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, Hideyuki Okano, Tetsuo Yamamori, Yoko Yamaguchi, Shin Ishii, Kenji Doya

Abstract

Diffusion-weighted magnetic resonance imaging (dMRI) allows non-invasive investigation of whole-brain connectivity, which can potentially help to reveal the brain's global network architecture and 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 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. We considered four criteria for goodness of fiber tracking: distance-weighted coverage, true/false positive ratio, projection coincidence, and commissural passage, applied using a multi-objective optimization algorithm. We implemented a variant of non-dominated sorting genetic algorithm II (NSGA-II) to optimize five major parameters of a global fiber-tracking algorithm over multiple brain samples in parallel. Using optimized parameters compared to the default parameters, dMRI-based fiber tracking performance was significantly improved, while minimizing false positives and impossible cross-hemisphere connections. Parameters optimized for 10 tracer injection sites showed good generalization capability for other brain samples. These results demonstrate the importance of data-driven adjustment of fiber-tracking algorithms and support the validity of dMRI-based tractography, if appropriate adjustments are employed.

Introduction

Mapping the brain's structural connectivity, or the "connectome," is fundamental to understanding neural function and pathology. Diffusion-weighted MRI (dMRI) is the primary tool for non-invasive connectomics in humans. However, dMRI tractography—the computational reconstruction of white matter pathways—is often limited by high false-positive rates and poor sensitivity to long-range connections. Historically, the parameters governing these algorithms (e.g., curvature thresholds, fiber length) have been tuned heuristically, often by "eye-balling" results. This lack of objective grounding leads to inconsistent and potentially biased findings. This study introduces a framework to mathematically optimize dMRI tracking parameters using high-resolution neural tracers as a "ground-truth" reference.

Methodology: A Data-Driven Framework

The researchers utilized the Japan Brain/MINDS Project dataset, which uniquely provides both dMRI and anterograde neural tracer data from the same common marmoset (Callithrix jacchus) brains. The framework maps dMRI data into a standard brain space and compares the resulting tractograms against 3D reconstructions of fluorescent tracer signals from specific injection sites.

Overview of the data-driven framework for optimizing dMRI tractography using neural tracer data as a reference.

The optimization process evaluates four distinct "goodness" criteria:

  1. Distance-weighted coverage ($f_1$): Ensures the reconstruction of long-range fibers.
  2. True/False positive ratio ($f_2$): Balances sensitivity with noise suppression.
  3. Projection coincidence ($f_3$): Measures the correlation between dMRI-predicted target areas and tracer-identified regions.
  4. Commissural passage ($f_4$): An anatomical constraint ensuring fibers only cross hemispheres through biologically valid structures like the corpus callosum.

Multi-Objective Optimization

The team employed a variant of the Non-dominated Sorting Genetic Algorithm II (NSGA-II) to optimize five major tracking parameters. To handle biological variability and accelerate convergence, they implemented a "Champion Sharing" strategy. In this parallel processing setup, optimization tasks for different brains shared their best-performing parameter sets ("champions"), ensuring the final parameters were robust across individuals. This approach addresses the fundamental difference between methods: tracers are unidirectional, while dMRI is bidirectional, necessitating a framework that allows for "reasonable" false positives representing incoming fibers the tracer wouldn't see.

Results and Validation

The optimization led to substantial improvements across all metrics. Most notably, the average fiber length increased from approximately 10mm to 17mm, and voxel-wise coverage of axonal tracts jumped from 0.9% to 15%.

Evolution of fiber length and number throughout the optimization process, showing significant improvement over generations.

The correlation between dMRI results and tracer ground truth ($f_3$) improved from 40% to 68%. Crucially, parameters optimized on a training set of 10 brains generalized effectively to 6 unseen test brains, suggesting that the optimized settings reflect fundamental biological properties of the marmoset brain rather than overfitting to specific samples.

Conclusion

This work demonstrates that dMRI tractography can be highly reliable when its parameters are optimized against biological reality. While dMRI still cannot determine the directionality of connections—a limitation inherent to water diffusion physics—this data-driven approach significantly mitigates the sensitivity and false-positive issues that have historically hampered the field. Future refinements could incorporate additional data types, such as myelin-based axonal orientation, to further enhance the accuracy of non-invasive brain mapping.

Summary updated

Abstract

Diffusion-weighted magnetic resonance imaging (dMRI) allows non-invasive investigation of whole-brain connectivity, which can potentially help to reveal the brain's global network architecture and 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 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. We considered four criteria for goodness of fiber tracking: distance-weighted coverage, true/false positive ratio, projection coincidence, and commissural passage, applied using a multi-objective optimization algorithm. We implemented a variant of non-dominated sorting genetic algorithm II (NSGA-II) to optimize five major parameters of a global fiber-tracking algorithm over multiple brain samples in parallel. Using optimized parameters compared to the default parameters, dMRI-based fiber tracking performance was significantly improved, while minimizing false positives and impossible cross-hemisphere connections. Parameters optimized for 10 tracer injection sites showed good generalization capability for other brain samples. These results demonstrate 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 Hideyuki Okano and Tetsuo Yamamori and Yoko Yamaguchi 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},
}

Loading related content...

Figure