Extracted PSG Channels

To facilitate sleep research and algorithm development, we provide pre-extracted physiological signals from the ISRUC-Sleep dataset in MATLAB format.

These files eliminate the need for users to parse the original polysomnography recordings and allow researchers to immediately focus on signal processing, feature extraction, machine learning, and deep learning applications.

Important note: In channel-extraction, for each subject, we did not consider last 30 epochs of data due to noise. The length of these files is 30 epochs smaller than their corresponding hypnograms.

Available Signal Channels

The extracted files include commonly used physiological signals recorded during overnight polysomnography, including:

  • Electroencephalography (EEG)
  • Electrooculography (EOG)
  • Electromyography (EMG)
  • Electrocardiography (ECG)
  • Respiratory signals
  • Oxygen saturation (SpO₂)
  • Additional PSG channels when available

Sleep-stage annotations corresponding to each recording are also provided.

Download Extracted Channels

The extracted channels are organized according to the three ISRUC-Sleep cohorts.

Cohort I Extracted Channels-Download

This cohort contains recordings from subjects with sleep-related disorders and includes extracted physiological signals and corresponding sleep-stage annotations.

Cohort II Extracted Channels- Download

This cohort contains extracted signals from healthy adult subjects and is suitable for sleep staging and physiological signal analysis studies.

Cohort III Extracted Channels- Download

This cohort provides an additional independent set of healthy subjects for benchmarking and validation studies.

File Format

The extracted signals are distributed as MATLAB (.mat) files.

Each subject folder contains:

  • Physiological signal recordings
  • Channel information
  • Sleep-stage labels
  • Recording metadata

The MATLAB format allows direct loading into MATLAB, Python, and other scientific computing environments.

Example Usage

MATLAB : load(‘subject.mat’)

Python: from scipy.io import loadmat data = loadmat(‘subject.mat’)

Notes

The extracted channels are provided as a convenience resource. The original polysomnography recordings remain available through the Dataset Download page for researchers requiring access to the complete raw recordings.

Citation

If you use the pre-extracted MATLAB channels provided on this page in your research, publications, software, or derived works, please cite the following publication:

Automatic sleep staging: A computer-assisted approach for optimal combination of features and polysomnographic channels. Khalighi S., Sousa T., Pires G., Nunes U.

The extracted channel files were generated and organized as part of the methodology described in this work. Proper citation helps acknowledge the effort involved in preparing and distributing these resources and supports the continued maintenance of the ISRUC-Sleep database.