TY - CHAP A1 - Ganz, Jonathan A1 - Ammeling, Jonas A1 - Rosbach, Emely A1 - Lausser, Ludwig A1 - Bertram, Christof A1 - Breininger, Katharina A1 - Aubreville, Marc ED - Palm, Christoph ED - Breininger, Katharina ED - Deserno, Thomas Martin ED - Handels, Heinz ED - Maier, Andreas ED - Maier-Hein, Klaus H. ED - Tolxdorff, Thomas T1 - Is Self-supervision Enough? BT - Benchmarking Foundation Models Against End-to-end Training for Mitotic Figure Classification T2 - Bildverarbeitung für die Medizin 2025: Proceedings, German Conference on Medical Image Computing, Regensburg March 09–11, 2025 UR - https://doi.org/10.1007/978-3-658-47422-5_15 Y1 - 2025 UR - https://doi.org/10.1007/978-3-658-47422-5_15 SN - 978-3-658-47422-5 SP - 63 EP - 68 PB - Springer Vieweg CY - Wiesbaden ER - TY - INPR A1 - Ammeling, Jonas A1 - Ganz, Jonathan A1 - Rosbach, Emely A1 - Lausser, Ludwig A1 - Bertram, Christof A1 - Breininger, Katharina A1 - Aubreville, Marc T1 - Benchmarking Foundation Models for Mitotic Figure Classification UR - https://doi.org/10.48550/arXiv.2508.04441 Y1 - 2025 UR - https://doi.org/10.48550/arXiv.2508.04441 PB - arXiv CY - Ithaca ER - TY - JOUR A1 - Puget, Chloé A1 - Ganz, Jonathan A1 - Bertram, Christof A1 - Conrad, Thomas A1 - Baeblich, Malte A1 - Voss, Anne A1 - Landmann, Katharina A1 - Haake, Alexander F. H. A1 - Spree, Andreas A1 - Hartung, Svenja A1 - Aeschlimann, Leonore A1 - Soto, Sara A1 - de Brot, Simone A1 - Dettwiler, Martina A1 - Aupperle-Lellbach, Heike A1 - Bolfa, Pompei A1 - Bartel, Alexander A1 - Kiupel, Matti A1 - Breininger, Katharina A1 - Aubreville, Marc A1 - Klopfleisch, Robert T1 - Artificial intelligence predicts c-KIT exon 11 genotype by phenotype in canine cutaneous mast cell tumors: Can human observers learn it? JF - Veterinary Pathology N2 - Canine cutaneous mast cell tumors (ccMCTs) are frequent neoplasms with variable biological behaviors. Internal tandem duplication mutations in c-KIT exon 11 (c-KIT-11-ITD) are associated with poor prognosis but predict therapeutic response to tyrosine kinase inhibitors. In a previous work, deep learning algorithms managed to predict the presence of c-KIT-11-ITD on digitalized hematoxylin and eosin-stained histological slides (whole-slide images, WSIs) in up to 87% of cases, suggesting the existence of morphological features characterizing ccMCTs carrying c-KIT-11-ITD. This 3-stage blinded study aimed to identify morphological features indicative of c-KIT-11-ITD and to evaluate the ability of human observers to learn this task. 17 untrained pathologists first classified 8 WSIs and 200 image patches (highly relevant for algorithmic classification) of ccMCTs as either positive or negative for c-KIT-11-ITD. Second, they self-trained to recognize c-KIT-11-ITD by looking at the same WSIs and patches correctly sorted. Third, pathologists classified 15 new WSIs and 200 new patches according to c-KIT-11-ITD status. In addition, participants reported microscopic features they considered relevant for their decision. Without training, participants correctly classified the c-KIT-11-ITD status of 63%–88% of WSIs and 43%–55% of patches. With self-training, 25%–38% of WSIs and 55%–56% of patches were correctly classified. High cellular pleomorphism, anisokaryosis, and sparse cytoplasmic granulation were commonly suggested as features associated with c-KIT-11-ITD-positive ccMCTs, none of which showed reliable predictivity in a follow-up study. The results indicate that transfer of algorithmic skills to the human observer is difficult. A c-KIT-11-ITD-specific morphological feature remains to be extracted from the artificial intelligence model. UR - https://doi.org/10.1177/03009858251380284 Y1 - 2025 UR - https://doi.org/10.1177/03009858251380284 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-63841 SN - 1544-2217 VL - 63 IS - 2 SP - 369 EP - 379 PB - Sage CY - London ER - TY - JOUR A1 - Rosbach, Emely A1 - Ammeling, Jonas A1 - Ganz, Jonathan A1 - Bertram, Christof A1 - Conrad, Thomas A1 - Riener, Andreas A1 - Aubreville, Marc T1 - Stuck on Suggestions: Automation Bias, the Anchoring Effect, and the Factors That Shape Them in Computational Pathology JF - Machine Learning for Biomedical Imaging N2 - Artificial intelligence (AI)-driven clinical decision support systems (CDSS) hold promise to improve diagnostic accuracy and efficiency in computational pathology. However, collaboration between human experts and AI may give rise to cognitive biases, such as automation and anchoring bias, wherein users may be inclined to blindly adopt system recommendations or be disproportionately influenced by the presence of AI predictions, even when they are inaccurate. These biases may be exacerbated under time pressure, pervasive in routine pathology diagnostics, or shaped by individual user characteristics. To investigate these effects, we conducted a web-based experiment in which trained pathology experts (n = 28) estimated tumor cell percentages twice: once independently and once with the aid of an AI. A subset of the estimates in each condition was performed under time constraints. Our findings indicate that AI integration generally enhances diagnostic performance. However, it also introduced a 7% automation bias rate, quantified as the number of accepted negative consultations, where a previously correct independent assessment gets overturned by inaccurate AI guidance. While time pressure did not increase the frequency of automation bias occurrence, it appeared to intensify its severity, as evidenced by a performance decline linked to increased automation reliance under cognitive load. A linear mixed-effects model (LMM) analysis, simulating weighted averaging, revealed a statistically significant positive coefficient for AI advice, indicating a moderate degree of anchoring on system output. This effect was further intensified under time pressure, suggesting that anchoring bias may become more pronounced when cognitive resources are limited. A secondary LMM evaluation assessing automation reliance, used as a proxy for both automation and anchoring bias, demonstrated that professional experience and self-efficacy were associated with reduced dependence on system support, whereas higher confidence during AI-assisted decision-making was linked to increased automation reliance. Together, these findings underscore the dual nature of AI integration in clinical workflows, offering performance benefits while also introducing risks of cognitive bias–driven diagnostic errors. As an initial investigation focused on a single medical specialty and diagnostic task, this study aims to lay the groundwork for future research to explore these phenomena across diverse clinical contexts, ultimately supporting the establishment of appropriate reliance on automated systems and the safe, effective integration of human–AI collaboration in medical decision-making. UR - https://doi.org/10.59275/j.melba.2026-87b1 Y1 - 2026 UR - https://doi.org/10.59275/j.melba.2026-87b1 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-67787 SN - 2766-905X VL - 3 IS - MELBA–BVM 2025 Special Issue SP - 126 EP - 147 PB - Melba editors CY - [s. l.] ER - TY - JOUR A1 - Puget, Chloé A1 - Ganz, Jonathan A1 - Ostermaier, Julian A1 - Conrad, Thomas A1 - Parlak, Eda A1 - Bertram, Christof A1 - Kiupel, Matti A1 - Breininger, Katharina A1 - Aubreville, Marc A1 - Klopfleisch, Robert T1 - Artificial intelligence can be trained to predict c-KIT-11 mutational status of canine mast cell tumors from hematoxylin and eosin-stained histological slides JF - Veterinary Pathology N2 - Numerous prognostic factors are currently assessed histologically and immunohistochemically in canine mast cell tumors (MCTs) to evaluate clinical behavior. In addition, polymerase chain reaction (PCR) is often performed to detect internal tandem duplication (ITD) mutations in exon 11 of the c-KIT gene ( c-KIT-11-ITD) to predict the therapeutic response to tyrosine kinase inhibitors. This project aimed at training deep learning models (DLMs) to identify MCTs with c-KIT-11-ITD solely based on morphology. Hematoxylin and eosin (HE) stained slides of 368 cutaneous, subcutaneous, and mucocutaneous MCTs (195 with ITD and 173 without) were stained consecutively in 2 different laboratories and scanned with 3 different slide scanners. This resulted in 6 data sets (stain-scanner variations representing diagnostic institutions) of whole-slide images. DLMs