Validation of diffusion MRI models and tractography algorithms using chemical tracing

Brain circuitry is still poorly understood, posing a challenge to the validation of diffusion MRI (dMRI) tractography. Many aspects of tractography algorithms, such as their choice of diffusion model or deterministic vs. probabilistic approach, can impact on their performance. Therefore, beyond a qualitative validation, we need quantitative metrics for comparing and optimizing these algorithms. In this work we perform a systematic evaluation of different diffusion models and tractography algorithms by assessing their accuracy with respect to chemical tracing in macaques. We find that the combination of probabilistic tractography and GQI/DSI model yields the best results, and that accuracy does not always improve with higher angular resolution.