@inproceedings{RosbachAmmelingKruegeletal.2025, author = {Rosbach, Emely and Ammeling, Jonas and Kr{\"u}gel, Sebastian and Kießig, Angelika and Fritz, Alexis and Ganz, Jonathan and Puget, Chlo{\´e} and Donovan, Taryn and Klang, Andrea and K{\"o}ller, Maximilian C. and Bolfa, Pompei and Tecilla, Marco and Denk, Daniela and Kiupel, Matti and Paraschou, Georgios and Kok, Mun Keong and Haake, Alexander F. H. and de Krijger, Ronald R. and Sonnen, Andreas F.-P. and Kasantikul, Tanit and Dorrestein, Gerry M. and Smedley, Rebecca C. and Stathonikos, Nikolas and Uhl, Matthias and Bertram, Christof and Riener, Andreas and Aubreville, Marc}, title = {"When Two Wrongs Don't Make a Right" - Examining Confirmation Bias and the Role of Time Pressure During Human-AI Collaboration in Computational Pathology}, pages = {528}, booktitle = {CHI'25: Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems}, editor = {Yamashita, Naomi and Evers, Vanessa and Yatani, Koji and Ding, Xianghua and Lee, Bongshin and Chetty, Marshini and Toups-Dugas, Phoebe}, publisher = {ACM}, address = {New York}, isbn = {979-8-4007-1394-1}, doi = {https://doi.org/10.1145/3706598.3713319}, year = {2025}, abstract = {Artificial intelligence (AI)-based decision support systems hold promise for enhancing diagnostic accuracy and efficiency in computational pathology. However, human-AI collaboration can introduce and amplify cognitive biases, like confirmation bias caused by false confirmation when erroneous human opinions are reinforced by inaccurate AI output. This bias may increase under time pressure, a ubiquitous factor in routine pathology, as it strains practitioners' cognitive resources. We quantified confirmation bias triggered by AI-induced false confirmation and examined the role of time constraints in a web-based experiment, where trained pathology experts (n=28) estimated tumor cell percentages. Our results suggest that AI integration fuels confirmation bias, evidenced by a statistically significant positive linear-mixed-effects model coefficient linking AI recommendations mirroring flawed human judgment and alignment with system advice. Conversely, time pressure appeared to weaken this relationship. These findings highlight potential risks of AI in healthcare and aim to support the safe integration of clinical decision support systems.}, language = {en} } @unpublished{GanzMarzahlAmmelingetal.2024, author = {Ganz, Jonathan and Marzahl, Christian and Ammeling, Jonas and Richter, Barbara and Puget, Chlo{\´e} and Denk, Daniela and Demeter, Elena A. and Tabaran, Flaviu A. and Wasinger, Gabriel and Lipnik, Karoline and Tecilla, Marco and Valentine, Matthew J. and Dark, Michael and Abele, Niklas and Bolfa, Pompei and Erber, Ramona and Klopfleisch, Robert and Merz, Sophie and Donovan, Taryn and Jabari, Samir and Bertram, Christof and Breininger, Katharina and Aubreville, Marc}, title = {On the Value of PHH3 for Mitotic Figure Detection on H\&E-stained Images}, publisher = {arXiv}, address = {Ithaca}, doi = {https://doi.org/10.48550/arXiv.2406.19899}, year = {2024}, abstract = {The count of mitotic figures (MFs) observed in hematoxylin and eosin (H\&E)-stained slides is an important prognostic marker as it is a measure for tumor cell proliferation. However, the identification of MFs has a known low inter-rater agreement. Deep learning algorithms can standardize this task, but they require large amounts of annotated data for training and validation. Furthermore, label noise introduced during the annotation process may impede the algorithm's performance. Unlike H\&E, the mitosis-specific antibody phospho-histone H3 (PHH3) specifically highlights MFs. Counting MFs on slides stained against PHH3 leads to higher agreement among raters and has therefore recently been used as a ground truth for the annotation of MFs in H\&E. However, as PHH3 facilitates the recognition of cells indistinguishable from H\&E stain alone, the use of this ground truth could potentially introduce noise into the H\&E-related dataset, impacting model performance. This study analyzes the impact of PHH3-assisted MF annotation on inter-rater reliability and object level agreement through an extensive multi-rater experiment. We found that the annotators' object-level agreement increased when using PHH3-assisted labeling. Subsequently, MF detectors were evaluated on the resulting datasets to investigate the influence of PHH3-assisted labeling on the models' performance. Additionally, a novel dual-stain MF detector was developed to investigate the interpretation-shift of PHH3-assisted labels used in H\&E, which clearly outperformed single-stain detectors. However, the PHH3-assisted labels did not have a positive effect on solely H\&E-based models. The high performance of our dual-input detector reveals an information mismatch between the H\&E and PHH3-stained images as the cause of this effect.