000293809 001__ 293809
000293809 005__ 20250107162137.0
000293809 0247_ $$2doi$$a10.1002/mp.17413
000293809 0247_ $$2pmid$$apmid:39348044
000293809 0247_ $$2ISSN$$a0094-2405
000293809 0247_ $$2ISSN$$a1522-8541
000293809 0247_ $$2ISSN$$a2473-4209
000293809 037__ $$aDKFZ-2024-01967
000293809 041__ $$aEnglish
000293809 082__ $$a610
000293809 1001_ $$0P:(DE-He78)9f72962d56560b57e5630885c8f8b31d$$aEulig, Elias$$b0$$eFirst author$$udkfz
000293809 245__ $$aReconstructing and analyzing the invariances of low-dose CT image denoising networks.
000293809 260__ $$aCollege Park, Md.$$bAAPM$$c2025
000293809 3367_ $$2DRIVER$$aarticle
000293809 3367_ $$2DataCite$$aOutput Types/Journal article
000293809 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1736263248_4144
000293809 3367_ $$2BibTeX$$aARTICLE
000293809 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000293809 3367_ $$00$$2EndNote$$aJournal Article
000293809 500__ $$a#EA:E025#LA:E025# / 2025 Jan;52(1):188-200
000293809 520__ $$aDeep learning-based methods led to significant advancements in many areas of medical imaging, most of which are concerned with the reduction of artifacts caused by motion, scatter, or noise. However, with most neural networks being black boxes, they remain notoriously difficult to interpret, hindering their clinical implementation. In particular, it has been shown that networks exhibit invariances w.r.t. input features, that is, they learn to ignore certain information in the input data.To improve the interpretability of deep learning-based low-dose CT image denoising networks.We learn a complete data representation of low-dose input images using a conditional variational autoencoder (cVAE). In this representation, invariances of any given denoising network are then disentangled from the information it is not invariant to using a conditional invertible neural network (cINN). At test time, image-space invariances are generated by applying the inverse of the cINN and subsequent decoding using the cVAE. We propose two methods to analyze sampled invariances and to find those that correspond to alterations of anatomical structures.The proposed method is applied to four popular deep learning-based low-dose CT image denoising networks. We find that the networks are not only invariant to noise amplitude and realizations, but also to anatomical structures.The proposed method is capable of reconstructing and analyzing invariances of deep learning-based low-dose CT image denoising networks. This is an important step toward interpreting deep learning-based methods for medical imaging, which is essential for their clinical implementation.
000293809 536__ $$0G:(DE-HGF)POF4-315$$a315 - Bildgebung und Radioonkologie (POF4-315)$$cPOF4-315$$fPOF IV$$x0
000293809 588__ $$aDataset connected to CrossRef, PubMed, , Journals: inrepo02.dkfz.de
000293809 650_7 $$2Other$$acomputed tomography
000293809 650_7 $$2Other$$adeep learning
000293809 650_7 $$2Other$$aexplainability
000293809 650_7 $$2Other$$ainvariances
000293809 650_7 $$2Other$$alow‐dose
000293809 650_7 $$2Other$$arobustness
000293809 7001_ $$0P:(DE-He78)ceed8085bb2faf0ec822fe89e8a085d3$$aJäger, Fabian$$b1$$udkfz
000293809 7001_ $$0P:(DE-He78)3c462b1378ce0906e7320c94e514abfa$$aMaier, Joscha$$b2$$udkfz
000293809 7001_ $$aOmmer, Björn$$b3
000293809 7001_ $$0P:(DE-He78)f288a8f92f092ddb41d52b1aeb915323$$aKachelriess, Marc$$b4$$eLast author$$udkfz
000293809 773__ $$0PERI:(DE-600)1466421-5$$a10.1002/mp.17413$$gp. mp.17413$$n1$$p188-200$$tMedical physics$$v52$$x0094-2405$$y2025
000293809 909CO $$ooai:inrepo02.dkfz.de:293809$$pVDB
000293809 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)9f72962d56560b57e5630885c8f8b31d$$aDeutsches Krebsforschungszentrum$$b0$$kDKFZ
000293809 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)ceed8085bb2faf0ec822fe89e8a085d3$$aDeutsches Krebsforschungszentrum$$b1$$kDKFZ
000293809 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)3c462b1378ce0906e7320c94e514abfa$$aDeutsches Krebsforschungszentrum$$b2$$kDKFZ
000293809 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)f288a8f92f092ddb41d52b1aeb915323$$aDeutsches Krebsforschungszentrum$$b4$$kDKFZ
000293809 9131_ $$0G:(DE-HGF)POF4-315$$1G:(DE-HGF)POF4-310$$2G:(DE-HGF)POF4-300$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$aDE-HGF$$bGesundheit$$lKrebsforschung$$vBildgebung und Radioonkologie$$x0
000293809 9141_ $$y2024
000293809 915__ $$0StatID:(DE-HGF)3001$$2StatID$$aDEAL Wiley$$d2023-10-21$$wger
000293809 915__ $$0StatID:(DE-HGF)0100$$2StatID$$aJCR$$bMED PHYS : 2022$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0200$$2StatID$$aDBCoverage$$bSCOPUS$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0300$$2StatID$$aDBCoverage$$bMedline$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0320$$2StatID$$aDBCoverage$$bPubMed Central$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0600$$2StatID$$aDBCoverage$$bEbsco Academic Search$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0030$$2StatID$$aPeer Review$$bASC$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0199$$2StatID$$aDBCoverage$$bClarivate Analytics Master Journal List$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0113$$2StatID$$aWoS$$bScience Citation Index Expanded$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0150$$2StatID$$aDBCoverage$$bWeb of Science Core Collection$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)1030$$2StatID$$aDBCoverage$$bCurrent Contents - Life Sciences$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)0160$$2StatID$$aDBCoverage$$bEssential Science Indicators$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)1110$$2StatID$$aDBCoverage$$bCurrent Contents - Clinical Medicine$$d2023-10-21
000293809 915__ $$0StatID:(DE-HGF)9900$$2StatID$$aIF < 5$$d2023-10-21
000293809 9202_ $$0I:(DE-He78)E025-20160331$$kE025$$lE025 Röntgenbildgebung und Computertomographie$$x0
000293809 9201_ $$0I:(DE-He78)E025-20160331$$kE025$$lE025 Röntgenbildgebung und Computertomographie$$x0
000293809 9200_ $$0I:(DE-He78)E025-20160331$$kE025$$lE025 Röntgenbildgebung und Computertomographie$$x0
000293809 980__ $$ajournal
000293809 980__ $$aVDB
000293809 980__ $$aI:(DE-He78)E025-20160331
000293809 980__ $$aUNRESTRICTED