Contributions
Abstract: EP1673
Type: Poster Sessions
Abstract Category: Therapy - Tools for detecting therapeutic response
Introduction: Brain atrophy is a promising marker of future disability in multiple sclerosis (MS) and is currently used as an outcome measure in clinical trials. As patients undergo frequent MRI exams, it would be beneficial if clinical scans could be exploited for research purposes. Unfortunately, scanners, protocols, and sequences vary substantially across MS centres and scan quality is adversely affected by time and cost constraints. In particular, we found that the brain is often not fully imaged, e.g. 84% of 76 randomly selected T1-weighted scans acquired on 5 scanners in 3 hospitals in the East Midlands had missing slices. This negatively impacts atrophy estimation approaches, such as SIENA. We previously proposed a SIENA-based method able to deal with missing slices (Pitiot et al. 2017). Here we report results of a preliminary validation study.
Methods: We selected 27 patients (2 CIS, 23 RRMS, 1 PPMS, 1 SPMS) from the Nottingham MS clinic who had at least two T1w scans acquired at different time points, with different protocols and brain coverages. We also simulated cases by cutting a varying number of slices from two complete T1w clinical scans from a healthy control (PSM). In all cases we created two “Frankenstein” scans by combining the potentially incomplete ones with a full scan from either the same patient or a different patient when necessary. We obtained brain atrophy maps from SIENA, visually assessed them for quality control, and estimated annual atrophy rates.
Results: For the real patients, mean age at the time of the first clinical scan was 35.24 years (±8.66) and mean scan interval was 4.78 years (±2.38). PSM scans' were acquired 7 years apart, with the number of missing slices in either or both scans ranging from 0% to 20%. SIENA was unable to process any of the original pairs of scans but could adequately handle 13 Frankenstein patient cases (50%) and all PSM cases (100%). For the patients, the average annual atrophy rate was -0.26%, and -0.19% (±0.03%) for PSM.
Conclusions: We previously proposed an approach enabling SIENA to estimate atrophy in clinical scans with missing slices. The estimated atrophy rates are in keeping with the literature for patients and very tightly grouped for the simulated cases. This suggests great promise for the analysis of abundant clinical data.
Pitiot et al. (2017). Brain atrophy estimation from incompletely acquired clinical MR scans in multiple sclerosis. Poster presented at ECTRIMS-ACTRIMS, Paris.
Disclosure: A. Pitiot: nothing to disclose; M. Clarke: nothing to disclose; P. Morgan: nothing to disclose; N. Evangelou has received funding and support from PCORI, MS society, Biogen, Novartis, Roche,Teva
Abstract: EP1673
Type: Poster Sessions
Abstract Category: Therapy - Tools for detecting therapeutic response
Introduction: Brain atrophy is a promising marker of future disability in multiple sclerosis (MS) and is currently used as an outcome measure in clinical trials. As patients undergo frequent MRI exams, it would be beneficial if clinical scans could be exploited for research purposes. Unfortunately, scanners, protocols, and sequences vary substantially across MS centres and scan quality is adversely affected by time and cost constraints. In particular, we found that the brain is often not fully imaged, e.g. 84% of 76 randomly selected T1-weighted scans acquired on 5 scanners in 3 hospitals in the East Midlands had missing slices. This negatively impacts atrophy estimation approaches, such as SIENA. We previously proposed a SIENA-based method able to deal with missing slices (Pitiot et al. 2017). Here we report results of a preliminary validation study.
Methods: We selected 27 patients (2 CIS, 23 RRMS, 1 PPMS, 1 SPMS) from the Nottingham MS clinic who had at least two T1w scans acquired at different time points, with different protocols and brain coverages. We also simulated cases by cutting a varying number of slices from two complete T1w clinical scans from a healthy control (PSM). In all cases we created two “Frankenstein” scans by combining the potentially incomplete ones with a full scan from either the same patient or a different patient when necessary. We obtained brain atrophy maps from SIENA, visually assessed them for quality control, and estimated annual atrophy rates.
Results: For the real patients, mean age at the time of the first clinical scan was 35.24 years (±8.66) and mean scan interval was 4.78 years (±2.38). PSM scans' were acquired 7 years apart, with the number of missing slices in either or both scans ranging from 0% to 20%. SIENA was unable to process any of the original pairs of scans but could adequately handle 13 Frankenstein patient cases (50%) and all PSM cases (100%). For the patients, the average annual atrophy rate was -0.26%, and -0.19% (±0.03%) for PSM.
Conclusions: We previously proposed an approach enabling SIENA to estimate atrophy in clinical scans with missing slices. The estimated atrophy rates are in keeping with the literature for patients and very tightly grouped for the simulated cases. This suggests great promise for the analysis of abundant clinical data.
Pitiot et al. (2017). Brain atrophy estimation from incompletely acquired clinical MR scans in multiple sclerosis. Poster presented at ECTRIMS-ACTRIMS, Paris.
Disclosure: A. Pitiot: nothing to disclose; M. Clarke: nothing to disclose; P. Morgan: nothing to disclose; N. Evangelou has received funding and support from PCORI, MS society, Biogen, Novartis, Roche,Teva