}, language = {en} } @unpublished{PugetGanzOstermaieretal.2024, author = {Puget, Chlo{\´e} and Ganz, Jonathan and Ostermaier, Julian and Konrad, Thomas and Parlak, Eda and Bertram, Christof and Kiupel, Matti and Breininger, Katharina and Aubreville, Marc and Klopfleisch, Robert}, title = {Deep Learning model predicts the c-Kit-11 mutational status of canine cutaneous mast cell tumors by HE stained histological slides}, publisher = {arXiv}, address = {Ithaca}, doi = {https://doi.org/10.48550/arXiv.2401.06169}, year = {2024}, abstract = {Numerous prognostic factors are currently assessed histopathologically in biopsies of canine mast cell tumors to evaluate clinical behavior. In addition, PCR analysis of the c-Kit exon 11 mutational status is often performed to evaluate the potential success of a tyrosine kinase inhibitor therapy. This project aimed at training deep learning models (DLMs) to identify the c-Kit-11 mutational status of MCTs solely based on morphology without additional molecular analysis. HE slides of 195 mutated and 173 non-mutated tumors were stained consecutively in two different laboratories and scanned with three different slide scanners. This resulted in six different datasets (stain-scanner variations) of whole slide images. DLMs were trained with single and mixed datasets and their performances was assessed under scanner and staining domain shifts. The DLMs correctly classified HE slides according to their c-Kit 11 mutation status in, on average, 87\% of cases for the best-suited stain-scanner variant. A relevant performance drop could be observed when the stain-scanner combination of the training and test dataset differed. Multi-variant datasets improved the average accuracy but did not reach the maximum accuracy of algorithms trained and tested on the same stain-scanner variant. In summary, DLM-assisted morphological examination of MCTs can predict c-Kit-exon 11 mutational status of MCTs with high accuracy. However, the recognition performance is impeded by a change of scanner or staining protocol. Larger data sets with higher numbers of scans originating from different laboratories and scanners may lead to more robust DLMs to identify c-Kit mutations in HE slides.}, language = {en} } @unpublished{GanzMarzahlAmmelingetal.2024, author = {Ganz, Jonathan and Marzahl, Christian and Ammeling, Jonas and Rosbach, Emely and Richter, Barbara and Puget, Chlo{\´e} and Denk, Daniela and Demeter, Elena A. and Tabaran, Flaviu A. and Wasinger, Gabriel and Lipnik, Karoline and Tecilla, Marco and Valentine, Matthew J. and Dark, Michael and Abele, Niklas and Bolfa, Pompei and Erber, Ramona and Klopfleisch, Robert and Merz, Sophie and Donovan, Taryn and Jabari, Samir and Bertram, Christof and Breininger, Katharina and Aubreville, Marc}, title = {Information Mismatch in PHH3-Assisted Mitosis Annotation Leads to Interpretation Shifts in H\&E Slide Analysis}, titleParent = {Research Square}, publisher = {Research Square}, address = {Durham}, doi = {https://doi.org/10.21203/rs.3.rs-4900505/v1}, year = {2024}, abstract = {The count of mitotic figures (MFs) observed in hematoxylin and eosin (H\&E)-stained slides is an important prognostic marker, as it is a measure for tumor cell proliferation. However, the identification of MFs has a known low inter-rater agreement. In a computer-aided setting, deep learning algorithms can help to mitigate this, but they require large amounts of annotated data for training and validation. Furthermore, label noise introduced during the annotation process may impede the algorithms' performance. Unlike H\&E, where identification of MFs is based mainly on morphological features, the mitosis-specific antibody phospho-histone H3 (PHH3) specifically highlights MFs. Counting MFs on slides stained against PHH3 leads to higher agreement among raters and has therefore recently been used