Respiration pattern analysis

What Does This Channel Measure?

The calibrated respiratory pattern can be recorded using inductive plethysmography (e.g., Respitrace, LifeShirt) or pneumatic bellows (e.g., Lafayette). Most other techniques (e.g., strain gauge with piezoelectric pressure-voltage conversion using the Hall effect) are not linear and thus cannot be calibrated accurately. This channel measures the subject’s breathing activity, including respiration rate and lung volume. anslab initially displays the raw data in five windows. For editing purposes, you are interested in figure 3, thoracic breathing, and figure 4, abdominal breathing.

You will notice that there is a basic cyclical pattern. Note: You must have done the respiration calibration analysis. anslab will base lung volume related measures on an accurate calibration.

   
What Kind of Artifacts are Common in this Channel?

There are two common problems with respiration files.  The first is incorrect calibration.  Problems with calibration will not show up immediately, so be careful with this first step. The second type of common problem is irregularities due to technical error or movement artifacts.  Both irregularities deviate from the cyclical pattern and may include extreme spikes in lung volume.
 
How Are Artifacts Removed?

Note: Do not ‘over-edit’ respiration.  In general, if respiration is properly calibrated and observes a cyclical pattern, editing is not required for many experiments.  Baseline shifts will be filtered out.







If you see any huge irregularities in the data, that is, periods where you do not see a cyclic, regular breathing pattern, you want to edit these data. Most of the time, irregularities will be resets, manifested by sharp, vertical lines. If there are very extreme spikes in the signal due to technical difficulties, exclude those by selecting “edit thorax” or “edit abdomen”. In the window that opens you can exclude data by "exclusion boxes" (drag the mouse to define a rectangle spanned by an upper left to a lower left corner; when you release the mouse, all values between the left and right margins that are above the upper or below the lower margin will be excluded; interpolate by clicking on the right side of the window). After editing one window, quit it by clicking in the right bottom corner of that window. Then edit the other channel or, if you’re done, select "both channels o.k."


If there are no extreme spikes due to technical difficulties, but irregularities in the signal probably due to movement, you want to mark them as artifacts later on. Next, 10 computed parameters are displayed: RR: respiratory rate, V: minute ventilation, duty: inspiration time/total time, Pe: expiratory pause, Ti: inspiratory time, Vt: tidal volume, Flow: inspiratory flow rate, Te: expiratory time, Pi: inspiratory pause and IbrI: interbreath interval. The value displayed at the y-axis is the average.







Eyeballing those can give you hints as to where to expect artifacts. For example, the green line showing inspiration time should be pretty constant, and spikes here are suspicious. To edit artifacts, you can either change inidiviual cycle onsets, offsets and maxima/minima manually in the 'Raw'-data-mode, or do a rather rough outlier exclusion in the 'Event'-data-mode (see 'the NAVIGATION section' for details about the Raw/Event-Modes). When zooming in sufficiently in 'Raw'-data-mode, respiration onset/offset/maximima/minima-markers are displayed as shown in the graph below:








The activity signal (displayed in green in the bottom window) and the thorax-abdomen desynchrony signal (in blue) can additionally help you recognize artifacts: If there’s a lot of activity or a lot of desynchrony, there might be artifacts in the respiration. You can take out entire breathing cycles using the 'delete cycle'-button, insert an entire cycle using the 'insert cycle'-button, or move onset/offset/minima/maxima-markers using the 'insert insp. end'-, the 'insert exp. end'-, the 'insert insp. onset'- and the 'insert exp. onset'-buttons.  Individual onset/offset points of cycles are edited by  clicking close to the point that you want to move, and a second time to the location where you want to move the point. This kind of editing is important if you are very interested in certain specialized parameters like expiratory pause time.


You can mark artifacts by using the 'define artifact'-button. After pushing this button, click twice on the data window to mark beginning and end of the artifact. A horizontal red bar is then displayed to symbolize the artifact definition. When variables are saved, artifact intervals will be filled with 'NaN's.




Once you have finished editing, push 'Resume' and choose 'Save reduced data'  if you want to save your changes.



   
What Qualities Must Be Preserved In Editing?
     
If you use respiration data for transfer function spectral analysis of respiratory sinus arrhythmia it must be calibrated carefully. The most important quality to preserve is the cyclic breathing pattern so that respiratory rate can be estimated correctly.  Many irregularities in wave height and frequency can be ignored if they conform to a general cyclical pattern and if the averaging over a minute or so will overcome much of the misestimation error.
Be conservative when editing baseline shifts and movement artifacts.  Areas you exclude are interpolated as straight lines.  Therefore, exclude only clear artifactual spikes, but do not exclude small baseline shifts and slower ‘unusual’ phenomena.