Contributions
Abstract: EP1508
Type: Poster Sessions
Abstract Category: Pathology and pathogenesis of MS - MRI and PET
Background: Cervical cord atrophy occurs in MS and is associated with disease severity (1-3).
Aim: To assess within-scanner reproducibility, between-scanner agreement, and sensitivity to MS lesions, for four software methods to measure mean upper cervical cord area (MUCCA): SCT-PropSeg, SCT-DeepSeg, NeuroQLab and ITK-Snap.
Methods: MUCCA was always assessed in the same manually selected 30-mm slice range starting at C1, to ensure comparability between results. Two archive datasets were used. Same-day scan-rescan 3DT1-weighted head imaging (head coil, including cervical spine as much as possible) performed on three 3-Tesla MR scanners (GE MR750, Philips Ingenuity, Toshiba Vantage Titan) in 21 MS patients and 6 healthy volunteers was used to assess within-scanner reproducibility and between-scanner agreement by intra-class correlation coefficient (ICC) and repeated measures ANOVA. To investigate lesion effects, we selected from a larger cohort (3T GE HDxT, head-neck-spine coil) archive cervical 3D T1-weighted images of 7 MS patients without cervical MS lesions and 14 MS patients with at least 12 cervical MS lesions, balanced on sex, disease types, and previously measured MUCCA. ICC and Dice overlap coefficient (DOC) compared to manual reference segmentations were assessed.
Results: MUCCA had high within-scanner reproducibility on all scanners for PropSeg, DeepSeg and NeuroQLab (ICC for absolute agreement all >0.98) and ITK-Snap (0.90-0.95). Systematic differences were observed between scanners and especially between software methods (interaction method*scanner p=0.002); PropSeg and DeepSeg produced lowest MUCCA values. ICC for consistency with manual MUCCA were largely similar for patients with and without lesions (PropSeg: 0.57 vs. 0.71; DeepSeg: 0.89 vs. 0.95; ITK: 0.88 vs. 0.69; NeuroQLab: 0.93 vs. 0.95). Somewhat unexpectedly, DOC were slightly higher for patients with many lesions compared to patients without lesions for PropSeg (0.81 vs. 0.79), DeepSeg (0.83 vs. 0.81) and ITK (0.96 vs. 0.94); DOC could not be assessed for NeuroQLab because it does not produce a binary segmentation.
Conclusion: MUCCA was highly reproducible within-scanner, was not systematically affected by lesions, but varied between scanners and software. Results should be confirmed in multi-center cervical scans and at different vertebral levels.
1. Kearney, Nat Rev Neurol 2015;11:327-38.
2. Lukas, Radiology 2013;269:542-52.
3. Rocca, Neurology 2011;76:2096-102.
Disclosure: Sander Middelkoop: nothing to disclose.
Merlin M. Weeda: nothing to disclose.
Martijn D. Steenwijk: nothing to disclose.
Marita Daams: nothing to disclose.
Houshang Amiri was sponsored by a research grant from Novartis Pharma.
Iman Brouwer was sponsored by research grants provided by Novartis and Teva.
Joep Killestein has accepted speaker and consultancy fees from Merck, Biogen, Teva, Genzyme, Roche and Novartis.
Bernard Uitdehaag reports personal fees from Genzyme, Biogen Idec, TEVA, Merck Serono, Roche, outside the submitted work.
Iris Dekker received speaking honoraria from Roche and receives funding from the Dutch MS Research Foundation, grant number 14-358e.
Carsten Lukas received a research grant by the German Federal Ministry for
Education and Research, BMBF, German Competence Network Multiple Sclerosis (KKNMS),
grant no.01GI1601I, has received consulting and speaker´s honoraria from Biogen Idec, Bayer
Schering, Daiichi Sanykyo, Merck Serono, Novartis, Sanofi, Genzyme and TEVA.
Barbara Bellenberg received financial support by the German Federal Ministry for Education and
Research, BMBF, German Competence Network Multiple Sclerosis (KKNMS), grant
no.01GI1601I.
Frederik Barkhof is supported by the NIHR UCLH biomedical research centre, serves as editorial board member of Brain, European Radiology, Neurology, Multiple Sclerosis Journal and Radiology and has accepted consulting fees from Bayer-Schering Pharma, Biogen-IDEC, TEVA, Merck-Serono, Novartis, Roche, Jansen Research, Genzyme-Sanofi, IXICO Ltd, GeNeuro, Apitope Ltd and speaker fees from Biogen-IDEC and IXICO.
P.J.W. Pouwels receives research support from the Dutch MS Research Foundation, grant number 14-876.
Hugo Vrenken has received research grants from Novartis, Teva and MerckSerono, and consulting fees from MerckSerono; all funds were paid directly to his institution.