as a ground truth for the annotation of MFs in H\&E. However, as PHH3 facilitates the recognition of cells indistinguishable from H\&E staining alone, the use of this ground truth could potentially introduce an interpretation shift and even label noise into the H\&E-related dataset, impacting model performance. This study analyzes the impact of PHH3-assisted MF annotation on inter-rater reliability and object level agreement through an extensive multi-rater experiment. Subsequently, MF detectors, including a novel dual-stain detector, were evaluated on the resulting datasets to investigate the influence of PHH3-assisted labeling on the models' performance. We found that the annotators' object-level agreement significantly increased when using PHH3-assisted labeling (F1: 0.53 to 0.74). However, this enhancement in label consistency did not translate to improved performance for H\&E-based detectors, neither during the training phase nor the evaluation phase. Conversely, the dual-stain detector was able to benefit from the higher consistency. This reveals an information mismatch between the H\&E and PHH3-stained images as the cause of this effect, which renders PHH3-assisted annotations not well-aligned for use with H\&E-based detectors. Based on our findings, we propose an improved PHH3-assisted labeling procedure.}, language = {en} } @article{WilmFragosoGarciaMarzahletal.2022, author = {Wilm, Frauke and Fragoso-Garcia, Marco and Marzahl, Christian and Qiu, Jingna and Puget, Chlo{\´e} and Diehl, Laura and Bertram, Christof and Klopfleisch, Robert and Maier, Andreas and Breininger, Katharina and Aubreville, Marc}, title = {Pan-tumor CAnine cuTaneous Cancer Histology (CATCH) dataset}, volume = {9}, pages = {588}, journal = {Scientific Data}, publisher = {Springer Nature}, address = {New York}, issn = {2052-4463}, doi = {https://doi.org/10.1038/s41597-022-01692-w}, year = {2022}, abstract = {Due to morphological similarities, the differentiation of histologic sections of cutaneous tumors into individual subtypes can be challenging. Recently, deep learning-based approaches have proven their potential for supporting pathologists in this regard. However, many of these supervised algorithms require a large amount of annotated data for robust development. We present a publicly available dataset of 350 whole slide images of seven different canine cutaneous tumors complemented by 12,424 polygon annotations for 13 histologic classes, including seven cutaneous tumor subtypes. In inter-rater experiments, we show a high consistency of the provided labels, especially for tumor annotations. We further validate the dataset by training a deep neural network for the task of tissue segmentation and tumor subtype classification. We achieve a class-averaged Jaccard coefficient of 0.7047, and 0.9044 for tumor in particular. For classification, we achieve a slide-level accuracy of 0.9857. Since canine cutaneous tumors possess various histologic homologies to human tumors the added value of this dataset is not limited to veterinary pathology but extends to more general fields of application.}, language = {en} } @article{GanzMarzahlAmmelingetal.2024, author = {Ganz, Jonathan and Marzahl, Christian and Ammeling, Jonas and Rosbach, Emely and Richter, Barbara and Puget, Chlo{\´e} and Denk, Daniela and Demeter, Elena A. and Tabaran, Flaviu A. and Wasinger, Gabriel and Lipnik, Karoline and Tecilla, Marco and Valentine, Matthew J. and Dark, Michael and Abele, Niklas and Bolfa, Pompei and Erber, Ramona and Klopfleisch, Robert and Merz, Sophie and Donovan, Taryn and Jabari, Samir and Bertram, Christof and Breininger, Katharina and Aubreville, Marc}, title = {Information mismatch in PHH3-assisted mitosis annotation leads to interpretation shifts in H\&E slide analysis}, volume = {14}, pages = {26273}, journal = {Scientific Reports}, number = {1}, publisher = {Springer Nature}, address = {London}, issn = {2045-2322}, doi = {https://doi.org/10.1038/s41598-024-77244-6}, year = {2024}, abstract = {The count of mitotic figures (MFs) observed in hematoxylin and eosin (H\&E)-stained slides is an important prognostic marker, as it is a measure for tumor cell proliferation. However, the identification of MFs has a known low inter-rater agreement. In a computer-aided setting, deep learning algorithms can help to mitigate this, but they require large