On hallucinations in tomographic image reconstruction

TL;DR

This paper investigates how deep learning-based priors influence tomographic image reconstruction, highlighting potential hallucinations and proposing a formalism to analyze the prior's effects on reconstructed images.

Contribution

It introduces a decomposition method and hallucination map to analyze the influence of learned priors in regularized tomographic reconstruction.

Findings

Deep neural network priors can cause false structures in images.

The proposed formalism helps understand the prior's impact on reconstruction.

Numerical studies demonstrate the effects of different methods on hallucinations.

Abstract

Tomographic image reconstruction is generally an ill-posed linear inverse problem. Such ill-posed inverse problems are typically regularized using prior knowledge of the sought-after object property. Recently, deep neural networks have been actively investigated for regularizing image reconstruction problems by learning a prior for the object properties from training images. However, an analysis of the prior information learned by these deep networks and their ability to generalize to data that may lie outside the training distribution is still being explored. An inaccurate prior might lead to false structures being hallucinated in the reconstructed image and that is a cause for serious concern in medical imaging. In this work, we propose to illustrate the effect of the prior imposed by a reconstruction method by decomposing the image estimate into generalized measurement and null components. The concept of a hallucination map is introduced for the general purpose of understanding the effect of the prior in regularized reconstruction methods. Numerical studies are conducted corresponding to a stylized tomographic imaging modality. The behavior of different reconstruction methods under the proposed formalism is discussed with the help of the numerical studies.