neurostuff / NiMARE

Coordinate- and image-based meta-analysis in Python
https://nimare.readthedocs.io
MIT License
178 stars 57 forks source link

Meta-analytic activation modeling-based parcellation #262

Open tsalo opened 4 years ago

tsalo commented 4 years ago

Add MAMP algorithm, as described in Yang et al. (2016).

References

Yang, Yong, et al. “Identifying functional subdivisions in the human brain using meta-analytic activation modeling-based parcellation.” Neuroimage 124 (2016): 300-309. https://doi.org/10.1016/j.neuroimage.2015.08.027

tsalo commented 4 years ago

Here is my understanding of the method:

MAMP works similarly to CBP, but skips the step of performing a MACM for each voxel. Here are the steps:

  1. Create an MA map for each study in the dataset.
  2. Concatenate MA maps across studies to create a 4D dataset.
  3. Extract values across studies for voxels in mask, resulting in n_voxels X n_studies array.
  4. Correlate "study series" between voxels to generate n_voxels X n_voxels correlation matrix.
  5. Convert correlation coefficients to correlation distance (1 -r) values.
  6. Perform clustering on correlation distance matrix.