Finding Peak Edges from Array of 116 Curves

Hello! I want to first say I'm incredibly appreciative of this community and its response to questions. People like me very much appreciate it. Here is my situation:

I have an array of 116 curves:

Each cuve is a single peak:

What I'm trying to do is find the edges of the peaks and the position of those edges:

I have successfully used findpeaks() to find the position, location, and peak width of each of the 116 curves/peaks, but the peak width is limited to either the bandwidth at halfheight or halfprominance. I was wondering if anyone could help me with an approach that could help me find the peak edges and what index/pixel it occurs at. I would greatly appreciate it.

Answer 1

[Mathieu NOE]

hello

I suggest to use the attached crossing function , with the y threshold corresponding the y value of the black line in your picture.

the function allows you to get the positive slope (left crossing) point and the negative slope (right crossing) point

hope it helps

[Christopher Gordon]

Hello!

I wanted to thank you for your timely feedback. I do have a question on the implimentation of this function to my data. When I perform this function with the demo example, the x and y inputs are both single vectors:

In my example, I have a single x vector, but a matrix of 116 (or even 1024) vectors/curves. Do I need to make a for/end loop in order for this function to work through each individual curve?

[Mathieu NOE]

hello

we could probably make the function work on a matrix and not only on vectors, but i guess it would require me to do also a for loop inside the function - so it's the same as to do it in the main code on each individual curve.

there is only one special case where you could use it as is even with multiple curves, but my idea would be to create the upper envelopp curve of all your data and apply the function on that envelopp

but that will work only if the threshold is above the y value of the crossing points between adjacent curves.

[Christopher Gordon]

Hello,

I did select the threshold so that each peak is above the threshold. Also so that any other features in the baseline are also below the threshold:

[Mathieu NOE]

do these curves share a common x axis or is each curve defined by specific x and y vectors ?

[Christopher Gordon]

Each curve uses the same x axis (1:1024)

[Mathieu NOE]

ok

attached the files for your data format

no need to do a for loop in the main code - this is done inside the function

enjoy !

[Christopher Gordon]

I worked beautifully. With careful choice of interpolation in generating the curves matrix and threshold setting, I was able to get the both edges of each curve.

My next goal that will take some trial/error is setting the x range between those edges = to 0 for each curve. I think that will most likely take a loop, right?

[Mathieu NOE]

Gald it helped

I am not sure to understand the 2nd part of your answer ,

for me once the threshold is choosen, the x values of the crossing points are defined , so I don't get what do you mean by "setting the x range between those edges"

tx

[Christopher Gordon]

Here is what I mean shown on 1 individual curve. Once the left and right edges are located, I need to set all signal values within that range to a value of 0.

[Mathieu NOE]

ok , it's clearer now

do you want to have this feature in the subfunction directly ? we can pass an argument with the target value (zero here)

[Christopher Gordon]

How would you set that kind of argument up as a for argument? I would like to have it as its own separate step in the function in case I want to move it around in the program.

[Christopher Gordon]

Again, thank you so much for all the help and sorry I keep asking so much. I went through the MatLAB crash course, but they only ever do simple examples and I have trouble expanding them, perhaps using other techniques that I haven't learned yet.

Up until now I have three matricies for this step:

t0_pos (116 x 1) that has the left side of each peak

t0_neg (116 x 1) that has the right side of each peak

AllCurves (1024 x 116) that has each curve

I would like the loop (or if there's a better way I'm open) to go through each curve and set that sections range of t0_pos to t0_neg equal to 0.

[Mathieu NOE]

hello Cristopher

no problem, I like to work on (interesting) posts !

One problem I figured out this night is that the t0_pos and t0_neg values are obtained by linear interpolation for maximal precision; this is good , but on the other side , these "new" time values do not belong to the original time vector , and they are different for eah individual curve;

so at the end, I cannot create a common time axis for all curves , unless I do not merge the crossing points coordinates into the original x and y datas

IMO, we should step back to what was the first implementation and add some steps :

• do a for loop in the main code
• do the crossing stuff on one y vector at a time
• merge the x and y original datas with the crossing points coordinates (and sort x in ascending order to avoid zig zags in plot)
• put the y section between x = t0_pos to t0_neg equal to 0
• repeat for all curves
• this implies each curve a a specific x vector
• plot with hold on