Sex classification using long-range temporal dependence of resting-state functional MRI time series

Elvisha Dhamala, Keith Jamison, Mert R. Sabuncu, Amy KuceyeskibioRxiv 2019 · DOI

TL;DR

This study explores sex differences in brain activity patterns using resting-state fMRI, finding that males exhibit more persistent long-range temporal dependence in certain brain regions. This temporal dependence, along with regional brain volume, can be used to classify sex with high accuracy, highlighting the importance of controlling for volume differences in such studies.

What Problem Does This Paper Address?

Understanding sex differences in the brain is crucial for comprehending neurological and psychiatric disorders, as many conditions exhibit sex-specific prevalence, onset, and symptoms. While previous research has identified sex differences in brain structure and function, there's a need to explore how the temporal dynamics of brain activity might differ between sexes. Specifically, this paper investigates whether long-range temporal dependence in resting-state fMRI signals can serve as a reliable marker for sex classification and whether these temporal differences persist even when controlling for overall brain volume.

Methods

The researchers employed the Hurst Exponent (HE) to quantify long-range temporal dependence in resting-state fMRI time series. HE is a measure that indicates how persistent or anti-persistent a time series is. They used Detrended Fluctuation Analysis (DFA) to compute HE. For sex classification, they utilized linear Support Vector Machines (SVMs), a common machine learning algorithm for binary classification. To account for potential biases, they performed analyses on multiple brain atlases and also explored ensemble models combining predictions from different atlases. A key methodological aspect was the strict matching of male and female participants for total grey matter volume to isolate the effects of temporal dependence and regional volume differences.

Datasets

The study utilized data from the Human Connectome Project (HCP) – Young Adult S1200 release. This dataset includes high-resolution resting-state fMRI data from a large cohort of healthy adults. For their primary analysis, they selected a subset of 390 participants, consisting of 190 non-overlapping male-female pairs, who were strictly matched for total grey matter volume. They also performed analyses on the entire HCP dataset (n=1003) and on randomly selected subsets to assess the impact of volume matching and sample size.

Key Insights and Ideas

  • Temporal Dynamics as a Sex Differentiator: The study demonstrates that the way brain activity fluctuates over time (long-range temporal dependence) carries significant information about an individual's sex, even in a resting state.
  • Male Brains: More Persistent Fluctuations: Males, on average, show more persistent long-range temporal dependence in regions like the temporal, parietal, and occipital cortices compared to females.
  • Volume Matters (A Lot): Strict volume matching is critical. Sex differences in brain volume are substantial and can confound findings if not controlled for, potentially inflating observed differences in functional measures.
  • Complementary Information: Both temporal dependence (HE) and regional brain volume provide distinct, yet complementary, information for sex classification, suggesting different underlying neurobiological substrates.

Key Results

  • HE-based Sex Classification: Using regional temporal dependence (HE), sex classification achieved a mean accuracy of up to 81.1% and a mean AUC of 0.86 across atlases. Key regions for this classification included the cerebellum, amygdala, frontal cortex, and occipital cortex.
  • Volume-based Sex Classification: Even after strict grey matter volume matching, regional volume differences allowed for sex classification with a mean accuracy of up to 85.3% and a mean AUC of 0.92. The most important features for volume-based classification were the cerebellum, cingulate cortex, and temporal cortex.
  • Impact of Volume Matching: Volume matching significantly reduced the classification performance of HE-based biomarkers, suggesting that some of the observed temporal differences might be indirectly related to volume.
  • Regional Volume Differences Persist: Despite global volume matching, significant sex differences in regional brain volume were observed, with males having larger volumes in certain areas (e.g., cerebellum, amygdala) and females in others (e.g., cingulate cortex, parietal cortex).

Limitations and Future Work

  • Curse of Dimensionality: The study acknowledges that neuroimaging data is high-dimensional, and while parcellation reduced dimensionality, some information might have been lost. Future work could explore voxel-wise analyses to mitigate this.
  • HE's Characterization Limits: The HE is a single scaling parameter and may not fully capture the complex temporal features of fMRI time series. Exploring alternatives like fractal Hölder exponents is suggested.
  • Hormonal Influences: The study did not evaluate the relationship between hormonal measures (e.g., menstrual cycle phase in females) and HE. Future research should collect hormonal data to investigate these effects.
  • Gender Identity and Fluidity: The study relied on self-reported biological sex and did not consider gender identity or fluidity, which could influence neuroimaging biomarkers. Future studies should incorporate this information.
  • Generalizability: The findings are based on HCP data, and time series can be sensitive to scan parameters and preprocessing pipelines. Generalizability to other datasets needs to be confirmed.
  • Biological Substrates: The underlying biological mechanisms driving the observed sex differences in temporal dependence are not yet understood. Future research is needed to elucidate these substrates and their behavioral or cognitive implications.

Summary updated

Abstract

Abstract A thorough understanding of sex differences, if any, that exist in the brains of healthy individuals is crucial for the study of neurological illnesses that exhibit differences in clinical and behavioural phenotypes between males and females. In this work, we evaluate sex differences in regional temporal dependence of resting-state brain activity using 195 male-female pairs (aged 22-37) from the Human Connectome Project. Male-female pairs are strictly matched for total grey matter volume. We find that males have more persistent long-range temporal dependence than females in regions within temporal, parietal, and occipital cortices. Machine learning algorithms trained on regional temporal dependence measures achieve sex classification accuracies of up to 81%. Regions with the strongest feature importance in the sex classification task included cerebellum, amygdala, frontal cortex, and occipital cortex. Additionally, we find that even after males and females are strictly matched on total grey matter volume, significant regional volumetric sex differences persist in many cortical and subcortical regions. Our results indicate males have larger cerebella, hippocampi, parahippocampi, thalami, caudates, and amygdalae while females have larger cingulates, precunei, frontal cortices, and parietal cortices. Sex classification based on regional volume achieves accuracies of up to 85%; cerebellum, cingulate cortex, and temporal cortex are the most important features. These findings highlight the important role of strict volume matching when studying brain-based sex differences. Differential patterns in regional temporal dependence between males and females identifies a potential neurobiological substrate underlying sex differences in functional brain activation patterns and the behaviours with which they correlate.
BibTeX
@misc{dhamala2019sex,
  author = {Elvisha Dhamala and Keith Jamison and Mert R. Sabuncu and Amy Kuceyeski},
  title = {{Sex classification using long-range temporal dependence of resting-state functional MRI time series}},
  year = {2019},
  doi = {10.1101/809954},
  howpublished = {bioRxiv},
}
Figure