SHORE-based detection and imputation of dropout in diffusion MRI.

Area of Science:

• Medical Imaging
• Neuroimaging
• Biophysics

Background:

• Signal dropout in diffusion MRI (dMRI) arises from bulk motion during diffusion encoding.
• Uncorrected dropout is misinterpreted as high diffusivity, compromising data integrity.
• Existing correction methods can be computationally intensive.

Purpose of the Study:

• To develop an automated method for detecting and imputing signal dropout in dMRI.
• To improve the accuracy and reliability of dMRI data analysis by addressing motion-induced artifacts.

Main Methods:

• Utilized a Simple Harmonic Oscillator-based Reconstruction and Estimation (SHORE) model to derive an outlier score.
• Identified dropout by detecting measurements significantly lower than SHORE predictions.
• Replaced detected outliers with values predicted by a secondary SHORE fit using inlier data.

Main Results:

• The proposed method reliably detects and accurately imputes dropout in simulated dMRI data.
• Plausible results were achieved in corrupted in vivo dMRI measurements.
• Demonstrated significantly lower computational effort compared to existing methods.

Conclusions:

• SHORE-based outlier scoring provides a fast and accurate approach for dMRI dropout detection and imputation.
• The method requires multi-shell or diffusion spectrum imaging (DSI) data but is model-agnostic for subsequent analysis.
• This technique enhances the robustness of diffusion MRI analyses without relying on specific modeling approaches.