.05). For the low-resolution scans, CS-SuperRes reconstructions matched the quality of standard-resolution traditional CS across all metrics. Overall image impression scored 4.3±0.4 with CS-SuperRes vs. 4.0±0.5 with traditional CS (p<0.05). This allowed a significant decrease in scan duration without image quality loss, from 11:01 minutes to just 4:46 minutes (a 57% reduction) for the entire protocol. Conclusion: The AI-aided CS-SuperRes reconstruction method can reduce MRI scan times by 57%, retaining image quality comparable to traditional CS. This allows for a complete 2D knee MRI scan in just 5 minutes with preserved image excellence, facilitating increased scanner throughput and improved patient workflow.Limitations: This was a single-centre study which used a single MRI scanner and a small sample of healthy volunteers. Nevertheless, the multiscale network ensures data consistency and objective findings corroborate subjective interpretations. Moreover, some of the scans showed incidental findings, which were visible with all reconstruction methods.Funding for this study: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.Has your study been approved by an ethics committee? YesEthics committee - additional information: This prospective single-centre study was carried out in accordance with the ethical standards in the 1964 Declaration of Helsinki and its later amendments and was approved by the institutional review board.
7 min
dGEMRIC evaluation of autologous conditioned adipose tissue with leukocyte-poor platelet rich plasma for treatment of mild knee osteoarthritis
Vid Matišić, Zagreb / Croatia
- Author Block: V. Matišić, V. Molnar, E. Pavelić, P. Brlek, Ž. Jeleč, E. Rod, I. Borić, D. Primorac; Zagreb/HRPurpose: This study was conducted to assess the clinical and radiomorphologic effects of autologous conditioned adipose tissue (ACA) with leukocyte-poor, platelet-rich (LP-PRP) plasma knee injections in treating patients with mild knee osteoarthritis (KOA). The main objective of this study was to compare the results of the treatment to the control group treated with the standard hyaluronic acid injection.Methods or Background: 16 patients (8 male and 8 female) with KOA were included in this study. Patients received an intraarticular, ultrasound-guided injection of ACA and LP-PRP. Contralateral knees of the same patients were treated with hyaluronic acid (HA) to compare the dGEMRIC indices with knees with KOA. Clinical results and dGEMRIC scores were noted by filling out questionnaires (VAS, WOMAC, KOOS) and imaging at 3-time points occurring prior to intervention and 3 and 6 months after the intervention. Each subject received
- 2 mmol/kg gadolinium diethylene triamine penta-acetic acid for the purpose of dGEMRIC per visit. Results or Findings: Clinical scores demonstrated improvement at both 3 and 6-month follow-ups. However, dGEMRIC indices showed no significant change at both time points. Furthermore, no difference between the affected and the unaffected knees treated with HA was demonstrated.Conclusion: The lack of objective findings after ortho-biologic treatment is a common finding across the published literature. Hence, the major orthopaedic guidelines do not include or advise against the use of these treatment modalities. Further research should standardise the dose of given mesenchymal stem cells and PRP products to try and produce higher-quality evidence for their use and identify the best patients to receive these treatments.Limitations: This study was limited by the small patient size and the undetermined amount of MSCs given to any of the patients.Funding for this study: No funding was received for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: This study was approved by the Ethics Committee of St. Catherine Specialty Hospital (protocol code 21/3-1, 9 March 2021)
7 min
Performance of low-field MR for knee cartilage quantification: results on comparison with arthroscopy
Francesco Pucciarelli, Rome / Italy
- Author Block: F. Pucciarelli, M. Zerunian, M. Polici, B. Masci, A. Del Gaudio, G. Argento, D. De Santis, D. Caruso, A. Laghi; Rome/ITPurpose: This study aimed to evaluate the reliability of T2-mapping on knee cartilage on low-field (
- 3 T) MRI to detect low-grade chondropathy and to compare it to arthroscopy. Methods or Background: 55 patients with planned arthroscopic surgery were prospectively enrolled. MRI examination was performed with a low-field scanner (
