Denoising of diffusion MRI improves peripheral nerve conspicuity and reproducibility

Area of Science:

• Medical Imaging
• Neuroscience
• Biophysics

Background:

• Quantitative diffusion MRI is crucial for assessing peripheral nerve integrity.
• Low signal-to-noise ratio (SNR) in diffusion MRI can limit measurement accuracy.
• Developing advanced denoising techniques is essential for improving peripheral nerve imaging.

Purpose of the Study:

• To evaluate the effectiveness of principal component analysis (PCA) and generalized spherical deconvolution (genSD) for denoising diffusion MRI data.
• To assess the impact of these denoising methods on within-subject reproducibility and peripheral nerve conspicuity.
• To determine if denoising techniques can improve the accuracy of quantitative diffusion MRI measurements in peripheral nerves.

Main Methods:

• Prospective study involving seven healthy volunteers and three patients with peripheral neuropathy.
• Diffusion MRI data acquired using a 3T/multiband single-shot echo planar sequence with a 55-direction scheme.
• Image processing included "original" (no denoising), "average" (10 repetitions), "PCA-only," and "PCA + genSD" methods.
• Quantitative analysis of SNR and contrast-to-noise ratio (CNR), and qualitative assessment of nerve conspicuity by three radiologists.

Main Results:

• The "PCA + genSD" method demonstrated significantly higher SNR and CNR compared to "original" and "PCA-only" methods (P < 0.001).
• "PCA + genSD" achieved superior within-subject reproducibility (lower coefficient of variation) for tibial and common peroneal nerves.
• Radiologist assessments showed "PCA + genSD" significantly improved tibial and common peroneal nerve conspicuity compared to other methods (Odds Ratios > 1.86, P < 0.001).
• The "PCA + genSD" method did not introduce significant bias in fractional anisotropy (FA) values.

Conclusions:

• The "PCA + genSD" denoising technique effectively enhances SNR, CNR, and within-subject reproducibility in peripheral nerve diffusion MRI.
• This method improves nerve conspicuity without biasing FA measurements, making it promising for reliable peripheral nerve imaging.
• The "PCA + genSD" approach holds potential for facilitating more accurate and dependable quantitative diffusion measurements of peripheral nerves.