Automated Nuclear Morphometry: A Deep Learning Approach for Prognostication in Canine Pulmonary Carcinoma to Enhance Reproducibility.

The integration of deep learning-based tools into diagnostic workflows is increasingly prevalent due to their efficiency and reproducibility in various settings. We investigated the utility of automated nuclear morphometry for assessing nuclear pleomorphism (NP), a criterion of malignancy in the current grading system in canine pulmonary carcinoma (cPC), and its prognostic implications. We developed a deep learning-based algorithm for evaluating NP (variation in size, i.e., anisokaryosis and/or shape) using a segmentation model. Its performance was evaluated on 46 cPC cases with comprehensive follow-up data regarding its accuracy in nuclear segmentation and its prognostic ability. Its assessment of NP was compared to manual morphometry and established prognostic tests (pathologists' NP estimates (n = 11), mitotic count, histological grading, and TNM-stage). The standard deviation (SD) of the nuclear area, indicative of anisokaryosis, exhibited good discriminatory ability for tumor-specific survival, with an area under the curve (AUC) of 0.80 and a hazard ratio (HR) of 3.38. The algorithm achieved values comparable to manual morphometry. In contrast, the pathologists' estimates of anisokaryosis resulted in HR values ranging from 0.86 to 34.8, with slight inter-observer reproducibility (k = 0.204). Other conventional tests had no significant prognostic value in our study cohort. Fully automated morphometry promises a time-efficient and reproducible assessment of NP with a high prognostic value. Further refinement of the algorithm, particularly to address undersegmentation, and application to a larger study population are required.

Synthesis and Characterization of DHODH Inhibitors Based on the Vidofludimus Scaffold with Pronounced Anti-SARS-CoV-2 Activity.

New strategies for the rapid development of broad-spectrum antiviral therapies are urgently required for emerging and re-emerging viruses like the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Host-directed antivirals that target universal cellular metabolic pathways necessary for viral replication present a promising approach with broad-spectrum activity and low potential for development of viral resistance. Dihydroorotate dehydrogenase (DHODH) was identified as one of those universal host factors essential for the replication of many clinically relevant human pathogenic viruses. DHODH is the rate-limiting enzyme catalyzing the fourth step in the de novo pyrimidine synthesis. Therefore, it is also developed as a therapeutic target for many diseases relying on cellular pyrimidine resources, such as cancer, autoimmune diseases and viral or bacterial infection. Thus, several DHODH inhibitors, including vidofludimus calcium (VidoCa, IMU-838), are currently in development or have been investigated in clinical trials for the treatment of virus infections such as SARS-CoV-2-mediated coronavirus disease 19 (COVID-19). Here, we report the medicinal chemistry optimization of VidoCa that resulted in metabolically more stable derivatives with improved DHODH target inhibition in various mammalian species, which translated into improved efficacy against SARS-CoV-2.

Characterization of anthracycline-induced cardiotoxicity by diffusion tensor magnetic resonance imaging.

Anthracyclines are highly potent anti-cancer drugs, but their clinical use is limited by severe cardiotoxic side effects. The impact of anthracycline-induced cardiotoxicity (AIC) on left ventricular (LV) microarchitecture and diffusion properties remains unknown. This study sought to characterize AIC by cardiovascular magnetic resonance diffusion tensor imaging (DTI). Mice were treated with Doxorubicin (DOX; n = 16) for induction of AIC or saline as corresponding control (n = 15). Cardiac function was assessed via echocardiography at the end of the study period. Whole hearts (n = 8 per group) were scanned ex vivo by high-resolution DTI at 7 T. Results were correlated with histopathology and mass spectrometry imaging. Mice with AIC demonstrated systolic dysfunction (LVEF 52 ± 3% vs. 43 ± 6%, P < 0.001), impaired global longitudinal strain (-19.6 ± 2.0% vs. -16.6 ± 3.0%, P < 0.01), and cardiac atrophy (LV mass index [mg/mm], 4.3 ± 0.1 vs. 3.6 ± 0.2, P < 0.01). Regional sheetlet angles were significantly lower in AIC, whereas helix angle and relative helicity remained unchanged. In AIC, fractional anisotropy was increased (0.12 ± 0.01 vs. 0.14 ± 0.02, P < 0.05). DOX-treated mice displayed higher planar and less spherical anisotropy (CPlanar 0.07 ± 0.01 vs. 0.09 ± 0.01, P < 0.01; CSpherical 0.89 ± 0.01 vs. 0.87 ± 0.02, P < 0.05). CPlanar and CSpherical yielded good discriminatory power to distinguish between mice with and without AIC (c-index 0.91 and 0.84, respectively, P for both < 0.05). AIC is associated with regional changes in sheetlet angle but no major abnormalities of global LV microarchitecture. The geometric shape of the diffusion tensor is altered in AIC. DTI may provide a new tool for myocardial characterization in patients with AIC, which warrants future clinical studies to evaluate its diagnostic utility.

