I'm looking for software that simulates/introduces CT noise coming from low dose scanning or a dataset of noiseless and corresponding noisy images.
I plan on using the data for training a neural network to perform denoising. Could you please give my some directions?
You might have a look at challenge data from, e.g. the low dose CT grand challenge [1] as one of its purposes is to assess the performance of denoising algorithms on synthetized low dose CT data. As they state, the low dose datasets were generated from full dose scans via Poisson noise insertion. Training data consists of scans and projections from 10 patients and from a phantom.
Hope that helps.
[1] https://www.aapm.org/grandchallenge/lowdoseCT/
I think it was available for download by the registered participants, but the registration is now closed.
Maybe CT sim can help for some preliminary testing: http://www.ctsim.org/
hi, do you solve the problem?I want to do the same simulation recently, can you give me some help?
I ended up adding a combination of Gaussian and Poisson noise with a range of weights. I was able to train a denoising neural network that performed acceptably and moved to other tasks.
Thank you for your reply. Are you add noise in the sinogram data? Can you share your code and your paper?
So just saw your post, actually im working on the same project. i added Poisson Noise into the images which is type of noise produced during Low Dose CT. So i can help you with that.
Saqib Nazir Hi Saqib, how did you measure the existing Poisson Noise level in the high Dose CT image? Thanks！
I found this code. I have tried it and it seems working:
https://github.com/smuzd/LD-CT-simulation
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Software