2473

Age and sex contributions to variance in resting state fMRI (rs-fMRI) temporal complexity analysis
Nicholas Maurice Simard1,2, Dinesh A Kumbhare3,4, Stephan Ulmer5,6, and Michael D Noseworthy1,2,7,8
1Electrical and Computer Engineering, McMaster University, Hamilton, ON, Canada, 2St. Joseph's Healthcare Hamilton, Imaging Research Centre, Hamilton, ON, Canada, 3Toronto Rehabilitation Institute, Toronto, ON, Canada, 4Department of Medicine, University of Toronto, Toronto, ON, Canada, 5neurorad.ch, Zurich, Switzerland, 6Department of Radiology and Neuroradiology, University hospital of Schleswig-Holstein, Kiel, Germany, 7School of Biomedical Engineering, McMaster University, Hamilton, ON, Canada, 8Radiology, McMaster University, Hamilton, ON, Canada

Synopsis

Understanding the process of aging and the differences in sex with regards to large data repositories can help improve the implementation of machine learning and artificial intelligence paradigms in neuroimaging. The following research presents data that identifies a relationship between aging and sex in resting state functional magnetic resonance imaging (rs-fMRI) data. Using over 10,000 age and sex matched healthy controls and performing a homebuilt processing pipeline for rs-fMRI data, significant relationships between aging and reduced temporal complexity (TC) was found (p=0.03058), along with women having a higher TC than men (p=0.000623).

Introduction

Because of its richness, magnetic resonance imaging (MRI) data is becoming increasingly used in machine learning (ML) and artificial intelligence (AI) studies. Understanding the relationship of healthy aging on quantitative MRI data (e.g. diffusion tensor imaging, DTI, and resting state fMRI) can improve design/execution of ML and AI algorithms for detecting/grading of neuropathology. With the evolution of large-scale data repositories (IDA, HCP, NITRC, UK BIOBANK, etc.)1,2,3,4, it is imperative to evaluate the effects of healthy aging and sex on any quantitative MRI metrics being collected.

Resting state fMRI (rs-fMRI) is one particular dataset that is available in massive volumes. It is typically analyzed as a between-region time domain correlation. Resting state fMRI signal complexity has also shown value where reduced complexity equates to reduction in function. The goal of this study was to assess the effect of age and sex on rs-fMRI signal complexity over brain ROIs defined by the Julich Histological Atlas (JHA)5,6.

Methods

Age related gray and white matter changes in normal adult brains, with regards to voxel-based morphometry and brain volume, have already been investigated in the context of structural and diffusion tensor imaging7. However, the effect of sex and age on rs-fMRI temporal complexity has never been assessed. Frequently, functional connectivity is evaluated with rs-fMRI data. However, biologically based time-varying signals can be classified as statistical fractals and enable functional complexity analysis. Investigating the physiological temporal complexity has shown utility in the evaluation of a number of neurological disorders as it can provide insight into the local function of specific gray matter regions8,9.

To assess functional complexity, over 10,000 healthy control 3T datasets (males and females between the ages of 18-70) were downloaded from open-source data repositories. A minimum of (n=88) for ages 30, 50, and 60 years old, sex matched, were collected to satisfy Cochran’s formula for 95% sensitivity with regards to complexity analysis. Data acquisition schemes and protocols varied across repositories; however, statistical map distributions were generated showing regional mean and standard deviations per age and sex. Statistical normality was also verified through skewness and kurtosis tests. A homebuilt python processing pipeline facilitated the voxel-wise complexity analysis by calculating a voxel’s temporal fractal dimension10,11. Gray matter regions of interest (ROIs) were then segmented in reference to the Julich Histological Atlas and each ROI was evaluated for each age and sex group5,6. For this particular study, individuals at age 30, 50, and 60yrs were selected for comparison.

Results

In rs-fMRI data, initial analysis demonstrates that there exist slight differences in mean values of temporal complexity (TC), for each ROI in the control data, between the ages of 30, 50, and 60 (Figure 1). A one-way ANOVA test identified that these results were significant (p=0.03058) demonstrating a reduction in TC associated with aging13. It is also interesting to note the TC reduction in the premotor cortex and medial geniculate bodies. The standard deviations for each age group and ROI show there are significant differences in the 30-year-old population (p=0.0198). However, there are limited differences in variability in temporal complexity between the 50 and 60 year old age groups. These standard deviations also support the idea that there is increased variability in temporal complexity with age.

Data in this study demonstrated that there also exist differences in mean TC values, for ROIs in the control data, between males and females in the 30 and 60 year old age groups (Figure 2). A one-way ANOVA identified that these results were highly significant (p=0.000623), implying that healthy women have higher brain TC compared to healthy men in these age groups. The caveat to this finding is that women also demonstrate a higher amount of variance compared to men, as demonstrated by the ANOVA of TC standard deviations for each age group, ROI, and sex (p=0.001649).

