Nihon Kohden V2.0 support

Aperta
  • Qualcosa non va?
#14,502 2 commenti 0 reazioni 0 assegnatari Vedi su GitHub

I maintainer di solito rispondono entro 1 giorno

Nessuno ha ancora preso questa issue.

Valutazione

Difficoltà
4/5
Tempo stimato
3-5 giorni
Idoneità per principianti
48/100
Tipo di issue
Bug
Chiarezza
Abbastanza chiara
Stato di attività
Attiva
Stack tecnologico
python
Ambito
data

Direzione di ricerca

Start with mne-io/nihon/nihon.py and trace how read_raw_nihon chooses the legacy parser; the issue identifies the uint32 at offset 0x3EE as a format discriminator. Reproduce with an EEG-1200A high-density file and compare channel count and duration against the expected results. Done means the reader handles the extended block chain correctly or raises NotImplementedError instead of silently returning incorrect data.

Scritto dal modello di indicizzazione a partire dal testo della issue.

Descrizione

BUG
Description of the problem

Currently, mne-python's nihon kohden file reader read_raw_nihon silently mis-parses high-density EEG-1200A files. These have an extended block layout on disk.

#13448 enabled the EEG-1200A V01.00 device string in mne-io/nihon/nihon.py but no parser for the file layout exists. The data is stored in an extended block chain. The current low-density parsing path reads the legacy data block, whose channel-count byte usually for just channel.

The result: mne.io.read_raw_nihon() accepts these files and returns a one-channel Raw object with the wrong duration. No warning or error.

How the layout differs (verified against real files)
  • A uint32 at file offset 0x3EE is 0 on every V1 file inspected, and
    points to an extended-block chain on every EEG-1200A high-density file
    inspected. It is a reliable discriminator; the device string is not (the
    same string also appears on files that must keep the V1 path).
Proposed fixes, two options:
  1. safety - raise NotImplementedError when we detect the extend-block chain format.
  2. Add extended block support. I have a working fix which is about 100 lines long which has been tested on real files. These are compared with NK's own EDF export files and are bit-for-bit matches. (This also covers a silent scaling bug which I will submit separately.) I will also submit a document with my current understanding of the file format if that is desired.
additions to mne-testing-data

I have prepared a few small file examples of the extended block format which have no patient data in them. I would be able to submit them for including in the mne-python test data. These files also caught a couple of bugs in the version 1 format parsing path.

Steps to reproduce
create a `EEG-1200A V01.00` file on an NK acquisition machne. with this amplifier. call name the stem of the file name XXXX. So have {"XXXX.eeg", "XXXX.21e", "XXXX.log", "XXXX.pnt"


import mne
raw = mne.io.read_raw_nihon("XXXX.eeg", preload=False)
print(raw.info["nchan"], raw.n_times)   # 1 channel, wrong length; no error
Link to data

Here is a link to three small studies data files on google drive

Expected results

I will put the outputs of expected results here with the test files as the generic result using the example above is not meaningful.

Actual results

ditto

Additional information

This is being tested on ubuntu 22.04, 24.01 and mac OS 27

Lingua principale
Python
Stelle
3.6k
Fork
1.6k
Merge medio
1g 18h
PR unite (30g)
71

Preparare l'ambiente

Come iniziare

  1. Leggi tutta la issue e poi la guida ai contributi del progetto.
  2. Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
  3. Fai un fork del repository e lavora su un branch.
  4. Apri una pull request che faccia riferimento al numero della issue.

Altre issue di mne-tools/mne-python

Tutte le issue di mne-tools/mne-python

Issue simili

Altre issue su Python

Ricevi le nuove issue nella tua casella

Un breve riepilogo di issue GitHub adatte ai principianti.