Optimization and validation of diffusion MRI-based fiber tracking with neural tracer data as a reference

Carlos Enrique Gutierrez, Henrik Skibbe, Ken Nakae, Hiromichi Tsukada, Jean Liénard, Akiya Watakabe, Junichi Hata, Marco Reisert, Alexander Woodward, Yoko Yamaguchi, Tetsuo Yamamori, Hideyuki Okano, Shin Ishii, Kenji Doya

Abstract

Diffusion-weighted magnetic resonance imaging (dMRI) allows non-invasive investigation of whole-brain connectivity. However, the reliability of dMRI-based tractography is often limited by low sensitivity, false positives, and heuristic parameter tuning. We present a data-driven framework using neural tracer data from the Brain/MINDS Project as a reference to optimize tracking parameters via a multi-objective genetic algorithm. Our results demonstrate significant improvements in fiber length, coverage, and target correlation, particularly for global tracking algorithms. The optimized parameters generalize well across subjects, supporting the validity of dMRI-based tractography when appropriately adjusted.

Introduction

Mapping the brain's "wiring diagram" is essential for understanding global network architecture and neurological disorders. Diffusion MRI (dMRI) is the primary non-invasive tool for this task, yet its accuracy is frequently questioned. Tractography algorithms often struggle with long-range connections and "hallucinated" false positives. Furthermore, the parameters governing these algorithms are typically chosen heuristically. This study addresses these challenges by leveraging a unique dataset of concurrent dMRI and invasive neural tracer data in common marmosets to establish a data-driven optimization framework.

Methods: A Multi-Objective Optimization Framework

The core of our approach is the use of the Non-dominated Sorting Genetic Algorithm II (NSGA-II) to perform multi-objective optimization (MOO). Instead of seeking a single optimal parameter set, the framework identifies a Pareto front—a set of solutions representing the best possible trade-offs between competing metrics, such as sensitivity (True Positive Rate) and specificity (False Positive Rate).

Overview of the optimization framework for iFOD2 and Global tracking algorithms, illustrating the integration of dMRI data with neural tracer "ground truth."

We evaluated two tracking approaches: a seed-based algorithm (iFOD2) and a global tracking algorithm. For the latter, we introduced refined objectives including distance-weighted coverage to favor long-range fibers and anatomical constraints (commissural passage) to penalize biologically impossible cross-hemisphere connections. The framework also accounts for the inherent directionality gap between tracers and dMRI by allowing for "reasonable" false positives that represent incoming fibers to the injection site.

Results and Validation

The optimization process led to dramatic improvements over default software settings. In the global tracking experiment, average fiber length increased from 10 mm to 17 mm, and voxel-wise coverage of axonal tracts improved from 0.9% to 15%. The correlation between dMRI-derived projection areas and tracer-defined targets rose from 40% to 68%.

Scatter plots showing the distribution of tracking performance across various parameter sets, highlighting the Pareto front for individual and group objectives.

Importantly, parameters optimized on a training set of 10 brains showed excellent generalization when applied to a separate test set of 6 brains. This suggests that the optimized settings capture fundamental anatomical properties rather than subject-specific noise.

Discussion

Our findings underscore the critical importance of data-driven parameter adjustment in fiber tracking. While the specific parameters found for marmosets may not translate directly to the human brain due to scale differences, the framework itself is highly generalizable. By incorporating biological "common sense" and high-resolution reference data, we can significantly enhance the reliability of non-invasive connectivity mapping. Future work will expand this approach to include more diverse injection sites and further refine the "ground truth" models as more complete tracer datasets become available.

Summary updated

Abstract

Diffusion-weighted magnetic resonance imaging (dMRI) allows non-invasive investigation of whole-brain connectivity, which can reveal the brain's global network architecture and also abnormalities involved in neurological and mental disorders. However, the reliability of connection inferences from dMRI-based fiber tracking is still debated, due to low sensitivity, dominance of false positives, and inaccurate and incomplete reconstruction of long-range connections. Furthermore, parameters of tracking algorithms are typically tuned in a heuristic way, which leaves room for manipulation of an intended result. Here we propose a general data-driven framework to optimize and validate parameters of dMRI-based fiber tracking algorithms using neural tracer data as a reference. Japan's Brain/MINDS Project provides invaluable datasets containing both dMRI and neural tracer data from the same primates. A fundamental difference when comparing dMRI-based tractography and neural tracer data is that the former cannot specify the direction of connectivity; therefore, evaluating the fitting of dMRI-based tractography becomes challenging. The framework implements multi-objective optimization based on the non-dominated sorting genetic algorithm II. Its performance is examined in two experiments using data from ten subjects for optimization and six for testing generalization. The first uses a seed-based tracking algorithm, iFOD2, and objectives for sensitivity and specificity of region-level connectivity. The second uses a global tracking algorithm and a more refined set of objectives: distance-weighted coverage, true/false positive ratio, projection coincidence, and commissural passage. In both experiments, with optimized parameters compared to default parameters, fiber tracking performance was significantly improved in coverage and fiber length. Improvements were more prominent using global tracking with refined objectives, achieving an average fiber length from 10 to 17 mm, voxel-wise coverage of axonal tracts from 0.9 to 15%, and the correlation of target areas from 40 to 68%, while minimizing false positives and impossible cross-hemisphere connections. Optimized parameters showed good generalization capability for test brain samples in both experiments, demonstrating the flexible applicability of our framework to different tracking algorithms and objectives. These results indicate the importance of data-driven adjustment of fiber tracking algorithms and support the validity of dMRI-based tractography, if appropriate adjustments are employed.
BibTeX
@inproceedings{gutierrez2020optimization,
  author = {Carlos Enrique Gutierrez and Henrik Skibbe and Ken Nakae and Hiromichi Tsukada and Jean Liénard and Akiya Watakabe and Junichi Hata and Marco Reisert and Alexander Woodward and Yoko Yamaguchi and Tetsuo Yamamori and Hideyuki Okano and Shin Ishii and Kenji Doya},
  title = {{Optimization and validation of diffusion MRI-based fiber tracking with neural tracer data as a reference}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-78284-4},
}
Figure