Hi everyone,
I am trying to run a spatial cluster-based permutation test to compare the abundance of alpha activity between two groups using MNE-Python. Since I only have one time point, I attempted to use both:
permutation_cluster_test()
spatio_temporal_cluster_test() (by adding one time-point as a newaxis)
I am creating the adjacency from my EEG montage:
montage = make_standard_montage(‘biosemi64’)
info = create_info(channels, sfreq=1, ch_types=‘eeg’)
info.set_montage(montage)
Create adjacency matrix based on the spatial layout of the electrodes
adjacency, ch_names = find_ch_adjacency(info, ch_type=‘eeg’)
But when running:
Perform the test
T_obs, clusters, cluster_p_values, H0 = permutation_cluster_test(
X, stat_fun=None, n_permutations=1000, threshold=None, adjacency=adjacency,
tail=0, n_jobs=1, seed=42
)
I got the following error:
Using a threshold of 3.942303
stat_fun(H1): min=0.14116639236174247 max=6.98171703693277
Running initial clustering …
C:\Users\tvanneau\AppData\Local\Temp\ipykernel_39684\3666213350.py:1: RuntimeWarning: Ignoring argument “tail”, performing 1-tailed F-test
T_obs, clusters, cluster_p_values, H0 = permutation_cluster_test(
ValueError: Buffer dtype mismatch, expected ‘ITYPE_t’ but got ‘long long’
Exception ignored in: ‘scipy.sparse.csgraph._traversal._connected_components_undirected’
Traceback (most recent call last):
File “C:\Users\tvanneau\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py”, line 304, in _get_components
_, components = connected_components(adjacency)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
ValueError: Buffer dtype mismatch, expected ‘ITYPE_t’ but got ‘long long’
Traceback (most recent call last):
Cell In[124], line 1
T_obs, clusters, cluster_p_values, H0 = permutation_cluster_test(
File :12 in permutation_cluster_test
File ~\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py:1242 in permutation_cluster_test
return _permutation_cluster_test(
File ~\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py:987 in _permutation_cluster_test
out = _find_clusters(
File ~\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py:455 in _find_clusters
out = _find_clusters_1dir_parts(
File ~\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py:493 in _find_clusters_1dir_parts
clusters, sums = _find_clusters_1dir(
File ~\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py:547 in _find_clusters_1dir
clusters = _get_components(x_in, adjacency)
File ~\AppData\Local\anaconda3\Lib\site-packages\mne\stats\cluster_level.py:311 in _get_components
comp_list[comp].append(ii)
IndexError: list index out of range
This is weird because everything is working well when I choose ‘adjacency=None’, but then I am loosing the 3D information and the spatial clustering is made only with neighbor rows, so in 2D unless I misunderstood something.
Thus, did you ever faced this issue and how are you performing spatial permutation cluster test if this is working for you?
Thank you in advance for your help and have a great day!