Datasets

This page provides details of publicly available datasets that may be useful for research into biomedical signal processing techniques.

Photoplethysmogram (PPG) Datasets

Dataset Signals Overview of protocol
CapnoBase PPG, ECG, resp Signals from patients during elective surgery and routine anaesthesia. (ref: 1 ).
MESA Polysomnography PPG, ECG, EEG, resp, others Recordings from overnight polysomnography (sleep) studies. (ref: 1 2 ).
MIMIC PPG, ECG, resp, BP, others Recordings during critical care admissions
PPG-DaLiA PPG, ECG, resp, accel, EDA Recordings during activities of daily living
PPG Diary 1 Dual wavelength PPG PPG signal measured at the thumb from a single participant for 28 days. The participant recorded their activities of daily living, and any technical problems (ref 1).
Pulse Wave Database PPG, BP, blood flow, blood vel Simulated arterial pulse waves at a range of common measurement sites for 4,374 virtual subjects aged 25-75 years, representative of healthy adults (ref 1)
Vortal PPG, ECG, resp signals acquired from young and elderly healthy subjects at rest (ref: 1 2 ).
WESAD PPG, ECG, resp, accel, others Recordings at baseline and during amusement, stress and meditation

Electrocardiography (ECG) Datasets

Dataset Signals Overview of protocol
Physionet CinC 2017 Challenge Dataset ECG Handheld, single-lead recordings alongside heart rhythm labels.

The Physionet CinC 2017 Challenge Dataset contains data recorded from handheld ECG devices, alongside reference labels of the rhythm for each recording. This Matlab Script collates the data into a single Matlab file for analysis. The file contains 30-second ECG recordings, alongside reference labels of the heart rhythm.

Fetal ECG Datasets

Table 1 in this article provides a list of fetal ECG datasets.

Other Datasets

Dataset Signals Overview of protocol
Virtual Database of Pulse Waves BP, blood flow, blood vel Pulse waves simulated at several arterial sites under a range of cardiovascular conditions (refs: 1 and 2). Simulated using the Nektar1D model of pulse wave propagation (ref: 3 ).

Finding datasets

I find the following websites helpful for finding datasets:

  • PhysioNet: An invaluable collection of physiological datasets, many of which contain physiological signals such as the ECG.

  • Google Dataset Search: A helpful search engine.


Table of contents


Copyright © 2021 Peter Charlton.