were trained with single and mixed data sets and their performances were assessed under stain-scanner variations (domain shifts). The DLM correctly classified HE slides according to their c-KIT-11-ITD status in up to 87% of cases with a 0.90 sensitivity and a 0.83 specificity. A relevant performance drop could be observed when the stain-scanner combination of training and test data set differed. Multi-institutional data sets improved the average accuracy but did not reach the maximum accuracy of algorithms trained and tested on the same stain-scanner variant (ie, intra-institutional). In summary, DLM-based morphological examination can predict c-KIT-11-ITD with high accuracy in canine MCTs in HE slides. However, staining protocol and scanner type influence accuracy. Larger data sets of scans from different laboratories and scanners may lead to more robust DLMs to identify c- KIT mutations in HE slides. UR - https://doi.org/10.1177/03009858241286806 Y1 - 2024 UR - https://doi.org/10.1177/03009858241286806 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-53323 SN - 1544-2217 SN - 0300-9858 VL - 62 IS - 2 SP - 152 EP - 160 PB - Sage CY - London ER - TY - JOUR A1 - Aubreville, Marc A1 - Wilm, Frauke A1 - Stathonikos, Nikolas A1 - Breininger, Katharina A1 - Donovan, Taryn A1 - Jabari, Samir A1 - Veta, Mitko A1 - Ganz, Jonathan A1 - Ammeling, Jonas A1 - van Diest, Paul J A1 - Klopfleisch, Robert A1 - Bertram, Christof T1 - A comprehensive multi-domain dataset for mitotic figure detection JF - Scientific Data N2 - The prognostic value of mitotic figures in tumor tissue is well-established for many tumor types and automating this task is of high research interest. However, especially deep learning-based methods face performance deterioration in the presence of domain shifts, which may arise from different tumor types, slide preparation and digitization devices. We introduce the MIDOG++ dataset, an extension of the MIDOG 2021 and 2022 challenge datasets. We provide region of interest images from 503 histological specimens of seven different tumor types with variable morphology with in total labels for 11,937 mitotic figures: breast carcinoma, lung carcinoma, lymphosarcoma, neuroendocrine tumor, cutaneous mast cell tumor, cutaneous melanoma, and (sub)cutaneous soft tissue sarcoma. The specimens were processed in several laboratories utilizing diverse scanners. We evaluated the extent of the domain shift by using state-of-the-art approaches, observing notable differences in single-domain training. In a leave-one-domain-out setting, generalizability improved considerably. This mitotic figure dataset is the first that incorporates a wide domain shift based on different tumor types, laboratories, whole slide image scanners, and species. UR - https://doi.org/10.1038/s41597-023-02327-4 Y1 - 2023 UR - https://doi.org/10.1038/s41597-023-02327-4 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-40068 SN - 2052-4463 N1 - Author Correction verfügbar unter https://doi.org/10.1038/s41597-024-03548-x VL - 10 PB - Springer CY - London ER - TY - INPR A1 - Aubreville, Marc A1 - Ganz, Jonathan A1 - Ammeling, Jonas A1 - Kaltenecker, Christopher A1 - Bertram, Christof T1 - Model-based Cleaning of the QUILT-1M Pathology Dataset for Text-Conditional Image Synthesis N2 - The QUILT-1M dataset is the first openly available dataset containing images harvested from various online sources. While it provides a huge data variety, the image quality and composition is highly heterogeneous, impacting its utility for text-conditional image synthesis. We propose an automatic pipeline that provides predictions of the most common impurities within the images, e.g., visibility of narrators, desktop environment and pathology software, or text within the image. Additionally, we propose to use semantic alignment filtering of the image-text pairs. Our findings demonstrate that by rigorously filtering the dataset, there is a substantial enhancement of image fidelity in text-to-image tasks. UR - https://doi.org/10.48550/arXiv.2404.07676 Y1 - 2024 UR - https://doi.org/10.48550/arXiv.2404.07676 UR - http://nbn-resolving.de/urn/resolver.pl?urn:nbn:de:bvb:573-48648 PB - arXiv CY - Ithaca ER -