compare images 2 signal and indentify difference

I need to compare outputs of ultrasonic scan of one with no damage to one with damage. The results are in image form and I used a graph digitizer to extract the graph and has saved it as a csv file.

I have absolutely no basis in matlab or any other cording enviornments but i was able to generate a plot of the signals.

na=readmatrix('Default Dataset.csv');
x= na(:,1);
y=na(:,2);
na2=readmatrix('Default Dataset (1).csv');
x1= na2(:,1);
y1=na2(:,2);
plot (x,y);
hold on
plot (x1,y1)

I tried comparing the same as imgaes but it doesnt give any desired results

When trying to create a signal, it says the data intervals are irregular and I'm stuck at this point. Is there another way to proceed forward and compare the peaks of the graphs ?

Answer 1

[Image Analyst]

Like with findpeaks?

Mathworks had a seminar/demo of this using Deep Learning to look at spectrograms of ECG signals. Maybe you could apply that but you'd need dozens or hundreds of signals that were labeled as normal or defective.

https://www.mathworks.com/help/signal/ug/classify-ecg-signals-using-long-short-term-memory-networks.html?s_tid=srchtitle_ecg%20spectrogram_1

Lots more hits here:

https://www.mathworks.com/support/search.html?q=ultrasound%20network&page=1

[Shruthimol]

I did try the ecg one but I do not have sufficient data to use deep learing. To use the ecg method, there needs to be a signal and the first hurdle too to generate my csv data as a signal. A plot is obtained as per my codes but are not accepted as a signal (saved as .mat file). when trying to create a signal from the time and amplitude data in the csv file, it says the time intervals are not equal and I dont know how to interpolate the data in the csv file first.

[Image Analyst]

Sounds like you have some data collection to do and some work to do.

[Walter Roberson]

Assuming that you have a 2D array Signal that has one column for each distinct signal (so time proceeds with increasing rows), and assuming you know the sample rate in Hz:

dt = 1/Sample_Rate;   %or assign dt directly if you know it
nsamp = size(Signal,1);
timevec = (0:nsamp-1).' * dt;
time_and_Signal = [timevec, Signal];

writematrix(time_and_Signal, 'NameOfCSVFileGoesHere.csv');