Similarity Measurements (Medical images) using CNN or any other techniquess

Hi, I have been working on a project which is based on unsupervised CNN. Two medical image datasets (CT and DynaCT images) needs to be segmented (Segmentation is already done) and then I've to find the similarity between these newly segmented datasets. It could be using any CNN models or any other techniques but it should be reasonable. I mean the output we are looking for that needs to be reasonable. What I meant by reasonable here? Well, I've worked on the project but unfortunately enough, the output that I got is not quite what I was looking for. I used SSIM technique where I put segmented CT images on X-axis and segmented DynaCT images on the y-axis. Since the two imaging datasets are of same person, so the graph should be a straight line or relatable. I meant to say is, for one point (let's say 20) in X-axis (ct), the DynaCT (y-axis), should be 20 or around 20 and not so far away from 20, right? They needs to follow a pattern what is missing in the output that I got. There's no pattern and hence the output is not reasonable. There are way too many outliers which needs to be reduced. Someone who has expertise in this, please come forward and help me get this done.


Remuneration: Any reasonable price can be negotiated after the job is done. Unfortunately, no advance or any sort of pre-payment is possible.

Skills needed: Expertise in Deep learning, Image processing and knowledge about SSIM would do just fine to finish this task.

Please contact me [Removed by Freelancer.com Admin]

Thank you.