使用neuromaps计算体数据空间空模型时遇内存不足问题求助

我需要为4D fMRI脑扫描的NIfTI格式体数据计算空间空模型，目前使用neuromaps工具包的burt2018()函数。但反复因存储空间不足出现内存错误，即使在高性能计算机上运行也无法解决。

错误信息 #

Traceback (most recent call last):
  File "/mnt/ghrelin/tvns_neuromaps/spatial_nulls.py", line 51, in <module>
    nulls = nulls.burt2018(combined_mask_img, atlas='MNI152', density='3mm', n_perm=3, seed=1234)
  File "/home/neuromadlab/.local/lib/python3.8/site-packages/neuromaps/nulls/nulls.py", line 600, in burt2018
    return _make_surrogates(data, 'burt2018', atlas=atlas, density=density,
  File "/home/neuromadlab/.local/lib/python3.8/site-packages/neuromaps/nulls/nulls.py", line 537, in _make_surrogates
    hdist, np.argsort(hind, axis=-1), axis=-1)
  File "<__array_function__ internals>", line 200, in argsort
  File "/home/neuromadlab/.local/lib/python3.8/site-packages/numpy/core/fromnumeric.py", line 1146, in argsort
    return _wrapfunc(a, 'argsort', axis=axis, kind=kind, order=order)
  File "/home/neuromadlab/.local/lib/python3.8/site-packages/numpy/core/fromnumeric.py", line 57, in _wrapfunc
    return bound(*args, **kwds)
numpy.core._exceptions.MemoryError: Unable to allocate 243. GiB for an array with shape (180671, 180671) and data type int64

已尝试的解决方法 #

• 将数据从int64类型转换为int32，但内存错误仍未解决

求助内容 #

1. 该计算所需的存储空间预估
2. 降低计算强度的可行方法

内容的提问来源于stack exchange，提问作者Lena Wiegert