Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Anatomically Constrained Tractography of the Fetal Brain

Camilo Calixto, Camilo Jaimes, Matheus D. Soldatelli, Simon K. Warfield, Ali Gholipour, Davood Karimi

TL;DR

This work addresses the challenge of accurate fetal brain tractography, which is hampered by low-quality in utero dMRI and high false-positive streamlines. It proposes an anatomically constrained tractography framework that relies on a deep learning model to segment fetal brain tissue directly in the dMRI space and to derive five-tissue-type (5TT) maps, enabling precise seed placement and termination of tractography based on diffusion-tensor–driven orientation with a sharpened dODF. The approach combines supervised and self-supervised training, uses a diffusion-tensor input suitable for routine fetal scans, and validates on independent data, achieving superior tract reconstruction (including curved tracts like optic radiations) compared with FACT while showing robustness to angle-threshold settings. The method substantially improves segmentation accuracy, tractography fidelity, and potential for reproducible quantitative assessment of the fetal structural connectome, with an open-source release to facilitate adoption and benchmarking in routine fetal dMRI studies.

Abstract

Diffusion-weighted Magnetic Resonance Imaging (dMRI) is increasingly used to study the fetal brain in utero. An important computation enabled by dMRI is streamline tractography, which has unique applications such as tract-specific analysis of the brain white matter and structural connectivity assessment. However, due to the low fetal dMRI data quality and the challenging nature of tractography, existing methods tend to produce highly inaccurate results. They generate many false streamlines while failing to reconstruct streamlines that constitute the major white matter tracts. In this paper, we advocate for anatomically constrained tractography based on an accurate segmentation of the fetal brain tissue directly in the dMRI space. We develop a deep learning method to compute the segmentation automatically. Experiments on independent test data show that this method can accurately segment the fetal brain tissue and drastically improve tractography results. It enables the reconstruction of highly curved tracts such as optic radiations. Importantly, our method infers the tissue segmentation and streamline propagation direction from a diffusion tensor fit to the dMRI data, making it applicable to routine fetal dMRI scans. The proposed method can lead to significant improvements in the accuracy and reproducibility of quantitative assessment of the fetal brain with dMRI.

Anatomically Constrained Tractography of the Fetal Brain

TL;DR

This work addresses the challenge of accurate fetal brain tractography, which is hampered by low-quality in utero dMRI and high false-positive streamlines. It proposes an anatomically constrained tractography framework that relies on a deep learning model to segment fetal brain tissue directly in the dMRI space and to derive five-tissue-type (5TT) maps, enabling precise seed placement and termination of tractography based on diffusion-tensor–driven orientation with a sharpened dODF. The approach combines supervised and self-supervised training, uses a diffusion-tensor input suitable for routine fetal scans, and validates on independent data, achieving superior tract reconstruction (including curved tracts like optic radiations) compared with FACT while showing robustness to angle-threshold settings. The method substantially improves segmentation accuracy, tractography fidelity, and potential for reproducible quantitative assessment of the fetal structural connectome, with an open-source release to facilitate adoption and benchmarking in routine fetal dMRI studies.

Abstract

Diffusion-weighted Magnetic Resonance Imaging (dMRI) is increasingly used to study the fetal brain in utero. An important computation enabled by dMRI is streamline tractography, which has unique applications such as tract-specific analysis of the brain white matter and structural connectivity assessment. However, due to the low fetal dMRI data quality and the challenging nature of tractography, existing methods tend to produce highly inaccurate results. They generate many false streamlines while failing to reconstruct streamlines that constitute the major white matter tracts. In this paper, we advocate for anatomically constrained tractography based on an accurate segmentation of the fetal brain tissue directly in the dMRI space. We develop a deep learning method to compute the segmentation automatically. Experiments on independent test data show that this method can accurately segment the fetal brain tissue and drastically improve tractography results. It enables the reconstruction of highly curved tracts such as optic radiations. Importantly, our method infers the tissue segmentation and streamline propagation direction from a diffusion tensor fit to the dMRI data, making it applicable to routine fetal dMRI scans. The proposed method can lead to significant improvements in the accuracy and reproducibility of quantitative assessment of the fetal brain with dMRI.
Paper Structure (18 sections, 7 equations, 7 figures, 2 tables, 1 algorithm)

This paper contains 18 sections, 7 equations, 7 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background and motivation
  3. Contributions of this work
  4. Methods
  5. Imaging data acquisition and preprocessing
  6. Estimation of local fiber orientation
  7. Segmentation of the brain tissue
  8. Generation of training and test labels
  9. Deep learning model development
  10. Supervised training.
  11. Self-supervised training.
  12. Streamline tracing
  13. Evaluation
  14. Results
  15. Assessment of automatic brain tissue segmentation
  16. ...and 3 more sections

Figures (7)

  • Figure 1: A schematic representation of the data processing steps to compute a whole-brain tractogram in this work. Super-resolved reconstruction of dMRI data volumes is performed using the method described in marami2016motion.
  • Figure 2: Example diffusion tensors and the corresponding streamline propagation probability maps ($p(u)$) for three different fetuses at three different gestational ages (GA).
  • Figure 3: Depiction of two diffusion tensor atlases at two different gestational weeks, namely 30 and 35 weeks. The left panel shows manually annotated labels, while the right panel shows the converted labels to the five-tissue-type (5TT) format. The 5TT image is used for anatomically constrained tractography. Please note that the bottom part of the brainstem is converted to a "gray matter" to ensure adequate termination of the streamlines in tractography computation.
  • Figure 4: Schematic representation of the deep learning model used for automatic fetal brain tissue segmentation.
  • Figure 5: Examples of segmentations predicted by our proposed method in three subjects at three different gestational ages (GA). Each panel shows the manual segmentation and the segmentation computed by the proposed method for four tissue types used in the five-tissue-type image. Please note that the last tissue type, reserved for pathological tissue smith2012anatomically, is omitted from this figure and excluded from this study, as it is absent in our cohort of healthy fetuses. GM stands for gray matter, and CSF stands for corticospinal fluid.
  • ...and 2 more figures