peak picking help

[douglas]

Hi All,

I'm wondering if the following is possible.

I have a set of data, lots of numbers against a time and the resulting x y scatter plot resembles a comb, with lots of peaks or teeth separated by lesser values. Each peak is made of about 100 points and there are say 120 peaks on the comb.

What I want to do is reduce the data set to 120 values, each one is the maximum height of each comb. This would involve breaking the data into 120 chunks and searching each chunk for the highest number, then recording it.

Is anyone able to help on how to do this?

Thanks alot.

[shg]

Welcome to the forum.

How noisy is the data? Post it.

[douglas]

Thanks shg. Not noisy at all. Here's a rough simulation. I'm looking for some advice on how to approach the problem.

[Andy Pope]

You could use a formula in column C with which to filter the peak values
Filter on Not Equal #N/A

C2: =IF(AND(B2>B3,B2>B1),B2,IF(AND(B2<B3,B2<B1),B2,NA()))

[douglas]

You could use a formula in column C with which to filter the peak values
Filter on Not Equal #N/A

C2: =IF(AND(B2>B3,B2>B1),B2,IF(AND(B2<B3,B2<B1),B2,NA()))

Thanks Andy that's brilliant. I've removed the valley finding part at the end because I don't need that. Is it possible now to extract only the real values so that I have in eg. D2:D7 the peak heights?

Thanks

[Andy Pope]

In D2:D7 you can use this array formula, use CTRL+SHIFT+ENTER to commit formula.

=LARGE(IF(ISNA(C2:C1000),0,C2:C1000),ROW()-1)

In E2:E7 use

=INDEX($A$2:$A$1000,MATCH(D2,$C$2:$C$1000,0))

[theoneness]

After reading this thread, and in advance, thank you for everyone's assistance...
I noticed Shg asked how noisy the data was: So, are there any excel tools that could facilitate peak picking in far noisier data? (Excel 2008 - OS X: Snow Leopard)
I ran an experiment to collect the oscilloscope data of a heart contracting as it was retrieved by an isotonic transducer. A baseline run of the data is attached. You'll see that It has peaks and troughs like any other wave, but since it's physiological data, it's not very clean.
There's also peaks roughly halfway between the larger peaks which correspond to the atrial systole (the highest peaks are the ventricle systole). In other words, there are peaks for both the Lub and the Dub in the Lub-Dub of the heartbeat (fyi, this is Rana catesbeiana ~ animal physiology lab). I need to locate both.
Cheers

edit: the two red highlighted rows represent an atrial peak and a ventricle peak, demonstrating just how undistinguished the atrial peak indeed is.

[teylyn]

Hi theoneness,

could you please take a moment to read the forum rules and then start your own thread? Feel welcome to link to this one if it helps provide background to your question.

cheers

[herbds7]

Excel 2010 Tables
Find peaks in noisy periodic signals.
http://c3017412.r12.cf0.rackcdn.com/04_08_11a.xlsx
If you get *.zip, don't unzip, just rename *.xlsx

[teylyn]

herbds7, please use the forum tools to upload files, rather than untrusted external sites.