Uncertainty Estimation in Medical Image Denoising with Bayesian Deep Image Prior

Understand

Uncertainty quantification in inverse medical imaging tasks with deep learning has received little attention.

• However, deep models trained on large data sets tend to hallucinate and create artifacts in the reconstructed output that are not anatomically present.
• We use a randomly initialized convolutional network as parameterization of the reconstructed image and perform gradient descent to match the observation, which is known as deep image prior.
• In this case, the reconstruction does not suffer from hallucinations as no prior training is performed.