SVD-DIP: Overcoming the Overfitting Problem in DIP-based CT Reconstruction

Marco Nittscher, Michael Falk Lameter, Riccardo Barbano, Johannes Leuschner, Bangti Jin, Peter Maass

Keywords: deep image prior, fine-tuning, computed tomography, singular value decomposition

TL;DR: Fine-tuning only the singular values of conv-layers prevents degradation of DIP+TV.

Abstract: The deep image prior (DIP) is a well-established unsupervised deep learning method for image reconstruction; yet it is far from being flawless. The DIP overfits to noise if not early stopped, or optimized via a regularized objective. We build on the regularized fine-tuning of a pretrained DIP, by adopting a novel strategy that restricts the learning to the adaptation of singular values. The proposed SVD-DIP uses ad hoc convolutional layers whose pretrained parameters are decomposed via the singular value decomposition. Optimizing the DIP then solely consists in the fine-tuning of the singular values, while keeping the left and right singular vectors fixed. We thoroughly validate the proposed method on real-measured μCT data of a lotus root as well as two medical datasets (LoDoPaB and Mayo). We report significantly improved stability of the DIP optimization, by overcoming the overfitting to noise.