|Next: GQRS(1) Up: WFDB Applications Guide Previous: FIR(1)||On This Page|
gqfuse produces a QRS annotation file based on two or more input QRS annotation files with annotator names annotator1, annotator2, etc. Each one-minute segment of the output annotation file is a copy of the corresponding segment of one of the input annotation files. In each segment, the program copies the input that best matches a predicted heart rate. If there are N inputs, the prediction is the median of N+1 values (the previous prediction and the number of beats marked within the current segment of each of the N input files). Although this process allows the input to be switched once per minute, the policy for resolving ties (within 2 beats) favors not switching if the previously chosen input is one of those belonging to the tie.
As its name suggests, gqfuse is intended to be used as a companion to the gqrs(1) QRS detector, but it is able to process annotations from any beat detector. Non-beat annotations (e.g., rhythm, signal quality, artifact, non-QRS waveforms, and notes) are copied to the output if present in the best matching input segments, but they are not counted as beats by gqfuse when it makes heart rate predictions.
One way to use gqfuse is to combine input annotation files for each available ECG signal in a record, made using a single detector such as gqrs. Another is to combine input annotation files made using a variety of QRS detectors. These ideas can be combined as desired.
A configuration file, which can be shared with gqrs and gqpost(1) , can be used to specify the expected heart rate. (In future versions, other parameters in the configuration file may also be used by gqfuse). The configuration file is unnecessary when processing adult human ECGs, but an appropriately constructed configuration file allows gqrs to analyze fetal, pediatric, and animal ECGs.
It may be necessary to set and export the shell variable WFDB (see setwfdb(1) ).
Table of Contents
Up: WFDB Applications Guide
Please e-mail your comments and suggestions to email@example.com, or post them to:PhysioNet
Updated 28 November 2018