{"entity": "researcher", "timestamp": "2026-08-20T20:36:40.050Z", "family": "Epstein", "given": "E", "initials": "E", "orcid": "0000-0003-2298-7785", "affiliations": ["Department of Clinical Science and Education, Karolinska Institutet, and Department of Obstetrics and Gynecology, S\u00f6dersjukhuset, Stockholm, Sweden."], "links": {"self": {"href": "https://publications-affiliated.scilifelab.se/researcher/d9e69297d7a644ca9a8520adbcb221aa.json"}, "display": {"href": "https://publications-affiliated.scilifelab.se/researcher/d9e69297d7a644ca9a8520adbcb221aa"}}, "publications": [{"entity": "publication", "iuid": "64786b7b1a6e442b802c4ef4d6fab84e", "links": {"self": {"href": "https://publications-affiliated.scilifelab.se/publication/64786b7b1a6e442b802c4ef4d6fab84e.json"}, "display": {"href": "https://publications-affiliated.scilifelab.se/publication/64786b7b1a6e442b802c4ef4d6fab84e"}}, "title": "International multicenter validation of AI-driven ultrasound detection of ovarian cancer.", "authors": [{"family": "Christiansen", "given": "Filip", "initials": "F", "orcid": "0000-0001-7206-9611", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/0493bb488d2b446093818d5aa859767e.json"}}, {"family": "Konuk", "given": "Emir", "initials": "E", "orcid": "0000-0001-9437-4553", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/0bf9d8966978447bbbe6593a9eb2f22f.json"}}, {"family": "Ganeshan", "given": "Adithya Raju", "initials": "AR", "orcid": "0000-0001-8216-6458", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/e7cb116a36204bd7aacae9716533917c.json"}}, {"family": "Welch", "given": "Robert", "initials": "R", "orcid": "0000-0003-1819-6120", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/c8f85ef18954415a84b7f2a357843762.json"}}, {"family": "Pal\u00e9s Huix", "given": "Joana", "initials": "J", "orcid": "0009-0008-4117-1638", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/368dcc1d7dd84663b19a3b704b23655e.json"}}, {"family": "Czekierdowski", "given": "Artur", "initials": "A"}, {"family": "Leone", "given": "Francesco Paolo Giuseppe", "initials": "FPG"}, {"family": "Haak", "given": "Lucia Anna", "initials": "LA", "orcid": "0000-0001-5749-7968", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/413b52105db74829a13f7f376c0d1b89.json"}}, {"family": "Fruscio", "given": "Robert", "initials": "R", "orcid": "0000-0001-5688-2194", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/4aba55b51c014a5eb5849a0f00a8b2bb.json"}}, {"family": "Gaurilcikas", "given": "Adrius", "initials": "A"}, {"family": "Franchi", "given": "Dorella", "initials": "D", "orcid": "0000-0002-3950-5538", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/feb275cf28ff4ac0919688768ae96593.json"}}, {"family": "Fischerova", "given": "Daniela", "initials": "D", "orcid": "0000-0002-7224-3218", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/059e2fdac40a45fab65d070e7e4c5180.json"}}, {"family": "Mor", "given": "Elisa", "initials": "E"}, {"family": "Savelli", "given": "Luca", "initials": "L"}, {"family": "Pascual", "given": "Maria \u00c0ngela", "initials": "M\u00c0", "orcid": "0000-0001-5095-6981", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/ec22fb2849e745159d4840111f4de04d.json"}}, {"family": "Kudla", "given": "Marek Jerzy", "initials": "MJ"}, {"family": "Guerriero", "given": "Stefano", "initials": "S"}, {"family": "Buonomo", "given": "Francesca", "initials": "F", "orcid": "0000-0002-6587-2622", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/53efb68e37734eae815c896babd2a3b3.json"}}, {"family": "Liuba", "given": "Karina", "initials": "K"}, {"family": "Montik", "given": "Nina", "initials": "N"}, {"family": "Alc\u00e1zar", "given": "Juan Luis", "initials": "JL"}, {"family": "Domali", "given": "Ekaterini", "initials": "E", "orcid": "0000-0001-8899-3040", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/be3702702ceb428e9ba6afe619619792.json"}}, {"family": "Pangilinan", "given": "Nelinda Catherine P", "initials": "NCP"}, {"family": "Carella", "given": "Chiara", "initials": "C", "orcid": "0009-0006-2690-5686", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/fc515c8b291242d98bc34683a2d04484.json"}}, {"family": "Munaretto", "given": "Maria", "initials": "M"}, {"family": "Saskova", "given": "Petra", "initials": "P", "orcid": "0000-0002-5367-8056", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/7dd63e6a856d45b3bda84f236d457278.json"}}, {"family": "Verri", "given": "Debora", "initials": "D", "orcid": "0000-0002-7406-8804", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/aaba8d0838a54b799a2bdee800af10bb.json"}}, {"family": "Visenzi", "given": "Chiara", "initials": "C"}, {"family": "Herman", "given": "Pawel", "initials": "P"}, {"family": "Smith", "given": "Kevin", "initials": "K"}, {"family": "Epstein", "given": "Elisabeth", "initials": "E", "orcid": "0000-0003-2298-7785", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/d9e69297d7a644ca9a8520adbcb221aa.json"}}], "type": "journal article", "published": "2025-01-00", "journal": {"title": "Nat. Med.", "issn": "1546-170X", "volume": "31", "issue": "1", "pages": "189-196", "issn-l": "1078-8956"}, "abstract": "Ovarian lesions are common and often incidentally detected. A critical shortage of expert ultrasound examiners has raised concerns of unnecessary interventions and