% IMPORTANT: The following is UTF-8 encoded. This means that in the presence
% of non-ASCII characters, it will not work with BibTeX 0.99 or older.
% Instead, you should use an up-to-date BibTeX implementation like “bibtex8” or
% “biber”.
@ARTICLE{Eulig:293809,
author = {E. Eulig$^*$ and F. Jäger$^*$ and J. Maier$^*$ and B.
Ommer and M. Kachelriess$^*$},
title = {{R}econstructing and analyzing the invariances of low-dose
{CT} image denoising networks.},
journal = {Medical physics},
volume = {52},
number = {1},
issn = {0094-2405},
address = {College Park, Md.},
publisher = {AAPM},
reportid = {DKFZ-2024-01967},
pages = {188-200},
year = {2025},
note = {#EA:E025#LA:E025# / 2025 Jan;52(1):188-200},
abstract = {Deep learning-based methods led to significant advancements
in many areas of medical imaging, most of which are
concerned with the reduction of artifacts caused by motion,
scatter, or noise. However, with most neural networks being
black boxes, they remain notoriously difficult to interpret,
hindering their clinical implementation. In particular, it
has been shown that networks exhibit invariances w.r.t.
input features, that is, they learn to ignore certain
information in the input data.To improve the
interpretability of deep learning-based low-dose CT image
denoising networks.We learn a complete data representation
of low-dose input images using a conditional variational
autoencoder (cVAE). In this representation, invariances of
any given denoising network are then disentangled from the
information it is not invariant to using a conditional
invertible neural network (cINN). At test time, image-space
invariances are generated by applying the inverse of the
cINN and subsequent decoding using the cVAE. We propose two
methods to analyze sampled invariances and to find those
that correspond to alterations of anatomical structures.The
proposed method is applied to four popular deep
learning-based low-dose CT image denoising networks. We find
that the networks are not only invariant to noise amplitude
and realizations, but also to anatomical structures.The
proposed method is capable of reconstructing and analyzing
invariances of deep learning-based low-dose CT image
denoising networks. This is an important step toward
interpreting deep learning-based methods for medical
imaging, which is essential for their clinical
implementation.},
keywords = {computed tomography (Other) / deep learning (Other) /
explainability (Other) / invariances (Other) / low‐dose
(Other) / robustness (Other)},
cin = {E025},
ddc = {610},
cid = {I:(DE-He78)E025-20160331},
pnm = {315 - Bildgebung und Radioonkologie (POF4-315)},
pid = {G:(DE-HGF)POF4-315},
typ = {PUB:(DE-HGF)16},
pubmed = {pmid:39348044},
doi = {10.1002/mp.17413},
url = {https://inrepo02.dkfz.de/record/293809},
}