NA-MIC Project Weeks

Back to Projects List

Anatomically informed UKF tractography

Key Investigators

• Fan Zhang (BWH, HMS)
• Yogesh Rathi (BWH, HMS)
• Lauren J O’Donnell (BWH, HMS)

Project Description

In this project, we will include brain tissue segmentation maps into the existing unscented Kalman filter (UKF) framework (Malcolm et al 2010, IEEE TMI; Reddy et al 2016, Front. Neuroscience) to inform fiber tracking seeding and stopping. Segmentations of WM, GM and CSF are computed using a deep learning based method that performs tissue segmentation using diffusion MRI data (Zhang et al 2021, Neuroimage). The WM segmentation will be used for tractography seeding, and the GM/CSF segmentations will be used for tractography stopping.

Objective

1. Add WM/GM/CSF segmentation maps.
2. Add seeding and stopping masks.
3. Improve input check to handle multiple input options.

Approach and Plan

1. UI design for better usage of the seeding/stopping options
2. CLI help documention

Progress and Next Steps

1. Coding part of the project is almost done.
2. Push requst to the master branch of UKF
3. Waiting for the final pull requst approval
4. Decide default (suggested) settings for each option
5. Testing on more datasets other than HCP data.

Illustrations

Background and References

Zhang, F., Breger, A., Cho, K. I. K., Ning, L., Westin, C. F., O’Donnell, L. J., & Pasternak, O. (2021). Deep learning based segmentation of brain tissue from diffusion MRI. NeuroImage, 233, 117934.

Reddy, C.P. and Rathi, Y., 2016. Joint Multi-Fiber NODDI Parameter Estimation and Tractography Using the Unscented Information Filter. Frontiers in Neuroscience, 10.

Malcolm, J.G., Shenton, M.E. and Rathi, Y., 2010. Filtered multitensor tractography. IEEE transactions on medical imaging, 29(9), pp.1664-1675.