Comparison of distortion correction preprocessing pipelines for DTI in the upper limb

Area of Science:

• Neuroimaging
• Diffusion Tensor Imaging (DTI)
• Peripheral Nerve Imaging

Background:

• Diffusion Tensor Imaging (DTI) is crucial for assessing nerve health.
• Echo-planar imaging (EPI) is common for DTI but prone to artifacts.
• Preprocessing DTI data can mitigate artifacts, but pipeline performance varies.

Purpose of the Study:

• To compare the performance of different DTI preprocessing pipelines.
• To evaluate their impact on diffusion parameter estimates in peripheral nerves.
• To assess the variability introduced by distinct correction methods.

Main Methods:

• DTI data from healthy volunteers' upper limbs were acquired at 3T.
• Four pipelines (FSL TOPUP & eddy, FSL TOPUP, DSI Studio, TORTOISE) were used for correction.
• DTI metrics (fractional anisotropy, radial diffusivity) were extracted and compared.

Main Results:

• Preprocessing improved geometric similarity between DTI and T1w images by 1%.
• DSI Studio and TORTOISE yielded significantly different fractional anisotropy and radial diffusivity estimates compared to FSL TOPUP & eddy.
• Agreement of DTI metrics across pipelines was poor.

Conclusions:

• DTI preprocessing enhances geometric similarity in upper limb nerve imaging.
• The selection of a DTI preprocessing pipeline significantly impacts diffusion parameter estimates.
• This variability has clinical implications for peripheral nerve assessment.