- 3 T) with a dedicated sequence (3D SHARC) for the evaluation of cartilage T2 relaxation time. Image analysis was performed by two radiologists with dedicated experimental software specifically set on MatLab. A colour map, superimposed on the cartilage, consisted of different colours reflecting different T2 values. Each knee was divided into 14 regions, according to the Whole-Organ Magnetic Resonance Imaging Score (WORMS) classification, and a single free-hand region of interest (ROI) was drawn in each region. Arthroscopic knee surgery was performed by an orthopaedic surgeon blinded to MRI data and chondropathy grade was assessed by Outerbridge classification. The T2-relaxation time of each ROI was annotated and compared to arthroscopy results. Results or Findings: A total of 52 participants were included and a total of 728 ROIs were analysed. Comparison of negative patients (grade 0) and positive patients (grade I, II, III and IV) showed significant (p<
- 001) lower T2-values for grade 0 (42,35±11,67 ms vs 48,36±12,36 ms, respectively). Sub-analysis of different grades showed significant (p<0.001) lower T2-values between grade 0 and I (42,35±11,67 ms vs 46,38±9,09 ms, respectively) and grade 0 and II (42,35±11,67 ms vs 47,42±10,51 ms, respectively). No significant difference (p=0.089) was found between grades I and II (46,38±9,09 ms vs 47,42±10,51 ms, respectively). No significant difference (all p>0.05) was found between the two readers. Conclusion: T2 mapping applied to low-field magnetic resonance allows discrimination between negative and positive patients with chondropathy and allows the detection of patients with low degrees of chondropathy.Limitations: This was a single-centre study with a small sample size.Funding for this study: No funding was received for this study.Has your study been approved by an ethics committee? YesEthics committee - additional information: Written informed consent was acquired for all patients and Institutional Review Board approval was obtained.
7 min
To evaluate the impact of cartilage damage on knee joint functionality through dynamic CT imaging
Manou Acke, Elsene / Belgium
- Author Block: M. Acke, T. Scheerlinck, B. Keelson, N. Van Vlasselaer, S. Héréus, G. Van Gompel, E. Cattrysse, N. Buls; Brussels/BEPurpose: This study aimed to evaluate the impact of cartilage damage on knee joint functionality through dynamic CT imaging.Methods or Background: In an ex-vivo thiel embalmed leg, cartilage was removed progressively from the patellofemoral joint in three stages: lateral facet, crista and medial facet. After each stage, dynamic CT scans were acquired while inducing a cyclic flexion-extension motion of the leg. Dynamic CT scans were acquired in cine mode with 16 cm z-coverage, 50 cm FOV, 80 kVp, 452 mA, 280 ms rotation time and 6 s total scan duration. Bones were segmented and registered through a multi-atlas approach using SimpleITK-Elastix. Surface proximity maps of the joint were created to visualise the intra-articular joint space. The intra-articular contact area was computed as the area with a distance to the adjacent bone smaller than the cartilage thickness (5,7 mm). Differences in contact areas between normal and damaged stages were evaluated using a Wilcoxon Signed Rank test.Results or Findings: The contact area after maximum intervention increased from 241 mm² to 613 mm² for extension, and from 67 mm² to 480 mm² for maximum flexion (p=
- 014). Apart from the first stage, each intervention resulted in an increased intra-articular contact area (p=0.02 – 0.014). Conclusion: Dynamic CT scans can indicate an increasing trend of intra-articular contact area with escalating cartilage degeneration.Limitations: The feasibility was shown for a single, ex-vivo sample.Funding for this study: Funding for this study was provided by the UZBrussels.Has your study been approved by an ethics committee? YesEthics committee - additional information: Prior to commencing this study, ethical approval (B.U.N. 1432023000100) was obtained from the VUB-UZ Brussels University Hospital’s ethical committee after submitting the necessary documentation. The cadaver used in the study was legally donated, and either the subject or their legal next of kin explicitly consented to the use for research purposes.
This session will not be streamed, nor will it be available on-demand!
This session is accredited with 1.5 CME credits.
Reto Sutter
Zürich / Switzerland
Roy P. Marcus
Zurich / SwitzerlandSeokhee Park
Seoul / Korea, Republic ofMuhammet Batuhan Gökhan
Ankara / TurkeyBjørn Morten Hofmann
Gjøvik / NorwayZiyi Tang
Chongqing / ChinaSriram Rajan
New Delhi / IndiaZaizhu Zhang
Beijing / ChinaRobert Angelo Terzis
Cologne / GermanyVid Matišić
Zagreb / CroatiaFrancesco Pucciarelli
Rome / Italy