Limits...
Marked effects of intracranial volume correction methods on sex differences in neuroanatomical structures: a HUNT MRI study.

Pintzka CW, Hansen TI, Evensmoen HR, Håberg AK - Front Neurosci (2015)

Bottom Line: In addition, sex-specific subsamples were created to investigate whether differences were an effect of head size or sex.Sex differences were detected in a few structures; amygdala, cerebellar cortex, and 3rd ventricle were larger in men, but the effect sizes were small.In conclusion, sex plays a minor role for neuroanatomical volume differences; most differences are related to ICV.

View Article: PubMed Central - PubMed

Affiliation: Department of Neuroscience, Norwegian University of Science and Technology Trondheim, Norway ; Department of Medical Imaging, St. Olav's University Hospital Trondheim, Norway.

ABSTRACT
To date, there is no consensus whether sexual dimorphism in the size of neuroanatomical structures exists, or if such differences are caused by choice of intracranial volume (ICV) correction method. When investigating volume differences in neuroanatomical structures, corrections for variation in ICV are used. Commonly applied methods are the ICV-proportions, ICV-residuals and ICV as a covariate of no interest, ANCOVA. However, these different methods give contradictory results with regard to presence of sex differences. Our aims were to investigate presence of sexual dimorphism in 18 neuroanatomical volumes unrelated to ICV-differences by using a large ICV-matched subsample of 304 men and women from the HUNT-MRI general population study, and further to demonstrate in the entire sample of 966 healthy subjects, which of the ICV-correction methods gave results similar to the ICV-matched subsample. In addition, sex-specific subsamples were created to investigate whether differences were an effect of head size or sex. Most sex differences were related to volume scaling with ICV, independent of sex. Sex differences were detected in a few structures; amygdala, cerebellar cortex, and 3rd ventricle were larger in men, but the effect sizes were small. The residuals and ANCOVA methods were most effective at removing the effects of ICV. The proportions method suffered from systematic errors due to lack of proportionality between ICV and neuroanatomical volumes, leading to systematic mis-assignment of structures as either larger or smaller than their actual size. Adding additional sexual dimorphic covariates to the ANCOVA gave opposite results of those obtained in the ICV-matched subsample or with the residuals method. The findings in the current study explain some of the considerable variation in the literature on sexual dimorphisms in neuroanatomical volumes. In conclusion, sex plays a minor role for neuroanatomical volume differences; most differences are related to ICV.

No MeSH data available.


Related in: MedlinePlus

Scatterplots of the volumes of the 18 neuroanatomical structures vs. ICV, women in orange and men in blue. The linear regression line for all subjects combined is superimposed in solid line with the 95% confidence interval in dashed. Y-axis: structural volume, x-axis: ICV. The letter A or B in the lower right corner of each neuroantomical volume indicates whether the y-intercept in the regression line is positive or negative. The two gray squares in the lower right corner involve imaginary data. In these cases, a sex difference is clearly not present. The proportions method would however show significant effect of sex because the regression line does not have zero y-intercept.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4496575&req=5

Figure 4: Scatterplots of the volumes of the 18 neuroanatomical structures vs. ICV, women in orange and men in blue. The linear regression line for all subjects combined is superimposed in solid line with the 95% confidence interval in dashed. Y-axis: structural volume, x-axis: ICV. The letter A or B in the lower right corner of each neuroantomical volume indicates whether the y-intercept in the regression line is positive or negative. The two gray squares in the lower right corner involve imaginary data. In these cases, a sex difference is clearly not present. The proportions method would however show significant effect of sex because the regression line does not have zero y-intercept.

Mentions: Figure 4 shows scatterplots of the raw neuroanatomical volumes vs. ICV for men and women. All structures had a linear relationship with ICV and the confidence intervals for the regression lines were generally narrow, with the smaller structures having somewhat wider bands than the larger ones. Although there was a strong correlation between most structures and ICV, no structure was directly proportionate to ICV (i.e., regression line did not go through the origin for any of the volumes). The structures were divided in two groups according to the constant in the regression line; group A had structures with positive y-intercept whereas group B had structures with negative y-intercept. Group A contained all the gray matter volumes (accumbens, amygdala, caudate, hippocampus, pallidum, putamen, thalamus, and the cerebral and cerebellar cortices) whereas group B contained ventricles and white matter structures. In the hypothetical dataset in Figure 4 (shaded squares), women had a relatively larger volume (expressed as proportions of ICV) of the structure with a positive y-intercept, whereas men had a relatively larger volume of the structure with a negative y-intercept. Whether the regression line has a positive or negative y-intercept affects results obtained with the proportions method and will be discussed further below.


