Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-dimensional Parameter Space Exploration for Streamline-specific Tractography

Ruben Vink, Anna Vilanova, Maxime Chamberland

TL;DR

Tractography currently relies on fixed, bundle- and dataset-specific parameters, which can fail in heterogeneous or pathological brains. The authors introduce streamline-specific parameter tracking (SSP), a probabilistic framework that samples parameters per streamline to explore the multi-dimensional parameter space, implemented on a tracking algorithm akin to iFOD2 and evaluated on synthetic ISMRM2015 data and real clinical datasets. SSP reveals how parameter distributions shape subparts of bundles, enables potential speedups, and demonstrates behavior near tumors compared with iFOD2 and TractSeg, highlighting both promise and limitations. This approach paves the way for automatic or clinician-guided parameter selection in tractography, potentially improving robustness and clinical applicability while requiring adaptations to ensure anatomical plausibility.

Abstract

One of the unspoken challenges of tractography is choosing the right parameters for a given dataset or bundle. In order to tackle this challenge, we explore the multi-dimensional parameter space of tractography using streamline-specific parameters (SSP). We 1) validate a state-of-the-art probabilistic tracking method using per-streamline parameters on synthetic data, and 2) show how we can gain insights into the parameter space by focusing on streamline acceptance using real-world data. We demonstrate the potential added value of SSP to the current state of tractography by showing how SSP can be used to reveal patterns in the parameter space.

Multi-dimensional Parameter Space Exploration for Streamline-specific Tractography

TL;DR

Tractography currently relies on fixed, bundle- and dataset-specific parameters, which can fail in heterogeneous or pathological brains. The authors introduce streamline-specific parameter tracking (SSP), a probabilistic framework that samples parameters per streamline to explore the multi-dimensional parameter space, implemented on a tracking algorithm akin to iFOD2 and evaluated on synthetic ISMRM2015 data and real clinical datasets. SSP reveals how parameter distributions shape subparts of bundles, enables potential speedups, and demonstrates behavior near tumors compared with iFOD2 and TractSeg, highlighting both promise and limitations. This approach paves the way for automatic or clinician-guided parameter selection in tractography, potentially improving robustness and clinical applicability while requiring adaptations to ensure anatomical plausibility.

Abstract

One of the unspoken challenges of tractography is choosing the right parameters for a given dataset or bundle. In order to tackle this challenge, we explore the multi-dimensional parameter space of tractography using streamline-specific parameters (SSP). We 1) validate a state-of-the-art probabilistic tracking method using per-streamline parameters on synthetic data, and 2) show how we can gain insights into the parameter space by focusing on streamline acceptance using real-world data. We demonstrate the potential added value of SSP to the current state of tractography by showing how SSP can be used to reveal patterns in the parameter space.
Paper Structure (19 sections, 1 equation, 5 figures, 2 tables, 1 algorithm)

This paper contains 19 sections, 1 equation, 5 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Methods
  3. Tracking Algorithm
  4. Parameters
  5. Automatic Validation of Tracking Algorithm
  6. Streamline-specific Experiments
  7. Per-streamline parameter distributions
  8. Clinical datasets
  9. Results
  10. Validation of Tracking Algorithm
  11. Parameter Space Exploration
  12. Clinical datasets
  13. Discussion
  14. Experiments
  15. Parameter space exploration
  16. ...and 4 more sections

Figures (5)

  • Figure 1: Trackings of the left CST of the healthy HCP subject. From left to right; results from iFOD2 with the default parameters for the pipeline ETZ-pipeline. Then the result of SSPT with the same ROIs. The last two images show a representation of the cluster representatives. The circled cluster numbers correspond to the data shown in Figure \ref{['fig:params_clusters']}.
  • Figure 2: Histograms of the amount of streamlines out of 10000 that were accepted with a specific parameter in the left CST of the healthy HCP subject.
  • Figure 3: Histograms for specific clusters as shown in Figure \ref{['fig:cst']} (left CST of healthy HCP subject). On the right a close up of the cluster representatives is shown.
  • Figure 4: All three slices are from the T1 image of patient A with a cross-sectional slab of the tracks of the right AF generated by iFOD2 (red), TractSeg (blue), and SSPT (green). The bottom right shows each result again in a 3D view, with a dashed circle highlighting the location of the tumor.
  • Figure 5: The top row shows sagittal and coronal slices of the T1 image of patient B with a cross-sectional slab of the tracks of the left CST generated by iFOD2 (red), TractSeg (blue), and SSPT (green). The bottom row shows the same tracks in a 3D view with a dashed circle highlighting the part of the bundle nearest to the tumor.