An Introduction to the PhysioBank Archives

From this site, you may download significant portions of many databases of annotated, digitized physiologic signals. Most of the databases currently in the PhysioBank Archives were developed at MIT and at Boston's Beth Israel Hospital (now the Beth Israel Deaconess Medical Center) and have previously been distributed in CD-ROM format. Many of these databases are available in their entirety from these archives. The support provided to PhysioBank by the NIH NCRR makes it possible for us to provide free access to these databases via PhysioNet to the research community.

The databases can still be obtained in CD-ROM format; since they are quite large, you may wish to consider purchasing one or more CD-ROMs. Please visit the MIT-BIH Database Distribution web site for further information.

We recommend that you begin by downloading and studying one or more of the excerpts described below before downloading any of the full-length recordings available from the PhysioBank Archives. Most of the excerpts are 10 minutes in length (exceptions are noted below). The .dat files contain the digitized signals, and are the largest (up to about 1.5 Mb each, though most are shorter). The full-length recordings also vary considerably in length; thirty minute recordings are typical, but some are in the range of 20 to 50 hours in length, with individual files up to 60 Mb, though most of longer recordings are divided into much shorter segments for convenience.

Most users will also need to download the WFDB Software Package (also available from this site) in order to view these files (there are alternatives, however).

The following excerpts are available for downloading. Be sure to download all of the files associated with each record (there are 2 to 4 such files in each case; see the individual listings below for details). If you are using Netscape, one easy way to download these files is to hold down the shift key while you click on the links, then choose the directory on your system where you wish to save the files.

The following databases are included on the MIT-BIH Arrhythmia Database CD-ROM; see mitdb.txt for further information. Note that most of the contents of the CD-ROM may be downloaded from the PhysioBank archives.

• MIT-BIH Arrhythmia Database (records x_100, x_119, and x_203)
  Download files x_100.atr (beat and rhythm annotations), x_100.dat, x_100.hea, x_119.atr, x_119.dat, x_119.hea, x_203.atr, x_203.dat, and x_203.hea. Record x_100 is relatively clean and uncomplicated; record x_119 contains many ventricular ectopic beats but is also relatively clean; and record x_203 is exceedingly complex (high-grade VEA together with AF and noise). Record 100s, a shorter (1 minute) excerpt of the same record as x_100, is also available. Download files 100s.atr, 100s.dat, and 100s.hea if you need record 100s.
• MIT-BIH Noise Stress Test Database (record x_119e12)
  Download files x_119e12.atr (beat and rhythm annotations), x_119e12.dat, and x_119e12.hea. Compare the final 5 minutes of this record with those of x_119 (the same data without added noise; the first 5 minutes are clean). The amount of added noise is moderately high (SNR = 12dB; other records in this database have SNRs ranging from 24dB to -6dB).
• Creighton University Ventricular Tachyarrhythmia Database (record cu01)
  Download files cu01.atr (beat and rhythm change annotations), cu01.dat, and cu01.hea. This is a complete 8.5 minute record from this database. Unlike the other databases, records in this database contain only 1 ECG lead.
• MIT-BIH ST Change Database (record x_300)
  Download files x_300.atr (beat annotations), x_300.dat, and x_300.hea. Note that ST changes are present but unannotated in this database, unlike the European ST-T Database. Record x_300 is a portion of an exercise stress test, and was recorded using an FM instrumentation tape recorder rather than an ambulatory ECG recorder (as was used for most of the other databases).
• MIT-BIH Malignant Ventricular Arrhythmia Database (record x_418)
  Download files x_418.atr (rhythm change annotations), x_418.dat, and x_418.hea. The annotations provided with this database here indicate only rhythm changes. Note that "(N" rhythm labels indicate all rhythms other than ventricular tachycardia ("(VT"), ventricular flutter ("(VFL"), and ventricular fibrillation ("(VFIB").
• MIT-BIH Atrial Fibrillation/Flutter Database (record x_04936)
  Download files x_04936.atr (rhythm change annotations), x_04936.dat, x_04936.hea, and x_04936.qrs (beat annotations). This is a 30-minute excerpt of a 10-hour record, one of 25 in this database. Note that this is the only record in this database for which the ECG signals are currently available; the other 24 records contain beat and rhythm change annotations only. Annotator qrs (file x_04936.qrs) contains beat annotations, and annotator atr (file x_04936.atr) contains rhythm change annotations. This database is intended primarily for development and evaluation of atrial fibrillation detectors that rely on R-R interval information only, and for study of R-R interval dynamics in the context of AF. All beats are labelled as normal in the qrs annotation files, although ectopic beats are present.
• MIT-BIH ECG Compression Test Database (record 08730_01)
  Download files 08730_01.dat and 08730_01.hea. This short (20.48 seconds) record is included in its entirety here. This database is unannotated.
• MIT-BIH Supraventricular Arrhythmia Database (record x_800)
  Download files x_800.atr (beat and rhythm change annotations), x_800.dat, and x_800.hea.
• MIT-BIH Long-Term Database (record x_15814)
  Download files x_15814.atr (beat annotations), x_15814.dat, and x_15814.hea. This 10-minute excerpt of a 22-hour record contains 3 ECG signals.
• MIT-BIH Normal Sinus Rhythm Database (record x_16265)
  Download files x_16265.atr (beat annotations), x_16265.dat, and x_16265.hea. This is a one-hour excerpt of a 21-hour record. Note that, because of space constraints, the ECG signals in this excerpt are not included on the MIT-BIH Arrhythmia Database CD-ROM, which contains only the beat annotations for the 18 records in this database. The ECG signals may be downloaded from the nsrdb directory of the PhysioBank Archives; they are also available on a separate CD-ROM.

