Sensitivity of diffusion-tensor and correlated diffusion imaging to white-matter microstructural abnormalities: application in COVID-19

Nick Teller, Jordan A. Chad, Alexander Wong, Hayden Gunraj, Xiang Ji, Bradley J. MacIntosh, Asaf Gilboa, Eugenie Roudaia, Allison B. Sekuler, Benjamin Lam, Chris Heyn, Sandra E. Black, Simon J. Graham, J. Jean ChenbioRxiv 2022 · DOI

TL;DR

This study introduces Correlated Diffusion Imaging (CDI) as a highly sensitive MRI technique for detecting subtle white matter changes in COVID-19 patients, outperforming traditional Diffusion Tensor Imaging (DTI). CDI reveals distinct patterns of altered water diffusion in the frontal and cerebellar regions, suggesting differential responses to viral infection.

What Problem Does This Paper Address?

COVID-19 is known to affect the brain, particularly the white matter, even in individuals who experience mild symptoms or self-isolate. However, these microstructural changes are often too subtle to be detected by conventional MRI. Existing diffusion MRI methods, like DTI, have shown mixed results in identifying these subtle abnormalities, especially in non-hospitalized patients. This paper aims to address this gap by evaluating the sensitivity of different single-shell diffusion MRI methods, including a novel approach called Correlated Diffusion Imaging (CDI), to detect these COVID-19-related white matter changes.

Methods

The researchers compared three single-shell diffusion MRI modeling methods:

  • Diffusion Tensor Imaging (DTI): A standard technique that models water diffusion using a tensor, providing metrics like Mean Diffusivity (MD) and Fractional Anisotropy (FA).
  • DT-DOME (Orthogonal Moment Decomposition): An extension of DTI that aims to provide more robust anisotropy measures (NA and MO) less confounded by MD.
  • Correlated Diffusion Imaging (CDI): A novel method that uses multiplicative diffusion-image intensities to detect restricted diffusion. The authors implemented CDI using a log transformation to compress its dynamic range.

They used both simulated data and experimental MRI data from COVID-19 patients and controls. Simulations were crucial for understanding the intrinsic sensitivities of each method to changes in diffusivity and anisotropy under varying signal-to-noise ratios (SNR) and contrast levels.

Datasets

The study utilized MRI data from:

  • 39 COVID+ participants: Individuals who tested positive for COVID-19 and self-isolated.
  • 14 COVID- participants: Individuals who experienced flu-like symptoms but tested negative for COVID-19.
  • Age- and sex-matched controls: Used for comparison.

Imaging was performed at baseline, with a subset of COVID+ participants also assessed at a 3-month follow-up.

Key Insights and Ideas

  • CDI's Superior Sensitivity: The most striking insight is CDI's significantly higher sensitivity in detecting COVID-19-related white matter changes compared to DTI and DT-DOME, especially when using single-shell acquisitions. This suggests CDI can pick up subtle microstructural alterations that other methods miss.
  • Opposing Diffusion Patterns: A key finding is the identification of contrasting diffusion patterns in different brain regions. COVID-19 patients showed less restricted diffusion in the frontal lobe but more restricted diffusion in the cerebellum. This highlights that the brain's response to the virus is not uniform.
  • Diffusivity as a Key Indicator: The results strongly suggest that changes in water diffusivity are a primary manifestation of COVID-19-related white matter pathology, rather than changes in anisotropy alone.
  • B-value Dependence: Different b-values (which relate to the diffusion weighting of the MRI signal) influenced the sensitivity of CDI, with lower b-values (e.g., 700 s/mm²) showing the most sensitive results for detecting less restricted diffusion in certain regions.

Key Results

  • CDI detects significant group differences: While DTI and DT-DOME metrics showed no significant differences between COVID+ and COVID- groups, CDI revealed highly significant differences.
  • Frontal lobe: Less restricted diffusion: In the frontal white matter, COVID+ patients exhibited lower log(CDI) values, indicating less restricted diffusion. This aligns with findings suggesting potential extracellular space enlargement or blood-brain barrier (BBB) leakage.
  • Cerebellum: More restricted diffusion: Conversely, in the cerebellar white matter, COVID+ patients showed higher log(CDI) values, suggesting more restricted diffusion. This is consistent with existing literature pointing to the cerebellum's vulnerability to COVID-19.
  • No significant follow-up differences: No significant differences were observed between baseline and 3-month follow-up in the limited cohort, suggesting that observed changes might be persistent or that the follow-up cohort was too small to detect changes.

Limitations and Future Work

  • Limited longitudinal data: The study acknowledges the constraint of limited longitudinal follow-up. While short-term effects were observed, long-term effects might be more widespread. Future work will involve larger sample sizes and expanded longitudinal evaluations.
  • Specificity of DTI: DTI, while sensitive to microstructural changes, is not specific. It's unclear if observed restricted diffusion in the long term reflects compensatory neurogenesis or persistent inflammation. Further research, potentially including autopsy, is needed.
  • CDI in grey matter and other conditions: The authors pose important questions for future research: How does CDI perform in grey matter? How does it reflect neurodegenerative processes in conditions like aging or Alzheimer's disease? Under what circumstances might CDI fail to detect pathology? These questions will drive future validation and characterization of the CDI technique.
  • Mechanism of diffusion changes: While the study identifies diffusion abnormalities, the precise mechanisms (e.g., BBB leakage, immune cell infiltration, debris accumulation) require further investigation.

Summary updated

Abstract

Abstract There has been growing attention on the effect of COVID-19 on white-matter microstructure, especially among those that self-isolated after being infected. There is also immense scientific interest and potential clinical utility to evaluate the sensitivity of single-shell diffusion MRI methods for detecting such effects. In this work, the sensitivities of three single-shell-compatible diffusion MRI modeling methods are compared for detecting the effect of COVID-19, including diffusion-tensor imaging, diffusion-tensor decomposition of orthogonal moments and correlated diffusion imaging. Imaging was performed on self-isolated patients at baseline and 3-month follow-up, along with age- and sex-matched controls. We demonstrate through simulations and experimental data that correlated diffusion imaging is associated with far greater sensitivity, being the only one of the three single-shell methods to demonstrate COVID-19-related brain effects. Results suggest less restricted diffusion in the frontal lobe in COVID-19 patients, but also more restricted diffusion in the cerebellar white matter, in agreement with several existing studies highlighting the vulnerability of the cerebellum to COVID-19 infection. These results, taken together with the simulation results, suggest that a significant proportion of COVID-19 related white-matter microstructural pathology manifests as a change in water diffusivity. Interestingly, different b-values also confer different sensitivities to the effects. No significant difference was observed in patients at the 3-month follow-up, likely due to the limited size of the follow-up cohort. To summarize, correlated diffusion imaging is shown to be a sensitive single-shell diffusion analysis approach that allows us to uncover opposing patterns of diffusion changes in the frontal and cerebellar regions of COVID-19 patients, suggesting the two regions react differently to viral infection.
BibTeX
@misc{teller2022sensitivity,
  author = {Nick Teller and Jordan A. Chad and Alexander Wong and Hayden Gunraj and Xiang Ji and Bradley J. MacIntosh and Asaf Gilboa and Eugenie Roudaia and Allison B. Sekuler and Benjamin Lam and Chris Heyn and Sandra E. Black and Simon J. Graham and J. Jean Chen},
  title = {{Sensitivity of diffusion-tensor and correlated diffusion imaging to white-matter microstructural abnormalities: application in COVID-19}},
  year = {2022},
  doi = {10.1101/2022.09.29.510004},
  howpublished = {bioRxiv},
}
Figure