000301270 001__ 301270
000301270 005__ 20250803021806.0
000301270 0247_ $$2doi$$a10.1007/s11548-025-03370-9
000301270 0247_ $$2pmid$$apmid:40343639
000301270 0247_ $$2ISSN$$a1861-6410
000301270 0247_ $$2ISSN$$a1861-6429
000301270 0247_ $$2altmetric$$aaltmetric:178328022
000301270 037__ $$aDKFZ-2025-00955
000301270 041__ $$aEnglish
000301270 082__ $$a610
000301270 1001_ $$0P:(DE-He78)494ff43d7941675bb715dbe497f23f22$$aKnopp, Marcel$$b0$$eFirst author$$udkfz
000301270 245__ $$aShortcut learning leads to sex bias in deep learning models for photoacoustic tomography.
000301270 260__ $$aHeidelberg [u.a.]$$bSpringer$$c2025
000301270 3367_ $$2DRIVER$$aarticle
000301270 3367_ $$2DataCite$$aOutput Types/Journal article
000301270 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1753779007_3430
000301270 3367_ $$2BibTeX$$aARTICLE
000301270 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000301270 3367_ $$00$$2EndNote$$aJournal Article
000301270 500__ $$a#EA:E130#LA:E130# / 2025 Jul;20(7):1325-1333
000301270 520__ $$aShortcut learning has been identified as a source of algorithmic unfairness in medical imaging artificial intelligence (AI), but its impact on photoacoustic tomography (PAT), particularly concerning sex bias, remains underexplored. This study investigates this issue using peripheral artery disease (PAD) diagnosis as a specific clinical application.To examine the potential for sex bias due to shortcut learning in convolutional neural network (CNNs) and assess how such biases might affect diagnostic predictions, we created training and test datasets with varying PAD prevalence between sexes. Using these datasets, we explored (1) whether CNNs can classify the sex from imaging data, (2) how sex-specific prevalence shifts impact PAD diagnosis performance and underdiagnosis disparity between sexes, and (3) how similarly CNNs encode sex and PAD features.Our study with 147 individuals demonstrates that CNNs can classify the sex from calf muscle PAT images, achieving an AUROC of 0.75. For PAD diagnosis, models trained on data with imbalanced sex-specific disease prevalence experienced significant performance drops (up to 0.21 AUROC) when applied to balanced test sets. Additionally, greater imbalances in sex-specific prevalence within the training data exacerbated underdiagnosis disparities between sexes. Finally, we identify evidence of shortcut learning by demonstrating the effective reuse of learned feature representations between PAD diagnosis and sex classification tasks.CNN-based models trained on PAT data may engage in shortcut learning by leveraging sex-related features, leading to biased and unreliable diagnostic predictions. Addressing demographic-specific prevalence imbalances and preventing shortcut learning is critical for developing models in the medical field that are both accurate and equitable across diverse patient populations.
