Contributions
Abstract: EP1505
Type: Poster Sessions
Abstract Category: Pathology and pathogenesis of MS - MRI and PET
Background: A great deal of imaging research in MS focuses on the segmentation and analysis of lesions. Yet it is known that MS is additionally characterized by diffuse damage in normal-appearing white matter (NAWM) that may precede overt lesion formation. Though these processes can sometimes be observed using summary statistics of voxel intensities, incorporating multiple modalities into these analyses can be difficult, and the appropriate summary statistic for a given situation is often unknown. As such, in this study we introduce a method for the naïve discovery of covariate effects in diffuse processes.
Methods: Brain magnetic resonance imaging (MRI) was performed on patients with relapsing-remitting (RR, n = 32) or secondary-progressive (SP, n = 23) MS. 3D T1-weighted (T1) and T2-weighted fluid-attenuated inversion-recovery (FLAIR) sequences were acquired. We then tested for differences in the joint distribution of T1 and FLAIR intensities in the NAWM across disease subtypes using the novel naïve method, and compared the results to the use of a conventional summary statistic approach.
Results: Using the novel screening method, significant differences were found between the two MS groups in the joint distributions of T1 and FLAIR intensities within NAWM (p = 0.038). Post-hoc analysis revealed that both subtypes showed negative associations between T1 and FLAIR intensities, but voxels more commonly deviated from the linear trend (i.e., were high/low on both T1 and FLAIR) in secondary-progressive patients. Using the correlation between T1 and FLAIR as an outcome measure did not yield a significant difference between subtypes, regardless of correlation method (pPearson = 0.062; pSpearman = 0.107).
Conclusion: This study introduces a novel method for naïvely detecting covariate effects on joint distributions of MRI intensities in NAWM of people with MS. This method detected differences between RR and SP subtypes and indicated that SP patients show less coherence between T1 and FLAIR intensities in NAWM, potentially indicating more damage. Importantly, the most relevant conventional summary statistic - the T1-FLAIR correlation - was unable to capture this difference, suggesting potential benefit of this novel approach when testing for unknown or complex changes in the brain.
Disclosure: Mr. Dworkin has nothing to disclose. Dr. Bakshi has received consulting fees from Bayer, EMD Serono, Genentech, Guerbet, Sanofi-Genzyme, and Shire, as well as research support from EMD Serono and Sanofi-Genzyme. Dr. Shinohara has received consulting fees from Genentech and Roche.
Abstract: EP1505
Type: Poster Sessions
Abstract Category: Pathology and pathogenesis of MS - MRI and PET
Background: A great deal of imaging research in MS focuses on the segmentation and analysis of lesions. Yet it is known that MS is additionally characterized by diffuse damage in normal-appearing white matter (NAWM) that may precede overt lesion formation. Though these processes can sometimes be observed using summary statistics of voxel intensities, incorporating multiple modalities into these analyses can be difficult, and the appropriate summary statistic for a given situation is often unknown. As such, in this study we introduce a method for the naïve discovery of covariate effects in diffuse processes.
Methods: Brain magnetic resonance imaging (MRI) was performed on patients with relapsing-remitting (RR, n = 32) or secondary-progressive (SP, n = 23) MS. 3D T1-weighted (T1) and T2-weighted fluid-attenuated inversion-recovery (FLAIR) sequences were acquired. We then tested for differences in the joint distribution of T1 and FLAIR intensities in the NAWM across disease subtypes using the novel naïve method, and compared the results to the use of a conventional summary statistic approach.
Results: Using the novel screening method, significant differences were found between the two MS groups in the joint distributions of T1 and FLAIR intensities within NAWM (p = 0.038). Post-hoc analysis revealed that both subtypes showed negative associations between T1 and FLAIR intensities, but voxels more commonly deviated from the linear trend (i.e., were high/low on both T1 and FLAIR) in secondary-progressive patients. Using the correlation between T1 and FLAIR as an outcome measure did not yield a significant difference between subtypes, regardless of correlation method (pPearson = 0.062; pSpearman = 0.107).
Conclusion: This study introduces a novel method for naïvely detecting covariate effects on joint distributions of MRI intensities in NAWM of people with MS. This method detected differences between RR and SP subtypes and indicated that SP patients show less coherence between T1 and FLAIR intensities in NAWM, potentially indicating more damage. Importantly, the most relevant conventional summary statistic - the T1-FLAIR correlation - was unable to capture this difference, suggesting potential benefit of this novel approach when testing for unknown or complex changes in the brain.
Disclosure: Mr. Dworkin has nothing to disclose. Dr. Bakshi has received consulting fees from Bayer, EMD Serono, Genentech, Guerbet, Sanofi-Genzyme, and Shire, as well as research support from EMD Serono and Sanofi-Genzyme. Dr. Shinohara has received consulting fees from Genentech and Roche.