dHCP

dHCP constructs a detailed 4-dimensional connectome of the developing human brain between 20-45 weeks post-menstrual age using multi-modal MRI, including in utero and ex utero neonatal fMRI from over 1,000 subjects.

Key Features:

• Automated resting-state functional processing framework: A pipeline tailored for neonatal fMRI that addresses low and variable tissue contrast and high head motion.
• Integrated Slice-to-Volume Motion Correction: Corrects motion artifacts prevalent in neonatal datasets to improve spatial alignment of fMRI slices.
• Dynamic Susceptibility Distortion Correction: Reduces distortions from magnetic field inhomogeneities to enhance image fidelity.
• Robust Multimodal Registration Approach: Ensures precise alignment across different imaging modalities for integrated analysis.
• Bespoke ICA-based Denoising: Uses Independent Component Analysis to separate signal from noise and improve detection of resting-state networks (RSNs).
• Automated Quality Control Framework: Provides rigorous quality assurance to minimize failure rates and select high-quality data for analysis.
• Large multi-modal neonatal cohort: Processes data from over 1,000 subjects scanned in utero and ex utero across 20-45 weeks post-menstrual age.
• Cohort-level performance improvements: Demonstrated increases in signal-to-noise ratio (SNR) and detection of high-quality RSNs on a large evaluated cohort.

Scientific Applications:

• 4D developing human connectome construction: Enables mapping of structural and functional connectivity changes over 20-45 weeks post-menstrual age.
• Neonatal resting-state network analysis: Facilitates identification and characterization of RSNs in early life using denoised, motion- and distortion-corrected fMRI.
• Developmental neuroimaging studies: Provides quality-controlled multi-modal MRI data for studies of brain development in utero and ex utero.
• Method validation and cohort analyses: Supports evaluation of preprocessing strategies via cohort-level SNR and RSN detection metrics.

Methodology:

Processing steps include integrated slice-to-volume motion correction, dynamic susceptibility distortion correction, robust multimodal registration, bespoke ICA-based denoising, and an automated quality control framework.