HAITCH: A Framework for Distortion and Motion Correction in Fetal Multi-Shell Diffusion-Weighted MRI

TL;DR

HAITCH delivers a comprehensive framework for distortion and motion correction in fetal multi-shell diffusion-weighted MRI by integrating a optimized multi-shell HARDI acquisition with a dynamic, model-free reconstruction that jointly estimates motion and SHORE-based diffusion coefficients. The method employs a dual-echo EPI sequence for dynamic distortion correction, three-layer outlier weighting, and iterative refinement of motion parameters and diffusion representations, followed by atlas-space normalization and tractography. Validation on real fetal data shows superior distortion correction, robust motion handling, and anatomically plausible diffusion metrics across diverse ages and motion levels, enabling more reliable fetal brain microstructure and connectivity analyses. By providing an open-source toolkit, HAITCH lowers barriers to advanced fetal dMRI analyses and promotes reproducibility and broader adoption in in-utero neuroimaging studies.

Abstract

Diffusion magnetic resonance imaging (dMRI) is pivotal for probing the microstructure of the rapidly-developing fetal brain. However, fetal motion during scans and its interaction with magnetic field inhomogeneities result in artifacts and data scattering across spatial and angular domains. The effects of those artifacts are more pronounced in high-angular resolution fetal dMRI, where signal-to-noise ratio is very low. Those effects lead to biased estimates and compromise the consistency and reliability of dMRI analysis. This work presents HAITCH, the first and the only publicly available tool to correct and reconstruct multi-shell high-angular resolution fetal dMRI data. HAITCH offers several technical advances that include a blip-reversed dual-echo acquisition for dynamic distortion correction, advanced motion correction for model-free and robust reconstruction, optimized multi-shell design for enhanced information capture and increased tolerance to motion, and outlier detection for improved reconstruction fidelity. The framework is open-source, flexible, and can be used to process any type of fetal dMRI data including single-echo or single-shell acquisitions, but is most effective when used with multi-shell multi-echo fetal dMRI data that cannot be processed with any of the existing tools. Validation experiments on real fetal dMRI scans demonstrate significant improvements and accurate correction across diverse fetal ages and motion levels. HAITCH successfully removes artifacts and reconstructs high-fidelity fetal dMRI data suitable for advanced diffusion modeling, including fiber orientation distribution function estimation. These advancements pave the way for more reliable analysis of the fetal brain microstructure and tractography under challenging imaging conditions.

Figures (8)

• Figure 1: HAITCH Framework Processing Stages. This flowchart illustrates the key steps involved in the HAITCH Framework. Boxes with bold borders refer to our main contributions. Grey boxes outline the acquisition stages, including the sampling scheme and the dual-echo sequence. Green boxes detail the pre-processing steps, distortion correction, B1 bias field correction, and fetal brain segmentation. Blue boxes refer to the iterative approach of motion correction and image reconstruction that includes slice & voxel weighting, model-free reconstruction, signal basis update, signal prediction, registration, and updating the gradient table. Purple boxes showcase the post-processing steps, encompassing T2-weighted image reconstruction, spatial normalization for atlas registration, and streamlined tractography. Refer to Section \ref{['sec:methods']} for details on the sampling scheme, dual-echo sequence, motion correction, and dynamic distortion correction. Section \ref{['sec:implementation']} discusses pre-processing, post-processing, and implementation specifics. Boxes with bold borders refer to our main contributions.
• Figure 2: (Top): Schematic diagram of the modified spin-echo EPI sequence. This sequence acquires two echoes (TE1 and TE2) with reversed phase encoding directions. Despite the extended readout, the dual-echo sequence maintains signal-to-noise ratio (SNR) efficiency comparable to a conventional EPI scan. Temporal proximity between the echoes (< 50 ms) ensures minimal motion artifacts between acquisitions. (Bottom): Illustration of the opposing susceptibility-induced distortions in the two echoes for an example. The first echo (TE1) exhibits stretching (arrow), while the second echo (TE2) shows signal pile-up in the corresponding location. This key feature allows for dynamic field map estimation and correction of geometric distortions even in the presence of fetal motion. Details of the implementation are discussed in Section \ref{['sec:implementation']}.
• Figure 3: Quantitative Evaluation of Distortion Correction Techniques using SSIM and PSNR. Boxplots summarize the distribution of SSIM (left) and PSNR (right) values across 27 subjects. Results are presented for raw data and data processed with: Slice-wise correction, Volume-wise correction (HAITCH approach), and Static field map correction obtained from the three echo reversal correction methods. Red boxes represent b=0 images, while blue boxes display diffusion-weighted images ($b>0$). Higher SSIM and PSNR values indicate better image quality and reduced distortion. As evident from the plots, the volume-wise dynamic correction method achieved the best results across both metrics and image types.
• Figure 4: Example fetal dMRI scans before and after motion correction. The left two columns display axial, coronal, and sagittal views of the raw data (pre-processed up to B1 bias field correction step) (Subject A) and corresponding motion-corrected data, respectively. Each row represents a specific volume (index: 5, 8, 15, 18, 27, 30, 31 from top to bottom). The right two columns show similar data for Subject B. Each row represents a specific volume (index: 7, 37, 38, 50, 58, 60, 87 from top to bottom).
• Figure 5: Estimated motion parameters over time and slice weights for Subject B: The top two panels show the estimated motion parameters: translation (mm) and rotation (Euler angles). The bottom panel displays the slice weights calculated using the modified Z-score method. Peaks in motion parameters coincide with low slice weights (outliers).