mne.io.read_raw_fif to read .fif file produces in matlab

💬 Support & Discussions
1

As a part of our workflow, .fif files were produced by a matlab code. Now I need to upload them in Python script.
However, I am getting an error that my .fif files are not the objects with the right attributes.
What is the best way to transfer the .fif files into the right object type?

Code:

"`raw = mne.io.read_raw_fif(fif_input_path, preload = True, verbose = False)`"

Error: `/tmp/ipykernel_4013990/1776892309.py:1: RuntimeWarning: Invalid tag with only 0/16 bytes at position 0 in file /gpfs/gibbs/project/pushkarskaya/hp245/Spaceship/Cache/Subject19/S19A_raw-1.fif
  raw = mne.io.read_raw_fif(fif_input_path, preload = True, verbose = False)
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[6], line 1
----> 1 raw = mne.io.read_raw_fif(fif_input_path, preload = True, verbose = False)

File ~/.conda/envs/mne/lib/python3.12/site-packages/mne/io/fiff/raw.py:532, in read_raw_fif(fname, allow_maxshield, preload, on_split_missing, verbose)
    491 @fill_doc
    492 def read_raw_fif(
    493     fname, allow_maxshield=False, preload=False, on_split_missing="raise", verbose=None
    494 ) -> Raw:
    495     """Reader function for Raw FIF data.
    496 
    497     Parameters
   (...)
    530     are updated accordingly.
    531     """
--> 532     return Raw(
    533         fname=fname,
    534         allow_maxshield=allow_maxshield,
    535         preload=preload,
    536         verbose=verbose,
    537         on_split_missing=on_split_missing,
    538     )

File <decorator-gen-203>:10, in __init__(self, fname, allow_maxshield, preload, on_split_missing, verbose)

File ~/.conda/envs/mne/lib/python3.12/site-packages/mne/io/fiff/raw.py:100, in Raw.__init__(self, fname, allow_maxshield, preload, on_split_missing, verbose)
     98 next_fname = fname
     99 while next_fname is not None:
--> 100     raw, next_fname, buffer_size_sec = self._read_raw_file(
    101         next_fname, allow_maxshield, preload, do_check_ext
    102     )
    103     do_check_ext = False
    104     raws.append(raw)

File <decorator-gen-204>:12, in _read_raw_file(self, fname, allow_maxshield, preload, do_check_ext, verbose)

File ~/.conda/envs/mne/lib/python3.12/site-packages/mne/io/fiff/raw.py:192, in Raw._read_raw_file(self, fname, allow_maxshield, preload, do_check_ext, verbose)
    190     whole_file = True
    191 fname_rep = _get_fname_rep(fname)
--> 192 ff, tree, _ = fiff_open(fname, preload=whole_file)
    193 with ff as fid:
    194     #   Read the measurement info
    196     info, meas = read_meas_info(fid, tree, clean_bads=True)

File <decorator-gen-11>:12, in fiff_open(fname, preload, verbose)

File ~/.conda/envs/mne/lib/python3.12/site-packages/mne/_fiff/open.py:133, in fiff_open(fname, preload, verbose)
    131 fid = _fiff_get_fid(fname)
    132 try:
--> 133     return _fiff_open(fname, fid, preload)
    134 except Exception:
    135     fid.close()

File ~/.conda/envs/mne/lib/python3.12/site-packages/mne/_fiff/open.py:151, in _fiff_open(fname, fid, preload)
    149 #   Check that this looks like a fif file
    150 prefix = f"file {repr(fname)} does not"
--> 151 if tag.kind != FIFF.FIFF_FILE_ID:
    152     raise ValueError(f"{prefix} start with a file id tag")
    154 if tag.type != FIFF.FIFFT_ID_STRUCT:

AttributeError: 'NoneType' object has no attribute 'kind'`
2

What did you save in a FIFF file? FIFF is a format which can store different kind of data structures. Each data structure has its own reader function. mne.io.read_raw_fif is for Raw (continuous recording). You also have readers for source spaces, ICA, epochs, evoked, etc…

Mathieu

3

Hi Mathieu,
it is raw continuous recording that was aligned with a trigger. We did alignment in Matlab.
Thank you, Helen.

4

Can you load the recording before alignment with a trigger?
If yes, the MATLAB code and saving did not produce a valid FIFF file.
If no, the file itself has an issue.

You could export the relevant information from matlab in a .mat file, then load the file with e.g. pymatreader and create the MNE data structure you need, e.g. with mne.create_info and mne.io.RawArray.

Mathieu