Overview of the WFDB File Structure

A WFDB (Waveform Database) record typically consists of multiple related files that together define a physiologic signal recording and its annotations.

The key file types are:

File Type Typical Extension Description
Header file .hea Describes the overall recording and each signal’s properties.
Signal file .dat, .bin, .16, etc. Contains the actual waveform sample data.
Annotation file (optional) .atr, .ann, etc. Contains event markers, labels, or comments aligned to the signal timeline.

How the Files Work Together

  • The header file (.hea) is the starting point. It specifies the record’s name, sampling frequency, signal files, gain settings, and other metadata.
  • The signal files store the waveform data, often in a binary-encoded format. Each signal may have its own file, or multiple signals may share a file.
  • Annotation files are optional and provide time-stamped events, such as heartbeat labels, rhythm changes, or clinical comments.

These files share a common record name, and are usually kept together in the same directory.

For example, for a record named 100, the typical files might be:

100.hea    (header file)
100.dat    (signal data file)
100.atr    (annotation file)

Multi-Segment Records

Some WFDB records are multi-segment. In this case, the main header file references a sequence of segments, each with its own associated files.

Multi-segment records allow for efficient storage of long recordings and variable signal structures.

Summary

To fully read or analyze a WFDB recording, you need:

  • The header file (.hea),
  • The associated signal files (e.g., .dat),
  • And optionally, any annotation files (e.g., .atr).

Each file provides a piece of the complete information needed to reconstruct the signals and their clinical context.