Revisiting the interpretation of axon diameter mapping using higher-order signal representations

Area of Science:

• Neuroimaging
• Biophysics
• Medical Physics

Background:

• Diffusion-weighted Magnetic Resonance Imaging (dMRI) is explored for non-invasive axon diameter measurement.
• High b-value dMRI signals are often modeled assuming water confined in "stick" geometries (axons).
• Current axon diameter mapping assumes deviations from power law scaling are exclusively axonal.

Purpose of the Study:

• To investigate the validity and robustness of axon diameter mapping techniques.
• To evaluate the impact of experimental variations, specifically RISH order, on diameter measurements.
• To explore potential confounding effects from non-axonal cellular structures like glial processes.

Main Methods:

• Comparison of axon diameter mapping using zeroth-order (spherical mean) and second-order (spherical variance) RISH features.
• Introduction of a novel, computationally efficient log-linear estimator for axon diameter mapping.
• Utilized Monte Carlo simulations to model signal contributions from different cellular structures.

Main Results:

• Axon diameter measurements were found to vary significantly with the RISH order used.
• The novel log-linear estimator demonstrated sensitivity to RISH order variations.
• Simulations suggest glial processes could contribute to high b-value dMRI signals, mimicking axonal properties.

Conclusions:

• The assumption that high b-value dMRI deviations are solely axonal may be flawed.
• Non-axonal cellular structures, such as glial processes, can confound dMRI-based axon diameter mapping.
• Increased caution is warranted when interpreting dMRI-derived "axon" diameters, considering potential glial contributions.