Skip to content

Latest commit

 

History

History
71 lines (51 loc) · 4.05 KB

README.md

File metadata and controls

71 lines (51 loc) · 4.05 KB

Cortico_Cereb_Connectivity

This repository includes the code to train and test cortico-cerebellar-connectivity models, as described in King et al. (2023). The original code underlying that paper is maintained in a different repository (https://github.com/maedbhk/cerebellum_connectivity). This repository is a cleaned-up version of the original code, that also uses the FunctionalFusion framework (https://github.com/DiedrichsenLab/Functional_Fusion) to integrate knowledge across different datasets.

Results with the new, updated connectivity models are reported in Nettekoven et al. (2024), and are being used in Shahshahani et al. (2024).

Dependencies

The code requires the following packages:

  • numpy
  • sklearn
  • nibabel
  • FunctionalFusion (for easy mapping of atlas spaces)

Models

The data-directory contains the connectivity weights in form of pdconn-cifti files. The neo-cortex is parcellated in a specific way. The cerebellum is voxel wise

A example of how to use these models to make predictions about new data is given in the notebooks/0.application_example.ipynb notebooks.

The models from Nettekoven et al. (2024):

  • Nettekoven_2024_MDTB_L2: L2 model (alpha = exp(8)), SUIT3 space, Icosahedron1002, full MDTB dataset
  • Nettekoven_2024_Demand_L2: L2 model (alpha = exp(8)), SUIT3 space, Icosahedron1002, full Multiple-demand dataset
  • Nettekoven_2024_HCP_L2: L2 model (alpha = exp(-2)), SUIT3 space, Icosahedron1002, resting state from 100 unrelated HCP subjects
  • Nettekoven_2024_IBC_L2: L2 model (alpha = exp(6)), SUIT3 space, Icosahedron1002, full IBC dataset
  • Nettekoven_2024_Nishimoto_L2: L2 model (alpha = exp(10)), SUIT3 space, Icosahedron1002, full Nishimoto dataset
  • Nettekoven_2024_Somatotopic_L2: L2 model (alpha = exp(8)), SUIT3 space, Icosahedron1002, full Somatotopic dataset
  • Nettekoven_2024_Fusion_L2: SUIT3 space, Icosahedron1002, Average model from above (excluding HCP)

The models used in Shahshahani et al. (2024). The first two models are very close to the models reported in King et al. (2023), but are trained in SUIT3 space.

  • Shahshahani_2024_MDTB_L2: L2 model (alpha = 8), SUIT3 space, Icosahedron1002, Task set A from MDTB dataset
  • Shahshahani_2024_MDTB_L1: L1 model (alpha = -5), SUIT3 space, Icosahedron1002, Task set A from MDTB dataset
  • Shahshahani_2024_Fusion_L2: SUIT3 space, Icosahedron1002, Average model from L2 models above (excluding WMFS and HCP)

References

King, M., Shahshahani, L., Ivry, R. B., & Diedrichsen, J. (2023). A task-general connectivity model reveals variation in convergence of cortical inputs to functional regions of the cerebellum. eLife, 12. https://doi.org/10.7554/eLife.81511

Nettekoven, C., Zhi, D., Ladan, S., Pinho, A. L., Saadon Grosmannn, N., Buckner, R., & Diedrichsen, J. (2023). A hierarchical atlas of the human cerebellum for functional precision mapping. BioRviv.

Shahshahani, L., King, M., Nettekoven, C., Ivry, R., & Diedrichsen, J. (2023). Selective recruitment: Evidence for task-dependent gating of inputs to the cerebellum. bioRxiv, 2023.01.25.525395.

Details on Fusion model in Nettekoven et al. (2024)

Models were trained evaluated ccc.run_model, which is called from ccc.scripts.script_train_eval_models.py

Models are then fused (i.e. simply averaged) using ccc.scripts.script_fuse_models.py

  • Model 4: Demand, HCP and MDTB
  • Model 5: all datasets including HCP
  • Model 6: all datasets excluding HCP
  • Model 7: all datasets excluding HCP and Somatotopic

The final model evaluation results reported in the paper can be found in ccc.notebooks.6.Evaluate_model_int.ipynb.

To summarize the connectivity pattern by cerebellar regions:

import cortico_cereb_connectivity.scripts.script_summarize_weights as csw
csw.make_weight_map('Fusion','06',method='L2Regression')

To summarize further by cortical ROI: T = csw.make_weight_table(dataset="Fusion",extension="06",cortical_roi="")

Summary figures (by MSHBM_Prior_15_fsLR32) notebooks/cortical_connectivity.ipynb

Full connectivity maps: notebooks/connectivity_weights.ipynb (Fig S5 & Fig S6)