Limits...
Assessing age-related gray matter decline with voxel-based morphometry depends significantly on segmentation and normalization procedures.

Callaert DV, Ribbens A, Maes F, Swinnen SP, Wenderoth N - Front Aging Neurosci (2014)

Bottom Line: Comparing GM decline between 21 young subjects (mean age 23) and 18 elderly (mean age 66) revealed that volumetric measurements differed significantly between methods.Our findings suggest that algorithms that provide only limited degrees of freedom for local deformations (such as the unified segmentation and normalization of SPM5/SPM8) tend to overestimate between-group differences in VBM results when compared to methods providing more flexible warping.This difference seems to be most pronounced if the anatomy of one of the groups deviates from custom templates, a finding that is of particular importance when results are compared across studies using different VBM methods.

View Article: PubMed Central - PubMed

Affiliation: Movement Control and Neuroplasticity Research Group, Department of Kinesiology KU Leuven, Belgium ; CNRS, INCIA, UMR 5287, University of Bordeaux Talence, France.

ABSTRACT
Healthy ageing coincides with a progressive decline of brain gray matter (GM) ultimately affecting the entire brain. For a long time, manual delineation-based volumetry within predefined regions of interest (ROI) has been the gold standard for assessing such degeneration. Voxel-Based Morphometry (VBM) offers an automated alternative approach that, however, relies critically on the segmentation and spatial normalization of a large collection of images from different subjects. This can be achieved via different algorithms, with SPM5/SPM8, DARTEL of SPM8 and FSL tools (FAST, FNIRT) being three of the most frequently used. We complemented these voxel based measurements with a ROI based approach, whereby the ROIs are defined by transforms of an atlas (containing different tissue probability maps as well as predefined anatomic labels) to the individual subject images in order to obtain volumetric information at the level of the whole brain or within separate ROIs. Comparing GM decline between 21 young subjects (mean age 23) and 18 elderly (mean age 66) revealed that volumetric measurements differed significantly between methods. The unified segmentation/normalization of SPM5/SPM8 revealed the largest age-related differences and DARTEL the smallest, with FSL being more similar to the DARTEL approach. Method specific differences were substantial after segmentation and most pronounced for the cortical structures in close vicinity to major sulci and fissures. Our findings suggest that algorithms that provide only limited degrees of freedom for local deformations (such as the unified segmentation and normalization of SPM5/SPM8) tend to overestimate between-group differences in VBM results when compared to methods providing more flexible warping. This difference seems to be most pronounced if the anatomy of one of the groups deviates from custom templates, a finding that is of particular importance when results are compared across studies using different VBM methods.

No MeSH data available.


Related in: MedlinePlus

Method × Age interaction of GM probabilities at the voxel level after common normalization to MNI space (see Methods section for details). Voxels exhibiting significantly larger GM differences between young versus elderly group when analyzed with SPM than with DARTEL (A), than with FSL (B) and than with ISBNRR (C). Also, FSL (D) and ISBNRR (E) revealed for some voxels larger GM differences between young versus elderly subjects than DARTEL. There were only minor differences in age-specific GM decline between FSL and ISBNRR (F,G). Statistical parametric maps were either thresholded at p < 0.001 uncorrected (red) or p < 0.05 FWE corrected (yellow).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4066859&req=5

Figure 3: Method × Age interaction of GM probabilities at the voxel level after common normalization to MNI space (see Methods section for details). Voxels exhibiting significantly larger GM differences between young versus elderly group when analyzed with SPM than with DARTEL (A), than with FSL (B) and than with ISBNRR (C). Also, FSL (D) and ISBNRR (E) revealed for some voxels larger GM differences between young versus elderly subjects than DARTEL. There were only minor differences in age-specific GM decline between FSL and ISBNRR (F,G). Statistical parametric maps were either thresholded at p < 0.001 uncorrected (red) or p < 0.05 FWE corrected (yellow).

Mentions: After previous analyses have established that the different segmentation methods have a significant influence on quantifying age specific changes in overall GM volumes, we asked where these method-specific differences are located in the brain. To this end we transformed all GM probability maps from native space to ICBM152 space and, importantly, applied for each individual the same transformation parameters to all GM probability maps. Subsequently we determined the Age × Meth interaction. Note that this will not reveal general differences in GM probability across methods or areas exhibiting GM decline with age. Instead this analysis was performed to identify where in the brain the age related GM decline is over- or under-estimated when compared across segmentation methods. Comparing SPM to DARTEL segmentation, SPM revealed significantly larger GM differences between young and elderly throughout the surface of the brain (Figure 3A). When SPM was compared to FSL, differences in the aging effect were most pronounced for dorsal cortical region, particularly around the central sulcus (Figure 3B). By contrast when comparing SPM to IBSNRR (Figure 3C), differences tended to be located more inferiorly, for example around the sylvian fissure. DARTEL segmentation revealed also a significantly smaller age effect on GM when compared to FSL (Figure 3D) and IBSNRR (Figure 3E): differences with both methods were particularly located around the interhemispheric and the sylvian fissure even though this effect was much more pronounced when compared to FSL than to IBSNRR. Finally, also FSL and IBSNRR segmentation exhibited minor differences when quantifying age related GM decline, such that FSL tended to reveal larger age effects for inferior located cortical surface areas (particularly around the sylvian fissure, Figure 3F) and lower age related effects for the dorsal cortical areas (Figure 3G). In summary, method specific differences seem to arise mainly from differential segmentation results for the cortical surface and particularly close to large sulci and fissures. By contrast, differences were only minor for subcortical gray matter structures.