amounts of annotated data for training and validation. Furthermore, label noise introduced during the annotation process may impede the algorithms' performance. Unlike H\&E, where identification of MFs is based mainly on morphological features, the mitosis-specific antibody phospho-histone H3 (PHH3) specifically highlights MFs. Counting MFs on slides stained against PHH3 leads to higher agreement among raters and has therefore recently been used as a ground truth for the annotation of MFs in H\&E. However, as PHH3 facilitates the recognition of cells indistinguishable from H\&E staining alone, the use of this ground truth could potentially introduce an interpretation shift and even label noise into the H\&E-related dataset, impacting model performance. This study analyzes the impact of PHH3-assisted MF annotation on inter-rater reliability and object level agreement through an extensive multi-rater experiment. Subsequently, MF detectors, including a novel dual-stain detector, were evaluated on the resulting datasets to investigate the influence of PHH3-assisted labeling on the models' performance. We found that the annotators' object-level agreement significantly increased when using PHH3-assisted labeling (F1: 0.53 to 0.74). However, this enhancement in label consistency did not translate to improved performance for H\&E-based detectors, neither during the training phase nor the evaluation phase. Conversely, the dual-stain detector was able to benefit from the higher consistency. This reveals an information mismatch between the H\&E and PHH3-stained images as the cause of this effect, which renders PHH3-assisted annotations not well-aligned for use with H\&E-based detectors. Based on our findings, we propose an improved PHH3-assisted labeling procedure.}, language = {en} } @article{WilmIhlingMehesetal.2023, author = {Wilm, Frauke and Ihling, Christian and M{\´e}hes, G{\´a}bor and Terracciano, Luigi and Puget, Chlo{\´e} and Klopfleisch, Robert and Sch{\"u}ffler, Peter and Aubreville, Marc and Maier, Andreas and Mrowiec, Thomas and Breininger, Katharina}, title = {Pan-tumor T-lymphocyte detection using deep neural networks: Recommendations for transfer learning in immunohistochemistry}, volume = {2023}, pages = {100301}, journal = {Journal of Pathology Informatics}, number = {14}, publisher = {Elsevier}, address = {Amsterdam}, issn = {2153-3539}, doi = {https://doi.org/10.1016/j.jpi.2023.100301}, year = {2023}, abstract = {The success of immuno-oncology treatments promises long-term cancer remission for an increasing number of patients. The response to checkpoint inhibitor drugs has shown a correlation with the presence of immune cells in the tumor and tumor microenvironment. An in-depth understanding of the spatial localization of immune cells is therefore critical for understanding the tumor's immune landscape and predicting drug response. Computer-aided systems are well suited for efficiently quantifying immune cells in their spatial context. Conventional image analysis approaches are often based on color features and therefore require a high level of manual interaction. More robust image analysis methods based on deep learning are expected to decrease this reliance on human interaction and improve the reproducibility of immune cell scoring. However, these methods require sufficient training data and previous work has reported low robustness of these algorithms when they are tested on out-of-distribution data from different pathology labs or samples from different organs. In this work, we used a new image analysis pipeline to explicitly evaluate the robustness of marker-labeled lymphocyte quantification algorithms depending on the number of training samples before and after being transferred to a new tumor indication. For these experiments, we adapted the RetinaNet architecture for the task of T-lymphocyte detection and employed transfer learning to bridge the domain gap between tumor indications and reduce the annotation costs for unseen domains. On our test set, we achieved human-level performance for almost all tumor indications with an average precision of 0.74 in-domain and 0.72-0.74 cross-domain. From our results, we derive recommendations for model development regarding annotation extent, training sample selection, and label extraction for the development of robust algorithms for immune cell scoring. By extending the task of marker-labeled lymphocyte quantification to a multi-class detection task, the pre-requisite for subsequent analyses, e.g., distinguishing lymphocytes in the tumor stroma from tumor-infiltrating lymphocytes, is met.