Contributions
Abstract: EP1456
Type: ePoster
Abstract Category: Pathology and pathogenesis of MS - Imaging
Background: Thalamic volume loss may be an important factor in MS-related clinical impairment as thalamic pathology occurs in early stages of MS. Thalamic volume changes may predict long-term accumulation of disability and may correlate with cognitive impairment. However, the thalamus poses challenges for anatomical segmentation. We found significant segmentation errors and biases in the thalamic volumes obtained through automatic segmentation using FreeSurfer. We investigated correlations between cognitive measures and thalamic volumes obtained through both manual and automatic segmentation.
Objective: To investigate the correlation of lateralized and combined thalamic volumes with frequently-affected cognitive domains of memory and information processing speed in MS patients.
Methods: 73 MS patients (RRMS, PPMS, SPMS) and 35 healthy controls were studied. Cognitive tests included Symbol Digit Modalities Test, Rey Auditory Verbal Learning Test, and Brief Visual Memory Test. Manual segmentation of the thalamus was performed using co-registered MPRAGE and proton density-weighted MR images, and FreeSurfer automatic thalamus segmentation volumes were measured using MPRAGE images. All volumes were normalized by the volume of the manually-segmented intracranial cavity.
Results: Compared with controls, MS patients had significantly lower scores on all cognitive measures and thalamus volumes except for the left thalamic volume obtained through FreeSurfer. This one exception appears to stem from FreeSurfer significantly overestimating MS patients" left thalamic volume by 14% compared with the manually obtained value. FreeSurfer thalamic volumes were generally greater and had a higher floor-value than manual segmentation. Furthermore, none of the thalamic volumes from FreeSurfer were significantly correlated with cognitive measures, whereas all of those based on manual segmentation were.
Conclusion: In studies centering on MS patients, automatic segmentation by FreeSurfer appears to be inadequate for obtaining regional volumes of the thalamus. The present findings demonstrate the need to reduce errors in thalamus volume analysis and the assessment of accurate thalamic pathology has important implications in clinical symptoms of MS patients.
Disclosure: Sharon G. Lynch receives grant funding from the National Multiple Sclerosis Society, and has participated in Multi-center MS trials and grants funded by Biogen, Teva, Novartis, Acorda, Opexa, Roche, Genentech, Genzyme, Sun Pharma, Vaccinex, Actelion, NMSS, and NIH. This work was partially funded by Grant RG 4495-A-4 from NMSS.
In-Young Choi: nothing to disclose.
Phil Lee: nothing to disclose.
Peter Adany: nothing to disclose.
Erica Sherry: nothing to disclose.
Abbey J. Hughes: nothing to disclose.
Douglas R. Denney: nothing to disclose.
Abstract: EP1456
Type: ePoster
Abstract Category: Pathology and pathogenesis of MS - Imaging
Background: Thalamic volume loss may be an important factor in MS-related clinical impairment as thalamic pathology occurs in early stages of MS. Thalamic volume changes may predict long-term accumulation of disability and may correlate with cognitive impairment. However, the thalamus poses challenges for anatomical segmentation. We found significant segmentation errors and biases in the thalamic volumes obtained through automatic segmentation using FreeSurfer. We investigated correlations between cognitive measures and thalamic volumes obtained through both manual and automatic segmentation.
Objective: To investigate the correlation of lateralized and combined thalamic volumes with frequently-affected cognitive domains of memory and information processing speed in MS patients.
Methods: 73 MS patients (RRMS, PPMS, SPMS) and 35 healthy controls were studied. Cognitive tests included Symbol Digit Modalities Test, Rey Auditory Verbal Learning Test, and Brief Visual Memory Test. Manual segmentation of the thalamus was performed using co-registered MPRAGE and proton density-weighted MR images, and FreeSurfer automatic thalamus segmentation volumes were measured using MPRAGE images. All volumes were normalized by the volume of the manually-segmented intracranial cavity.
Results: Compared with controls, MS patients had significantly lower scores on all cognitive measures and thalamus volumes except for the left thalamic volume obtained through FreeSurfer. This one exception appears to stem from FreeSurfer significantly overestimating MS patients" left thalamic volume by 14% compared with the manually obtained value. FreeSurfer thalamic volumes were generally greater and had a higher floor-value than manual segmentation. Furthermore, none of the thalamic volumes from FreeSurfer were significantly correlated with cognitive measures, whereas all of those based on manual segmentation were.
Conclusion: In studies centering on MS patients, automatic segmentation by FreeSurfer appears to be inadequate for obtaining regional volumes of the thalamus. The present findings demonstrate the need to reduce errors in thalamus volume analysis and the assessment of accurate thalamic pathology has important implications in clinical symptoms of MS patients.
Disclosure: Sharon G. Lynch receives grant funding from the National Multiple Sclerosis Society, and has participated in Multi-center MS trials and grants funded by Biogen, Teva, Novartis, Acorda, Opexa, Roche, Genentech, Genzyme, Sun Pharma, Vaccinex, Actelion, NMSS, and NIH. This work was partially funded by Grant RG 4495-A-4 from NMSS.
In-Young Choi: nothing to disclose.
Phil Lee: nothing to disclose.
Peter Adany: nothing to disclose.
Erica Sherry: nothing to disclose.
Abbey J. Hughes: nothing to disclose.
Douglas R. Denney: nothing to disclose.