000302803 001__ 302803
000302803 005__ 20250801105012.0
000302803 0247_ $$2doi$$a10.1186/s44398-025-00003-8
000302803 0247_ $$2pmid$$apmid:40605934
000302803 0247_ $$2pmc$$apmc:PMC12212421
000302803 037__ $$aDKFZ-2025-01343
000302803 041__ $$aEnglish
000302803 1001_ $$0P:(DE-HGF)0$$aMeneghetti, Asier Rabasco$$b0
000302803 245__ $$aEnd-to-end prediction of clinical outcomes in head and neck squamous cell carcinoma with foundation model-based multiple instance learning.
000302803 260__ $$aLondon$$bSpringer$$c2025
000302803 3367_ $$2DRIVER$$aarticle
000302803 3367_ $$2DataCite$$aOutput Types/Journal article
000302803 3367_ $$0PUB:(DE-HGF)16$$2PUB:(DE-HGF)$$aJournal Article$$bjournal$$mjournal$$s1754038170_19094
000302803 3367_ $$2BibTeX$$aARTICLE
000302803 3367_ $$2ORCID$$aJOURNAL_ARTICLE
000302803 3367_ $$00$$2EndNote$$aJournal Article
000302803 500__ $$aBMC Artificial Intelligence (BMC Artif. Intell.) = 3005-1924 
000302803 520__ $$aFoundation models have shown promise in medical AI by learning flexible features from large datasets, offering new opportunities for improving endpoint prediction. However, usage of foundation models for endpoint prediction using routine imaging in head and neck squamous cell carcinoma patients remains unexplored. Within this study, we evaluated the potential of foundation-model based multiple instance learning for prediction of 2-year overall survival, locoregional control and freedom from distant metastasis across three external head and neck squamous cell carcinoma patient cohorts using 2D, multiview and 3D approaches while comparing prediction and stratification performance with handcrafted radiomics and clinical baselines.2D multiple-instance learning models achieved 2-year test area under the receiver-operator curve (AUROC) range of 0.75-0.84 for 2-year overall survival, 0.66-0.75 for 2-year locoregional control and 0.71-0.78 for 2-year freedom from distant metastasis across three different external cohorts, outperforming multiview and 3D multiple instance learning models (AUROC range: 0.50-0.77, p ≥ 0.15) and showing comparable or superior performance to handcrafted radiomics (AUROC range: 0.64-0.74, p ≥ 0.012). Significant stratification was observed from the 2D MIL models (hazard ratios: 2.14-4.77, p ≤ 0.039). 2D MIL models were also shown to learn endpoint-specific correlation patterns such as N-stage for 2-year freedom from distant metastasis prognosis. Multimodal enhancement of 2-year OS/FFDM (AUROC range: 0.82-0.87, p ≤ 0.018) for patients without human papilloma virus positive tumors.FM-based 2D MIL demonstrates promise in HNSCC risk prediction as well as stratification of clinical outcomes. The models match or outperform radiomics baselines, learning clinically-related patterns and showing enhancement of clinical baselines in non-human papilloma virus positive patients.The online version contains supplementary material available at 10.1186/s44398-025-00003-8.
000302803 536__ $$0G:(DE-HGF)POF4-313$$a313 - Krebsrisikofaktoren und Prävention (POF4-313)$$cPOF4-313$$fPOF IV$$x0
000302803 588__ $$aDataset connected to CrossRef, PubMed, , Journals: inrepo02.dkfz.de
000302803 650_7 $$2Other$$aFoundation models
000302803 650_7 $$2Other$$aHead and neck cancer
000302803 650_7 $$2Other$$aMultimodality
000302803 650_7 $$2Other$$aPrognosis
000302803 650_7 $$2Other$$aRadiomics
000302803 7001_ $$aHernández, Marta Ligero$$b1
000302803 7001_ $$aKühn, Jens-Peter$$b2
000302803 7001_ $$0P:(DE-HGF)0$$aLöck, Steffen$$b3
000302803 7001_ $$aCarrero, Zunamys Itzell$$b4
000302803 7001_ $$aPerez-Lopez, Raquel$$b5
000302803 7001_ $$aBressem, Keno K$$b6
000302803 7001_ $$0P:(DE-He78)1e33961c8780aca9b76d776d1fdc1ebb$$aBrinker, Titus$$b7$$udkfz
000302803 7001_ $$aPearson, Alexander T$$b8
000302803 7001_ $$aTruhn, Daniel$$b9
000302803 7001_ $$aNebelung, Sven$$b10
000302803 7001_ $$aKather, Jakob Nikolas$$b11
000302803 773__ $$0PERI:(DE-600)3219788-3$$a10.1186/s44398-025-00003-8$$gVol. 1, no. 1, p. 3$$n1$$p3$$tBMC artificial intelligence$$v1$$x3005-1924$$y2025
000302803 909CO $$ooai:inrepo02.dkfz.de:302803$$pVDB
000302803 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-HGF)0$$aDeutsches Krebsforschungszentrum$$b0$$kDKFZ
000302803 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-HGF)0$$aDeutsches Krebsforschungszentrum$$b3$$kDKFZ
000302803 9101_ $$0I:(DE-588b)2036810-0$$6P:(DE-He78)1e33961c8780aca9b76d776d1fdc1ebb$$aDeutsches Krebsforschungszentrum$$b7$$kDKFZ
000302803 9131_ $$0G:(DE-HGF)POF4-313$$1G:(DE-HGF)POF4-310$$2G:(DE-HGF)POF4-300$$3G:(DE-HGF)POF4$$4G:(DE-HGF)POF$$aDE-HGF$$bGesundheit$$lKrebsforschung$$vKrebsrisikofaktoren und Prävention$$x0
000302803 9141_ $$y2025
000302803 9201_ $$0I:(DE-He78)DD01-20160331$$kDD01$$lDKTK Koordinierungsstelle Dresden$$x0
000302803 9201_ $$0I:(DE-He78)C140-20160331$$kC140$$lDigitale Prävention, Diagnostik und Therapiesteuerung$$x1
000302803 980__ $$ajournal
000302803 980__ $$aVDB
000302803 980__ $$aI:(DE-He78)DD01-20160331
000302803 980__ $$aI:(DE-He78)C140-20160331
000302803 980__ $$aUNRESTRICTED