Assessing age-related gray matter decline with voxel-based morphometry depends significantly on segmentation and normalization procedures.

Callaert DV, Ribbens A, Maes F, Swinnen SP, Wenderoth N - Front Aging Neurosci (2014)

Method × Age interaction of GM probabilities at the voxel level after common normalization to MNI space (see Methods section for details). Voxels exhibiting significantly larger GM differences between young versus elderly group when analyzed with SPM than with DARTEL (A), than with FSL (B) and than with ISBNRR (C). Also, FSL (D) and ISBNRR (E) revealed for some voxels larger GM differences between young versus elderly subjects than DARTEL. There were only minor differences in age-specific GM decline between FSL and ISBNRR (F,G). Statistical parametric maps were either thresholded at p < 0.001 uncorrected (red) or p < 0.05 FWE corrected (yellow).
© Copyright Policy - open-access
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4066859&req=5

Figure 3: Method × Age interaction of GM probabilities at the voxel level after common normalization to MNI space (see Methods section for details). Voxels exhibiting significantly larger GM differences between young versus elderly group when analyzed with SPM than with DARTEL (A), than with FSL (B) and than with ISBNRR (C). Also, FSL (D) and ISBNRR (E) revealed for some voxels larger GM differences between young versus elderly subjects than DARTEL. There were only minor differences in age-specific GM decline between FSL and ISBNRR (F,G). Statistical parametric maps were either thresholded at p < 0.001 uncorrected (red) or p < 0.05 FWE corrected (yellow).
Mentions: After previous analyses have established that the different segmentation methods have a significant influence on quantifying age specific changes in overall GM volumes, we asked where these method-specific differences are located in the brain. To this end we transformed all GM probability maps from native space to ICBM152 space and, importantly, applied for each individual the same transformation parameters to all GM probability maps. Subsequently we determined the Age × Meth interaction. Note that this will not reveal general differences in GM probability across methods or areas exhibiting GM decline with age. Instead this analysis was performed to identify where in the brain the age related GM decline is over- or under-estimated when compared across segmentation methods. Comparing SPM to DARTEL segmentation, SPM revealed significantly larger GM differences between young and elderly throughout the surface of the brain (Figure 3A). When SPM was compared to FSL, differences in the aging effect were most pronounced for dorsal cortical region, particularly around the central sulcus (Figure 3B). By contrast when comparing SPM to IBSNRR (Figure 3C), differences tended to be located more inferiorly, for example around the sylvian fissure. DARTEL segmentation revealed also a significantly smaller age effect on GM when compared to FSL (Figure 3D) and IBSNRR (Figure 3E): differences with both methods were particularly located around the interhemispheric and the sylvian fissure even though this effect was much more pronounced when compared to FSL than to IBSNRR. Finally, also FSL and IBSNRR segmentation exhibited minor differences when quantifying age related GM decline, such that FSL tended to reveal larger age effects for inferior located cortical surface areas (particularly around the sylvian fissure, Figure 3F) and lower age related effects for the dorsal cortical areas (Figure 3G). In summary, method specific differences seem to arise mainly from differential segmentation results for the cortical surface and particularly close to large sulci and fissures. By contrast, differences were only minor for subcortical gray matter structures.

Bottom Line: Comparing GM decline between 21 young subjects (mean age 23) and 18 elderly (mean age 66) revealed that volumetric measurements differed significantly between methods.Our findings suggest that algorithms that provide only limited degrees of freedom for local deformations (such as the unified segmentation and normalization of SPM5/SPM8) tend to overestimate between-group differences in VBM results when compared to methods providing more flexible warping.This difference seems to be most pronounced if the anatomy of one of the groups deviates from custom templates, a finding that is of particular importance when results are compared across studies using different VBM methods.

View Article: PubMed Central - PubMed

Affiliation: Movement Control and Neuroplasticity Research Group, Department of Kinesiology KU Leuven, Belgium ; CNRS, INCIA, UMR 5287, University of Bordeaux Talence, France.

ABSTRACT
Healthy ageing coincides with a progressive decline of brain gray matter (GM) ultimately affecting the entire brain. For a long time, manual delineation-based volumetry within predefined regions of interest (ROI) has been the gold standard for assessing such degeneration. Voxel-Based Morphometry (VBM) offers an automated alternative approach that, however, relies critically on the segmentation and spatial normalization of a large collection of images from different subjects. This can be achieved via different algorithms, with SPM5/SPM8, DARTEL of SPM8 and FSL tools (FAST, FNIRT) being three of the most frequently used. We complemented these voxel based measurements with a ROI based approach, whereby the ROIs are defined by transforms of an atlas (containing different tissue probability maps as well as predefined anatomic labels) to the individual subject images in order to obtain volumetric information at the level of the whole brain or within separate ROIs. Comparing GM decline between 21 young subjects (mean age 23) and 18 elderly (mean age 66) revealed that volumetric measurements differed significantly between methods. The unified segmentation/normalization of SPM5/SPM8 revealed the largest age-related differences and DARTEL the smallest, with FSL being more similar to the DARTEL approach. Method specific differences were substantial after segmentation and most pronounced for the cortical structures in close vicinity to major sulci and fissures. Our findings suggest that algorithms that provide only limited degrees of freedom for local deformations (such as the unified segmentation and normalization of SPM5/SPM8) tend to overestimate between-group differences in VBM results when compared to methods providing more flexible warping. This difference seems to be most pronounced if the anatomy of one of the groups deviates from custom templates, a finding that is of particular importance when results are compared across studies using different VBM methods.

No MeSH data available.


Related in: MedlinePlus