The following databases are not included on the MIT-BIH Arrhythmia Database CD-ROM, but are on other CD-ROMs available from us. See the notes in the individual entries for further information.

• MIT-BIH Polysomnographic Database (record x_slpdb)
  Download files x_slpdb.dat, x_slpdb.ecg (beat annotations), x_slpdb.hea, and x_slpdb.st (sleep stage annotations). This database is available from the slpdb directory of the PhysioBank Archives, or on its own CD-ROM (see slpdb.txt).
• MIMIC Database (excerpts of record 216 and 237)
  There are many files associated with each record in the MIMIC Database. The sample of record 237 available here includes the first three 10-minute segments of that record (which, in its entirety, is over 42 hours in length); record 237n, which contains heart rate, blood pressure, respiration rate, and O₂ saturation measurements for the entire 42 hours of record 237; and a set of hypertext files containing the clinical data (laboratory reports, abstracted data from the medical record) with links to the signals and measurements in records 237 and 237n. In all, the sample includes 102 files, which may be downloaded as a gzip-compressed tar archive (3180347 bytes), a PKZIP archive (3190250 bytes), or as individual files (approximately 11.3 Mb in all). For further information about this sample, see mimic.txt (included with the sample files). The complete 42-hour record 237 may be downloaded from the PhysioBank Archives; it is also available on a separate CD-ROM.

  The sample of record 216 includes two 10-minute segments of that record, without accompanying clinical data (10 files in all), which may be downloaded as a gzip-compressed tar archive (1865150 bytes) or as individual files (approximately 3 Mb in all).

Other databases are available on CD-ROMs in the same format as ours. Here are samples of two of these databases.

• European ST-T Database (record x_edb)
  Download files x_edb.atr (beat, rhythm, & ST change annotations), x_edb.dat, and x_edb.hea. Forty-eight of the ninety records, and all of the annotation files, in this database are available from PhysioNet. The remaining 42 signal files are available only on a CD-ROM from the developers of this database; please write to Alessandro Taddei <taddei@ifc.cnr.it> for further information.
• MGH/MF Waveform Database (records x_mgh020 and x_mgh022)
  Download files x_mgh020.ari (beat annotations), x_mgh020.dat, x_mgh020.hea, x_mgh022.ari, x_mgh022.dat, and x_mgh022.hea. These excerpts are each 1 minute long; the original records are typically about 2 hours each. This database must be obtained from its developers (see mghdb.txt).

What else do I need to look at these recordings?

The WFDB software that can be downloaded from this site can read any of these databases.

Download the file dbcal , which contains information about the customary scales used by software such as WAVE, WVIEW, and pschart for displaying or plotting signals of various types. Install it in the same directory as the sample data files.

Alternative methods for viewing these excerpts

If you wish to write your own software to read these files, you are encouraged to download the WFDB library (a portable set of functions usable from C, C++, or Fortran programs) and to use it with your own software. The advantage of doing so (apart from the immediate savings of effort) is that support for new file formats is added to the library from time to time, and your software can then incorporate this support simply by recompiling or relinking. As an example, support for reading remote files via HTTP or FTP has very recently been added to the WFDB library.

Jose Garcia Moros and Salvador Olmos have contributed a set of m-files for reading and writing PhysioBank signal and annotation files using Matlab (or the compatible and freely available Octave).

A workable alternative is to use rdsamp(1) and rdann(1) to convert any desired portions of these files into an easy-to-process text format. Sources for these programs are included in the WFDB Software Package; binaries are also available for several popular operating systems. You may also run these utilities on our web server without downloading the data files or software first; in this case, you can capture their output using your browser, or have it sent to you by e-mail. Visit the rdsamp-O-matic to obtain digitized samples as text, or the rdann-O-matic to obtain annotations as text.

The digitized signals contained in sample record x_100 are also available in uncompressed text format (5.2 Mb) and in gzip-compressed text format (1 Mb), if you wish to obtain a sample without the need to use any of our software to read it. The text file was prepared from the binary data using the command

    rdsamp -r x_100 -p -v >x_100.txt

Your comments and suggestions are welcome. We encourage you to use our feedback form to comment on this page. If you would like to receive a reply, please send your comments by email to webmaster@physionet.org, or post them to: PhysioNet MIT Room E25-505A 77 Massachusetts Avenue Cambridge, MA 02139 USA

Updated Tuesday, 05-Aug-2003 08:50:34 EDT