delayed cancer diagnoses. Deep learning has shown promising results in the detection of ovarian cancer in ultrasound images; however, external validation is lacking. In this international multicenter retrospective study, we developed and validated transformer-based neural network models using a comprehensive dataset of 17,119 ultrasound images from 3,652 patients across 20 centers in eight countries. Using a leave-one-center-out cross-validation scheme, for each center in turn, we trained a model using data from the remaining centers. The models demonstrated robust performance across centers, ultrasound systems, histological diagnoses and patient age groups, significantly outperforming both expert and non-expert examiners on all evaluated metrics, namely F1 score, sensitivity, specificity, accuracy, Cohen's kappa, Matthew's correlation coefficient, diagnostic odds ratio and Youden's J statistic. Furthermore, in a retrospective triage simulation, artificial intelligence (AI)-driven diagnostic support reduced referrals to experts by 63% while significantly surpassing the diagnostic performance of the current practice. These results show that transformer-based models exhibit strong generalization and above human expert-level diagnostic accuracy, with the potential to alleviate the shortage of expert ultrasound examiners and improve patient outcomes.", "doi": "10.1038/s41591-024-03329-4", "pmid": "39747679", "labels": [], "xrefs": [{"db": "pmc", "key": "PMC11750711"}, {"db": "pii", "key": "10.1038/s41591-024-03329-4"}], "notes": [], "created": "2026-08-20T09:02:02.307Z", "modified": "2026-08-20T09:02:02.936Z"}, {"entity": "publication", "iuid": "0e5500571e1c4c20ba054f89b83cca59", "links": {"self": {"href": "https://publications-affiliated.scilifelab.se/publication/0e5500571e1c4c20ba054f89b83cca59.json"}, "display": {"href": "https://publications-affiliated.scilifelab.se/publication/0e5500571e1c4c20ba054f89b83cca59"}}, "title": "Ultrasound image analysis using deep neural networks for discriminating between benign and malignant ovarian tumors: comparison with expert subjective assessment.", "authors": [{"family": "Christiansen", "given": "F", "initials": "F"}, {"family": "Epstein", "given": "E L", "initials": "EL"}, {"family": "Smedberg", "given": "E", "initials": "E"}, {"family": "\u00c5kerlund", "given": "M", "initials": "M"}, {"family": "Smith", "given": "K", "initials": "K"}, {"family": "Epstein", "given": "E", "initials": "E", "orcid": "0000-0003-2298-7785", "researcher": {"href": "https://publications-affiliated.scilifelab.se/researcher/d9e69297d7a644ca9a8520adbcb221aa.json"}}], "type": "journal article", "published": "2021-01-00", "journal": {"title": "Ultrasound Obstet Gynecol", "issn": "1469-0705", "volume": "57", "issue": "1", "pages": "155-163", "issn-l": null}, "abstract": "To develop and test the performance of computerized ultrasound image analysis using deep neural networks (DNNs) in discriminating between benign and malignant ovarian tumors and to compare its diagnostic accuracy with that of subjective assessment (SA) by an ultrasound expert.\n\nWe included 3077 (grayscale, n = 1927; power Doppler, n = 1150) ultrasound images from 758 women with ovarian tumors, who were classified prospectively by expert ultrasound examiners according to IOTA (International Ovarian Tumor Analysis) terms and definitions. Histological outcome from surgery (n = 634) or long-term (\u2265 3 years) follow-up (n = 124) served as the gold standard. The dataset was split into a training set (n = 508; 314 benign and 194 malignant), a validation set (n = 100; 60 benign and 40 malignant) and a test set (n = 150; 75 benign and 75 malignant). We used transfer learning on three pre-trained DNNs: VGG16, ResNet50 and MobileNet. Each model was trained, and the outputs calibrated, using temperature scaling. An ensemble of the three models was then used to estimate the probability of malignancy based on all images from a given case. The DNN ensemble classified the tumors as benign or malignant (Ovry-Dx1 model); or as benign, inconclusive or malignant (Ovry-Dx2 model). The diagnostic performance of the DNN models, in terms of sensitivity and specificity, was compared to that of SA for classifying ovarian tumors in the test set.\n\nAt a sensitivity of 96.0%, Ovry-Dx1 had a specificity similar to that of SA (86.7% vs 88.0%; P = 1.0). Ovry-Dx2 had a sensitivity of 97.1% and a specificity of 93.7%, when designating 12.7% of the lesions as inconclusive. By complimenting Ovry-Dx2 with SA in inconclusive cases, the overall sensitivity (96.0%) and specificity (89.3%) were not significantly different from using SA in all cases (P = 1.0).\n\nUltrasound image analysis using DNNs can predict ovarian malignancy with a diagnostic accuracy comparable to that of human expert examiners, indicating that these models may have a role in the triage of women with an ovarian tumor. \u00a9 2020 The Authors. Ultrasound in Obstetrics & Gynecology published by John Wiley & Sons Ltd on behalf of International Society of Ultrasound in Obstetrics and Gynecology.", "doi": "10.1002/uog.23530", "pmid": "33142359", "labels": [], "xrefs": [{"db": "pmc", "key": "PMC7839489"}], "notes": [], "created": "2026-08-20T06:35:48.474Z", "modified": "2026-08-20T06:35:48.587Z"}]}