000301270 536__ $$0G:(DE-HGF)POF4-315$$a315 - Bildgebung und Radioonkologie (POF4-315)$$cPOF4-315$$fPOF IV$$x0
000301270 588__ $$aDataset connected to CrossRef, PubMed, , Journals: inrepo02.dkfz.de
000301270 650_7 $$2Other$$aPeripheral artery disease (PAD)
000301270 650_7 $$2Other$$aPhotoacoustic tomography (PAT)
000301270 650_7 $$2Other$$aSex Bias in AI
000301270 650_7 $$2Other$$aShortcut learning
000301270 7001_ $$0P:(DE-He78)d9c9ea92e3b697685f4b4c3bd6d063ad$$aBender, Christoph Julien$$b1$$eFirst author$$udkfz
000301270 7001_ $$0P:(DE-He78)1c47bf7bdef42ec57b194723ccfb2946$$aHolzwarth, Niklas$$b2$$udkfz
000301270 7001_ $$aLi, Yi$$b3
000301270 7001_ $$aKempf, Julius$$b4
000301270 7001_ $$aCaranovic, Milenko$$b5
000301270 7001_ $$00000-0002-3535-2626$$aKnieling, Ferdinand$$b6
000301270 7001_ $$00000-0003-4114-7589$$aLang, Werner$$b7
000301270 7001_ $$00000-0002-4016-5673$$aRother, Ulrich$$b8
000301270 7001_ $$0P:(DE-He78)a83df473f58a6a8ef43263ec9783ecf0$$aSeitel, Alexander$$b9$$eLast author$$udkfz
000301270 7001_ $$0P:(DE-He78)26a1176cd8450660333a012075050072$$aMaier-Hein, Lena$$b10$$eLast author$$udkfz
000301270 7001_ $$0P:(DE-He78)84acbc6406dd178828f87a8150d40951$$aDreher, Kris$$b11$$eLast author$$udkfz
000301270 773__ $$0PERI:(DE-600)2235881-X$$a10.1007/s11548-025-03370-9$$n7$$p1325-1333$$tInternational journal of computer assisted radiology and surgery$$v20$$x1861-6410$$y2025
000301270 909CO $$ooai:inrepo02.dkfz.de:301270$$pVDB
000301270 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)494ff43d7941675bb715dbe497f23f22$$aDeutsches Krebsforschungszentrum$$b0$$kDKFZ
000301270 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)d9c9ea92e3b697685f4b4c3bd6d063ad$$aDeutsches Krebsforschungszentrum$$b1$$kDKFZ
000301270 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)1c47bf7bdef42ec57b194723ccfb2946$$aDeutsches Krebsforschungszentrum$$b2$$kDKFZ
000301270 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)a83df473f58a6a8ef43263ec9783ecf0$$aDeutsches Krebsforschungszentrum$$b9$$kDKFZ
000301270 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)26a1176cd8450660333a012075050072$$aDeutsches Krebsforschungszentrum$$b10$$kDKFZ
000301270 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)84acbc6406dd178828f87a8150d40951$$aDeutsches Krebsforschungszentrum$$b11$$kDKFZ
000301270 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
000301270 9141_ $$y2025
000301270 915__ $$0StatID:(DE-HGF)3002$$2StatID$$aDEAL Springer$$d2024-12-18$$wger
000301270 915__ $$0StatID:(DE-HGF)3002$$2StatID$$aDEAL Springer$$d2024-12-18$$wger
000301270 915__ $$0StatID:(DE-HGF)0200$$2StatID$$aDBCoverage$$bSCOPUS$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)0300$$2StatID$$aDBCoverage$$bMedline$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)0199$$2StatID$$aDBCoverage$$bClarivate Analytics Master Journal List$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)0160$$2StatID$$aDBCoverage$$bEssential Science Indicators$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)1110$$2StatID$$aDBCoverage$$bCurrent Contents - Clinical Medicine$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)0113$$2StatID$$aWoS$$bScience Citation Index Expanded$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)0150$$2StatID$$aDBCoverage$$bWeb of Science Core Collection$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)0100$$2StatID$$aJCR$$bINT J COMPUT ASS RAD : 2022$$d2024-12-18
000301270 915__ $$0StatID:(DE-HGF)9900$$2StatID$$aIF < 5$$d2024-12-18
000301270 9202_ $$0I:(DE-He78)E130-20160331$$kE130$$lE130 Intelligente Medizinische Systeme$$x0
000301270 9201_ $$0I:(DE-He78)E130-20160331$$kE130$$lE130 Intelligente Medizinische Systeme$$x0
000301270 9200_ $$0I:(DE-He78)E130-20160331$$kE130$$lE130 Intelligente Medizinische Systeme$$x0
000301270 980__ $$ajournal
000301270 980__ $$aVDB
000301270 980__ $$aI:(DE-He78)E130-20160331
000301270 980__ $$aUNRESTRICTED