Marked effects of intracranial volume correction methods on sex differences in neuroanatomical structures: a HUNT MRI study.

Pintzka CW, Hansen TI, Evensmoen HR, Håberg AK - Front Neurosci (2015)

Scatterplots of the volumes of the 18 neuroanatomical structures vs. ICV, women in orange and men in blue. The linear regression line for all subjects combined is superimposed in solid line with the 95% confidence interval in dashed. Y-axis: structural volume, x-axis: ICV. The letter A or B in the lower right corner of each neuroantomical volume indicates whether the y-intercept in the regression line is positive or negative. The two gray squares in the lower right corner involve imaginary data. In these cases, a sex difference is clearly not present. The proportions method would however show significant effect of sex because the regression line does not have zero y-intercept.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4496575&req=5

Figure 4: Scatterplots of the volumes of the 18 neuroanatomical structures vs. ICV, women in orange and men in blue. The linear regression line for all subjects combined is superimposed in solid line with the 95% confidence interval in dashed. Y-axis: structural volume, x-axis: ICV. The letter A or B in the lower right corner of each neuroantomical volume indicates whether the y-intercept in the regression line is positive or negative. The two gray squares in the lower right corner involve imaginary data. In these cases, a sex difference is clearly not present. The proportions method would however show significant effect of sex because the regression line does not have zero y-intercept.
Mentions: Figure 4 shows scatterplots of the raw neuroanatomical volumes vs. ICV for men and women. All structures had a linear relationship with ICV and the confidence intervals for the regression lines were generally narrow, with the smaller structures having somewhat wider bands than the larger ones. Although there was a strong correlation between most structures and ICV, no structure was directly proportionate to ICV (i.e., regression line did not go through the origin for any of the volumes). The structures were divided in two groups according to the constant in the regression line; group A had structures with positive y-intercept whereas group B had structures with negative y-intercept. Group A contained all the gray matter volumes (accumbens, amygdala, caudate, hippocampus, pallidum, putamen, thalamus, and the cerebral and cerebellar cortices) whereas group B contained ventricles and white matter structures. In the hypothetical dataset in Figure 4 (shaded squares), women had a relatively larger volume (expressed as proportions of ICV) of the structure with a positive y-intercept, whereas men had a relatively larger volume of the structure with a negative y-intercept. Whether the regression line has a positive or negative y-intercept affects results obtained with the proportions method and will be discussed further below.

Bottom Line: In addition, sex-specific subsamples were created to investigate whether differences were an effect of head size or sex.Sex differences were detected in a few structures; amygdala, cerebellar cortex, and 3rd ventricle were larger in men, but the effect sizes were small.In conclusion, sex plays a minor role for neuroanatomical volume differences; most differences are related to ICV.

View Article: PubMed Central - PubMed

Affiliation: Department of Neuroscience, Norwegian University of Science and Technology Trondheim, Norway ; Department of Medical Imaging, St. Olav's University Hospital Trondheim, Norway.

ABSTRACT
To date, there is no consensus whether sexual dimorphism in the size of neuroanatomical structures exists, or if such differences are caused by choice of intracranial volume (ICV) correction method. When investigating volume differences in neuroanatomical structures, corrections for variation in ICV are used. Commonly applied methods are the ICV-proportions, ICV-residuals and ICV as a covariate of no interest, ANCOVA. However, these different methods give contradictory results with regard to presence of sex differences. Our aims were to investigate presence of sexual dimorphism in 18 neuroanatomical volumes unrelated to ICV-differences by using a large ICV-matched subsample of 304 men and women from the HUNT-MRI general population study, and further to demonstrate in the entire sample of 966 healthy subjects, which of the ICV-correction methods gave results similar to the ICV-matched subsample. In addition, sex-specific subsamples were created to investigate whether differences were an effect of head size or sex. Most sex differences were related to volume scaling with ICV, independent of sex. Sex differences were detected in a few structures; amygdala, cerebellar cortex, and 3rd ventricle were larger in men, but the effect sizes were small. The residuals and ANCOVA methods were most effective at removing the effects of ICV. The proportions method suffered from systematic errors due to lack of proportionality between ICV and neuroanatomical volumes, leading to systematic mis-assignment of structures as either larger or smaller than their actual size. Adding additional sexual dimorphic covariates to the ANCOVA gave opposite results of those obtained in the ICV-matched subsample or with the residuals method. The findings in the current study explain some of the considerable variation in the literature on sexual dimorphisms in neuroanatomical volumes. In conclusion, sex plays a minor role for neuroanatomical volume differences; most differences are related to ICV.

No MeSH data available.


Related in: MedlinePlus