Significant clusters obtained from permutation_cluster_1samp_test

💬 Support & Discussions
1

If you have a question or issue with MNE-Python, please include the following info:

  • MNE-Python version: 0.24.dev0
  • operating system: linux

Hi! guys,

long time no see :grin:

I was trying to find the significant clusters of decoding performance obtained from GeneralizingEstimator(). I’m confused about the cluster the permutation_cluster_1samp_test() gave, here is my code snippet:

# significance tests
globals()[f'clust_fea_{fea}'] = permutation_cluster_1samp_test(globals()[f'arr_cate_spec_fea_{fea}'], n_permutations=1024, tail=1, n_jobs=12, seed=6)

# outputs of the significance tests above
globals()[f't_obs_fea_{fea}'] = globals()[f'clust_fea_{fea}'][0] # (n_time_points,n_time_points), (700,700)
globals()[f'clusters_fea_{fea}'] = globals()[f'clust_fea_{fea}'][1] 
globals()[f'cluster_pv_fea_{fea}'] = globals()[f'clust_fea_{fea}'][2]

globals()[f'arr_cate_spec_fea_{fea}'] is a 3-dimensional array of generalized decoding performance with the shape of (40,700,700), 40 is the number of subjects and 700 is the number of time points, and the obtained globals()[f'clusters_fea_{fea}'] is a tuple which I converted to an array, its shape is (2, 490000), here is its content:

In [132]: globals()[f'sig_clust_fea_{fea}']
Out[132]: 
(array([  0,   0,   0, ..., 699, 699, 699]),
 array([  0,   1,   2, ..., 697, 698, 699]))

I don’t know what’s this array stands for, the indices of significant time points? but why the size is 490000 instead of 700?

Looking forward to hearing from you guys!
Thanks a lot! :sparkling_heart: :sparkling_heart: :sparkling_heart:

2

Hi @YuZhou,
I think the tuple elements tells you the coordinates (for each dimension) of points belonging to clusters. It might be more intuitive for you if you specify out_type='mask' then the cluster membership will be specified with boolean arrays - one per cluster.
BTW, I think you would be better off using a dictionary instead of globals()[var_name] :slight_smile:

1 Like
3

Thanks a lot for your suggestions! Mikolaj, I’ll try them immediately!
Best