Discussion

This study identified decreases in brain resting state BOLD temporal complexity related to aging. The lack of concordance between age groups should be investigated on a more microscopic scale as only 3 age groups were selected for this initial study. It is interesting to note that some specific ROIs (i.e. medial geniculate body and premotor cortex) may require further investigation with regards to their physiology as it relates to functional complexity. This preliminary study demonstrates how essential it is to establish functional variations due to aging.

Conclusion

This study demonstrates the nuances that must be investigated when considering ‘Big Data’. These results have implications in the implementation of machine learning and artificial intelligence models in neuroimaging. Further investigation is also required to ascertain whether MRI vendor contributes significant variance towards these large datasets. Each vendor has slightly different image encoding approaches and fat suppression techniques that could be significant with regards to this type of data. Lastly, more in-depth statistical investigation is required to establish specific relationships between altered temporal functional complexity with aging and sex.

Acknowledgements

Special thanks to my supervisors and mentors Dr. Michael D. Noseworthy, Dr. Dinesh Kumbhare, Dr. Stephan Ulmer, and Norm Konyer for their abilities and guidance. Thank you to my PhD committee members Dr. Carol DeMatteo, Dr. Jim Reilly, and Dr. Nicholas Bock for their unique perspective and research insights. Thank you to my fellow Brain Trust colleagues Ethan Danielli, Ethan Samson, Neil MacPhee, Bhanu Sharma, Mahnaz Tajik, and Lauren Anderson. Thank you for the financial support from McMaster University, the National Sciences and Engineering Research Council of Canada (NSERC), and the MITACS Accelerate program. Lastly, thank you to my family and friends who have tremendously supported me along the way.

References

  1. Crawford KL, Neu SC, Toga AW. The image and data archive at the laboratory of neuro imaging. Neuroimage. 2016 Jan 1;124:1080-3.
  2. Van Essen DC, Smith SM, Barch DM, Behrens TE, Yacoub E, Ugurbil K, Wu-Minn HCP Consortium. The WU-Minn human connectome project: an overview. Neuroimage. 2013 Oct 15;80:62-79.
  3. Sudlow C, Gallacher J, Allen N, Beral V, Burton P, Danesh J, Downey P, Elliott P, Green J, Landray M, Liu B. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS medicine. 2015 Mar 31;12(3):e1001779.
  4. Kennedy DN, Haselgrove C, Riehl J, Preuss N, Buccigrossi R. The NITRC image repository. NeuroImage. 2016 Jan 1;124:1069-73.
  5. Bürgel U, Schormann T, Schleicher A, Zilles K. Mapping of histologically identified long fiber tracts in human cerebral hemispheres to the MRI volume of a reference brain: position and spatial variability of the optic radiation. Neuroimage. 1999 Nov 1;10(5):489-99.
  6. Bürgel U, Amunts K, Hoemke L, Mohlberg H, Gilsbach JM, Zilles K. White matter fiber tracts of the human brain: three-dimensional mapping at microscopic resolution, topography and intersubject variability. Neuroimage. 2006 Feb 15;29(4):1092-105.
  7. Farokhian F, Yang C, Beheshti I, Matsuda H, Wu S. Age-related gray and white matter changes in normal adult brains. Aging and disease. 2017 Dec;8(6):899.
  8. Eke A, Herman P, Bassingthwaighte J, Raymond G, Percival D, Cannon M, Balla I, Ikrényi C. Physiological time series: distinguishing fractal noises from motions. Pflügers Archiv. 2000 Feb;439(4):403-15.
  9. Eke A, Herman P, Kocsis L, Kozak LR. Fractal characterization of complexity in temporal physiological signals. Physiological measurement. 2002 Jan 28;23(1):R1.
  10. Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Computers and Biomedical research. 1996 Jun 1;29(3):162-73.
  11. Jenkinson M, Beckmann CF, Behrens TE, Woolrich MW, Smith SM. Fsl. Neuroimage. 2012 Aug 15;62(2):782-90.
  12. Weber AM, Soreni N, Noseworthy MD. A preliminary study on the effects of acute ethanol ingestion on default mode network and temporal fractal properties of the brain. Magnetic Resonance Materials in Physics, Biology and Medicine. 2014 Aug;27(4):291-301.

Figures

Top, A colour coded visualization of the temporal complexity (TC) mean values across each ROI for the 30, 50, and 60 year old age groups. A reduction in complexity is demonstrated within older age groups due to prospective reduction in brain activity due to aging. Bottom, A similar colour coded visualization of the temporal complexity (TC) standard deviation values across each ROI for the 30, 50, and 60 year old age groups. A higher variance in TC is demonstrated due to prospectively higher variance in aging populations.

Top, A colour coded visualization of temporal complexity (TC) mean values across each ROI for the 30 and 60 year old age groups and in both Male and Female sex. A reduction in temporal complexity is seen with males, which warrants further investigation. Bottom, Similarly, a colour coded visualization of the temporal complexity (TC) standard deviation values across each ROI for the 30 and 60 year old age groups and in both Male and Female sex. A higher variance in TC is demonstrated within Males due to prospectively higher variance in the aging women population.

Proc. Intl. Soc. Mag. Reson. Med. 30 (2022)
2473
DOI: https://doi.org/10.58530/2022/2473