# Suggestion for finding the stable regions in spiral galaxy data?

I am working with a data set that consists of the actual pitch angle (given as PA(Y)) and the pitch angle at each radii (listed from 1 to 217). In the image below, you can only see radii 1 through 16. The Mode(Y) in the image below is not of relevance at the moment.

There are regions that range between certain radii in which the pitch angle measurement does not change (in the image, you'll notice this happens for all the radii values, but they do change after a certain radius that's cut off in the image). These are known as stable regions. My goal is to capture all the ranges in the data in which the pitch angle measurement does not change, and create a program that returns those values.

Is there a machine learning method in which this is possible, or is this just a non-machine learning problem? I have tried creating plots and have considered creating a CNN that can identify these flat regions, but I feel like this is overkill. My PIs want to use a machine learning method and they have proposed neural networks, hence why I tried the CNN, but I just am not sure if that is possible?

I should add, usually stable regions radii ranges are unknown, so the goal is to try to see if certain radii ranges usually can predict where a stable region are located.

Moreover, I've thought of using a classifier to determine whether a region is flat or not. I am just very confused as to how to approach this. Are there any similar examples to the problem I'm currently working on that someone can point me to?