| import os |
| import mne |
| import numpy as np |
| import lmdb |
| import pickle |
|
|
| |
| def iter_files(rootDir): |
| |
| files_H, files_MDD = [], [] |
| for file in os.listdir(rootDir): |
| if 'TASK' not in file: |
| if 'MDD' in file: |
| files_MDD.append(file) |
| else: |
| files_H.append(file) |
| return files_H, files_MDD |
|
|
|
|
| selected_channels = ['EEG Fp1-LE', 'EEG Fp2-LE', 'EEG F3-LE', 'EEG F4-LE', 'EEG C3-LE', 'EEG C4-LE', 'EEG P3-LE', |
| 'EEG P4-LE', 'EEG O1-LE', 'EEG O2-LE', 'EEG F7-LE', 'EEG F8-LE', 'EEG T3-LE', 'EEG T4-LE', |
| 'EEG T5-LE', 'EEG T6-LE', 'EEG Fz-LE', 'EEG Cz-LE', 'EEG Pz-LE'] |
| rootDir = '/data/datasets/MDDPHCED/files' |
| files_H, files_MDD = iter_files(rootDir) |
| files_H = sorted(files_H) |
| files_MDD = sorted(files_MDD) |
| print(files_H) |
| print(files_MDD) |
| print(len(files_H), len(files_MDD)) |
|
|
|
|
| files_dict = { |
| 'train':[], |
| 'val':[], |
| 'test':[], |
| } |
|
|
| dataset = { |
| 'train': list(), |
| 'val': list(), |
| 'test': list(), |
| } |
|
|
| files_dict['train'].extend(files_H[:40]) |
| files_dict['train'].extend(files_MDD[:42]) |
| files_dict['val'].extend(files_H[40:48]) |
| files_dict['val'].extend(files_MDD[42:52]) |
| files_dict['test'].extend(files_H[48:]) |
| files_dict['test'].extend(files_MDD[52:]) |
|
|
| print(files_dict['train']) |
| print(files_dict['val']) |
| print(files_dict['test']) |
|
|
|
|
| db = lmdb.open('/data/datasets/MDDPHCED/processed_lmdb_75hz', map_size=1273741824) |
|
|
| for files_key in files_dict.keys(): |
| for file in files_dict[files_key]: |
| raw = mne.io.read_raw_edf(os.path.join(rootDir, file), preload=True) |
| print(raw.info['ch_names']) |
| raw.pick_channels(selected_channels, ordered=True) |
| print(raw.info['ch_names']) |
| raw.resample(200) |
| raw.filter(l_freq=0.3, h_freq=75) |
| raw.notch_filter((50)) |
| |
| eeg_array = raw.to_data_frame().values |
| |
| eeg_array = eeg_array[:, 1:] |
| points, chs = eeg_array.shape |
| print(eeg_array.shape) |
| a = points % (5 * 200) |
| print(a) |
| if a != 0: |
| eeg_array = eeg_array[:-a, :] |
| eeg_array = eeg_array.reshape(-1, 5, 200, chs) |
| eeg_array = eeg_array.transpose(0, 3, 1, 2) |
| print(eeg_array.shape) |
| label = 1 if 'MDD' in file else 0 |
| for i, sample in enumerate(eeg_array): |
| sample_key = f'{file[:-4]}_{i}' |
| data_dict = { |
| 'sample': sample, 'label': label |
| } |
| txn = db.begin(write=True) |
| txn.put(key=sample_key.encode(), value=pickle.dumps(data_dict)) |
| txn.commit() |
| dataset[files_key].append(sample_key) |
|
|
|
|
| txn = db.begin(write=True) |
| txn.put(key='__keys__'.encode(), value=pickle.dumps(dataset)) |
| txn.commit() |
| db.close() |