Abstract: EP1508
Type: Poster Sessions
Abstract Category: Pathology and pathogenesis of MS - MRI and PET
Background: Cervical cord atrophy occurs in MS and is associated with disease severity (1-3).
Aim: To assess within-scanner reproducibility, between-scanner agreement, and sensitivity to MS lesions, for four software methods to measure mean upper cervical cord area (MUCCA): SCT-PropSeg, SCT-DeepSeg, NeuroQLab and ITK-Snap.
Methods: MUCCA was always assessed in the same manually selected 30-mm slice range starting at C1, to ensure comparability between results. Two archive datasets were used. Same-day scan-rescan 3DT1-weighted head imaging (head coil, including cervical spine as much as possible) performed on three 3-Tesla MR scanners (GE MR750, Philips Ingenuity, Toshiba Vantage Titan) in 21 MS patients and 6 healthy volunteers was used to assess within-scanner reproducibility and between-scanner agreement by intra-class correlation coefficient (ICC) and repeated measures ANOVA. To investigate lesion effects, we selected from a larger cohort (3T GE HDxT, head-neck-spine coil) archive cervical 3D T1-weighted images of 7 MS patients without cervical MS lesions and 14 MS patients with at least 12 cervical MS lesions, balanced on sex, disease types, and previously measured MUCCA. ICC and Dice overlap coefficient (DOC) compared to manual reference segmentations were assessed.
Results: MUCCA had high within-scanner reproducibility on all scanners for PropSeg, DeepSeg and NeuroQLab (ICC for absolute agreement all >0.98) and ITK-Snap (0.90-0.95). Systematic differences were observed between scanners and especially between software methods (interaction method*scanner p=0.002); PropSeg and DeepSeg produced lowest MUCCA values. ICC for consistency with manual MUCCA were largely similar for patients with and without lesions (PropSeg: 0.57 vs. 0.71; DeepSeg: 0.89 vs. 0.95; ITK: 0.88 vs. 0.69; NeuroQLab: 0.93 vs. 0.95). Somewhat unexpectedly, DOC were slightly higher for patients with many lesions compared to patients without lesions for PropSeg (0.81 vs. 0.79), DeepSeg (0.83 vs. 0.81) and ITK (0.96 vs. 0.94); DOC could not be assessed for NeuroQLab because it does not produce a binary segmentation.
Conclusion: MUCCA was highly reproducible within-scanner, was not systematically affected by lesions, but varied between scanners and software. Results should be confirmed in multi-center cervical scans and at different vertebral levels.
1. Kearney, Nat Rev Neurol 2015;11:327-38.
2. Lukas, Radiology 2013;269:542-52.
3. Rocca, Neurology 2011;76:2096-102.
Disclosure: Sander Middelkoop: nothing to disclose.
Merlin M. Weeda: nothing to disclose.
Martijn D. Steenwijk: nothing to disclose.
Marita Daams: nothing to disclose.
Houshang Amiri was sponsored by a research grant from Novartis Pharma.
Iman Brouwer was sponsored by research grants provided by Novartis and Teva.
Joep Killestein has accepted speaker and consultancy fees from Merck, Biogen, Teva, Genzyme, Roche and Novartis.
Bernard Uitdehaag reports personal fees from Genzyme, Biogen Idec, TEVA, Merck Serono, Roche, outside the submitted work.
Iris Dekker received speaking honoraria from Roche and receives funding from the Dutch MS Research Foundation, grant number 14-358e.
Carsten Lukas received a research grant by the German Federal Ministry for
Education and Research, BMBF, German Competence Network Multiple Sclerosis (KKNMS),
grant no.01GI1601I, has received consulting and speaker´s honoraria from Biogen Idec, Bayer
Schering, Daiichi Sanykyo, Merck Serono, Novartis, Sanofi, Genzyme and TEVA.
Barbara Bellenberg received financial support by the German Federal Ministry for Education and
Research, BMBF, German Competence Network Multiple Sclerosis (KKNMS), grant
no.01GI1601I.
Frederik Barkhof is supported by the NIHR UCLH biomedical research centre, serves as editorial board member of Brain, European Radiology, Neurology, Multiple Sclerosis Journal and Radiology and has accepted consulting fees from Bayer-Schering Pharma, Biogen-IDEC, TEVA, Merck-Serono, Novartis, Roche, Jansen Research, Genzyme-Sanofi, IXICO Ltd, GeNeuro, Apitope Ltd and speaker fees from Biogen-IDEC and IXICO.
P.J.W. Pouwels receives research support from the Dutch MS Research Foundation, grant number 14-876.
Hugo Vrenken has received research grants from Novartis, Teva and MerckSerono, and consulting fees from MerckSerono; all funds were paid directly to his institution.