}, language = {en} } @article{PugetGanzOstermaieretal.2024, author = {Puget, Chlo{\´e} and Ganz, Jonathan and Ostermaier, Julian and Conrad, Thomas and Parlak, Eda and Bertram, Christof and Kiupel, Matti and Breininger, Katharina and Aubreville, Marc and Klopfleisch, Robert}, title = {Artificial intelligence can be trained to predict c-KIT-11 mutational status of canine mast cell tumors from hematoxylin and eosin-stained histological slides}, volume = {62}, journal = {Veterinary Pathology}, number = {2}, publisher = {Sage}, address = {London}, issn = {1544-2217}, doi = {https://doi.org/10.1177/03009858241286806}, pages = {152 -- 160}, year = {2024}, abstract = {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.}, language = {en} } @article{FragosoGarciaWilmBertrametal.2023, author = {Fragoso-Garcia, Marco and Wilm, Frauke and Bertram, Christof and Merz, Sophie and Schmidt, Anja and Donovan, Taryn and Fuchs-Baumgartinger, Andrea and Bartel, Alexander and Marzahl, Christian and Diehl, Laura and Puget, Chloe and Maier, Andreas and Aubreville, Marc and Breininger, Katharina and Klopfleisch, Robert}, title = {Automated diagnosis of 7 canine skin tumors using machine learning on H\&E-stained whole slide images}, volume = {60}, journal = {Veterinary Pathology}, number = {6}, publisher = {Sage}, address = {London}, issn = {0300-9858}, doi = {https://doi.org/10.1177/03009858231189205}, pages = {865 -- 875}, year = {2023}, abstract = {Microscopic evaluation of hematoxylin and eosin-stained slides is still the diagnostic gold standard for a variety of diseases, including neoplasms. Nevertheless, intra- and interrater variability are well documented among pathologists. So far, computer assistance via automated image analysis has shown potential to support pathologists in improving accuracy and reproducibility of quantitative tasks. In this proof of principle study, we describe a machine-learning-based algorithm for the automated diagnosis of 7 of the most common canine skin tumors: trichoblastoma, squamous cell carcinoma, peripheral nerve sheath tumor, melanoma, histiocytoma, mast cell tumor, and plasmacytoma. We selected, digitized, and annotated 350 hematoxylin and eosin-stained slides (50 per tumor type) to create a database divided into training, n = 245 whole-slide images (WSIs), validation ( n = 35 WSIs), and test sets ( n = 70 WSIs). Full annotations included the 7 tumor classes and 6 normal skin structures. The data set was used to train a convolutional neural network (CNN) for the automatic segmentation of tumor and nontumor classes. Subsequently, the detected tumor regions were classified patch-wise into 1 of the 7 tumor classes. A majority of patches-approach led to a tumor classification accuracy of the network on the slide-level of 95\% (133/140 WSIs), with a patch-level precision of 85\%. The same 140 WSIs were provided to 6 experienced pathologists for diagnosis, who achieved a similar slide-level accuracy of 98\% (137/140 correct majority votes). Our results highlight the feasibility of artificial intelligence-based methods as a support tool in diagnostic oncologic pathology with future applications in other species and tumor types.}, language = {en} } @article{PugetGanzBertrametal.2025, author = {Puget, Chlo{\´e} and Ganz, Jonathan and Bertram, Christof and Conrad, Thomas and Baeblich, Malte and Voss, Anne and Landmann, Katharina and Haake, Alexander F. H. and Spree, Andreas and Hartung, Svenja and Aeschlimann, Leonore and Soto, Sara and de Brot, Simone and Dettwiler, Martina and Aupperle-Lellbach, Heike and Bolfa, Pompei and Bartel, Alexander and Kiupel, Matti and Breininger, Katharina and Aubreville, Marc and Klopfleisch, Robert}, title = {Artificial intelligence predicts c-KIT exon 11 genotype by phenotype in canine cutaneous mast cell tumors: Can human observers learn it?}, volume = {63}, journal = {Veterinary Pathology}, number = {2}, publisher = {Sage}, address = {London}, issn = {1544-2217}, doi = {https://doi.org/10.1177/03009858251380284}, pages = {369 -- 379}, year = {2025}, abstract = {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.}, language = {en} }