Anatomically constrained tractography of the fetal brain

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Medical Image Analysis

Background:

• Diffusion-weighted Magnetic Resonance Imaging (dMRI) is vital for in utero fetal brain studies.
• Streamline tractography, a key dMRI computation, aids white matter analysis but suffers from inaccuracies in fetal data.
• Existing fetal tractography methods yield many false streamlines and fail to reconstruct major white matter tracts.

Purpose of the Study:

• To develop an anatomically constrained tractography method for fetal brain dMRI.
• To improve the accuracy of streamline tractography by using precise fetal brain tissue segmentation.
• To enable reliable reconstruction of fetal white matter tracts for research.

Main Methods:

• A deep learning approach was developed for automatic segmentation of fetal brain tissue directly in dMRI space.
• The method utilizes a diffusion tensor fit to dMRI data for tissue segmentation and streamline propagation direction inference.
• Anatomically constrained tractography was applied using the generated segmentation.

Main Results:

• The deep learning segmentation method accurately segmented fetal brain tissue on independent test data.
• Tractography results showed drastic improvements, enabling reconstruction of highly curved tracts like optic radiations.
• The method demonstrated applicability to routine fetal dMRI scans.

Conclusions:

• The proposed method significantly enhances the accuracy of fetal brain streamline tractography.
• Accurate segmentation directly in dMRI space is crucial for reliable fetal white matter tract reconstruction.
• This approach facilitates advanced studies of fetal brain white matter development using dMRI.