Removing artifact and outlier values

[Mayflower2400]

I am a behavioral researcher who is just learning Labview. I just purchased the full Professional Development program after using signal express and a DAQ board for the past 2 years. I now have hundreds of files collected that I would like to automate processing for as well as create a program to process new data collections. I was able to create some of the steps but am really having trouble getting out bad data first. I have tried numerous examples found on the forums and in the manuals - but my inexperience is showing. I am hoping someone can help me past this step (and if you have suggestions for future steps all the better!).

My signal is from an actocardiograph machine that records two signals on the same time scale: fetal heart rate in beats per minute (generally 100-190 bpms) not an electrical signal so no QRS wave, just a doppler measurement) and a second signal of activity in arbitrary units (0-100). Movement artifact is present throughout the recording in the heart rate signal. I need to delete artifact and interpolate if possible.  I cant seem to get any of the VIs to process the signal for this - and I think its because I dont know the best one to use or how to configure it properly. There are also instances of artifact that I need to remove completely due to interference from my ultrasound machine that is  within this range but clearly a repeating pattern of a steady 146-148 bpms.

Any suggestions or examples of exactly HOW to use VIs such as removing outliers (e.g. how do I specify the upper and lower range for the Y values of 100 to 190) or wavelets would be GREATLY appreciated!

[A_Patel]

Have you seen the following forum?

http://forums.ni.com/ni/board/message?board.id=170&requireLogin=False&thread.id=186974

Do any of the VIs do what you are asking for? 

Since you are reading from a file, you may have to add thea read values in section to the code.

Cheers,

Flash P

Applications Engineering

National Instruments

[CoastalMaineBird]

I'm pretty experienced at processing all sorts of scientific data, but I'm having trouble understanding what exactly you need.

Can you post a picture of a graph of a portion of this data?

And then point out the wheat and the chaff? 

[Mayflower2400]

Thank you! Yes I had seen this posting and also thought it should work but when I tried to add in the vi "read from measurement file" my wiring to the existing input did not seem to work as nothing changed after it read in the file - not even the original array. I am attaching a picture of what the signals look like. I still need to delete bad sections of data before averaging.  If you think after seeing the signals this is the example you suggested is the way to go, then I will work on learning how to modify this properly.

 

[Mayflower2400]

Hopefully the previously posted picture helps to explain better.

[CoastalMaineBird]

I don't get it.  the strip chart from the machine looks reasonable - the signal is not varying all that much.

Yet your DAQ signal looks as if it suffers from dropouts. As if the wire was loose, or a connection was poor.

My first recommendation would be to fix that, and get your acquisition up to snuff. 

Assuming you must process the files you have, then you need to clip it, it seems to me. Find the average, assume a +/- 10% (or something) variation, and set clip levels to those values.  Anything clipped should be set to the latest good value you had.

Still, that's trying to make a silk purse from a sow's ear. 

[Coq_Rouge]

Yes I can se you have a problem. The human hart rate do not jump around like that. I have some ideas. They are quite simple. How often is the heart rate updated. Can you post some data. Then you have some data you will share copy the chart to a new VI. Then right click on the chart and selct  data operations>>Make current vaules default. Then save, zip and post

[Mayflower2400]

The printed data doesnt print out the drop outs but rather show as blank spaces - but they are always there. Both signals are obtained from the same doppler so it is very prone to movement artifact - though not usually as plentiful as these files are. Here is the sample chart you requested. The data are sampled at 10 per second - a rather low frequency due to my former hand processing and needing to handle the lowest possible.

[Coq_Rouge]

Ok first try. It is better now

[Mayflower2400]

Yes! That looks great! - now how do I do that?