Deep Image Prior for Computed Tomography Reconstruction

TL;DR

This paper reviews the Deep Image Prior framework for CT reconstruction, highlighting its unsupervised approach, various regularisation strategies, and computational enhancements tested on real measurements.

Contribution

It provides a comprehensive overview of DIP applications in CT, detailing algorithmic choices, regularisation techniques, and computational improvements for better reconstruction.

Findings

DIP can effectively reconstruct CT images without large datasets.

Regularisation and early stopping mitigate overfitting in DIP.

Computational strategies reduce reconstruction time significantly.

Abstract

We present a comprehensive overview of the Deep Image Prior (DIP) framework and its applications to image reconstruction in computed tomography. Unlike conventional deep learning methods that rely on large, supervised datasets, the DIP exploits the implicit bias of convolutional neural networks and operates in a fully unsupervised setting, requiring only a single measurement, even in the presence of noise. We describe the standard DIP formulation, outline key algorithmic design choices, and review several strategies to mitigate overfitting, including early stopping, explicit regularisation, and self-guided methods that adapt the network input. In addition, we examine computational improvements such as warm-start and stochastic optimisation methods to reduce the reconstruction time. The discussed methods are tested on real $\mu$CT measurements, which allows examination of trade-offs among the different modifications and extensions.