Domain generalization across tumor types, laboratories, and species - Insights from the 2022 edition of the Mitosis Domain Generalization Challenge.

Recognition of mitotic figures in histologic tumor specimens is highly relevant to patient outcome assessment. This task is challenging for algorithms and human experts alike, with deterioration of algorithmic performance under shifts in image representations. Considerable covariate shifts occur when assessment is performed on different tumor types, images are acquired using different digitization devices, or specimens are produced in different laboratories. This observation motivated the inception of the 2022 challenge on MItosis Domain Generalization (MIDOG 2022). The challenge provided annotated histologic tumor images from six different domains and evaluated the algorithmic approaches for mitotic figure detection provided by nine challenge participants on ten independent domains. Ground truth for mitotic figure detection was established in two ways: a three-expert majority vote and an independent, immunohistochemistry-assisted set of labels. This work represents an overview of the challenge tasks, the algorithmic strategies employed by the participants, and potential factors contributing to their success. With an F1 score of 0.764 for the top-performing team, we summarize that domain generalization across various tumor domains is possible with today's deep learning-based recognition pipelines. However, we also found that domain characteristics not present in the training set (feline as new species, spindle cell shape as new morphology and a new scanner) led to small but significant decreases in performance. When assessed against the immunohistochemistry-assisted reference standard, all methods resulted in reduced recall scores, with only minor changes in the order of participants in the ranking.

Knockdown of TPI in human dermal microvascular endothelial cells and its impact on angiogenesis in vitro.

INTRODUCTION: Angiogenic behaviour has been shown as highly versatile among Endothelial cells (ECs) causing problems of in vitro assays of angiogenesis considering their reproducibility. It is indispensable to investigate influencing factors of the angiogenic potency of ECs. OBJECTIVE: The present study aimed to analyse the impact of knocking down triosephosphate isomerase (TPI) on in vitro angiogenesis and simultaneously on vimentin (VIM) and adenosylmethionine synthetase isoform type 2 (MAT2A) expression. Furthermore, native expression profiles of TPI, VIM and MAT2A in the course of angiogenesis in vitro were examined. METHODS: Two batches of human dermal microvascular ECs were cultivated over 50 days and stimulated to undergo angiogenesis. A shRNA-mediated knockdown of TPI was performed. During cultivation, time-dependant morphological changes were detected and applied for EC-staging as prerequisite for quantifying in vitro angiogenesis. Additionally, mRNA and protein levels of all proteins were monitored. RESULTS: Opposed to native cells, knockdown cells were not able to enter late stages of angiogenesis and primarily displayed a downregulation of VIM and an uprise in MAT2A expression. Native cells increased their TPI expression and decreased their VIM expression during the course of angiogenesis in vitro. For MAT2A, highest expression was observed to be in the beginning and at the end of angiogenesis. CONCLUSION: Knocking down TPI provoked expressional changes in VIM and MAT2A and a deceleration of in vitro angiogenesis, indicating that TPI represents an angiogenic protein. Native expression profiles lead to the assumption of VIM being predominantly relevant in beginning stages, MAT2A in beginning and late stages and TPI during the whole course of angiogenesis in vitro.

Histological classification of canine and feline lymphoma using a modular approach based on deep learning and advanced image processing.

Histopathological examination of tissue samples is essential for identifying tumor malignancy and the diagnosis of different types of tumor. In the case of lymphoma classification, nuclear size of the neoplastic lymphocytes is one of the key features to differentiate the different subtypes. Based on the combination of artificial intelligence and advanced image processing, we provide a workflow for the classification of lymphoma with regards to their nuclear size (small, intermediate, and large). As the baseline for our workflow testing, we use a Unet++ model trained on histological images of canine lymphoma with individually labeled nuclei. As an alternative to the Unet++, we also used a publicly available pre-trained and unmodified instance segmentation model called Stardist to demonstrate that our modular classification workflow can be combined with different types of segmentation models if they can provide proper nuclei segmentation. Subsequent to nuclear segmentation, we optimize algorithmic parameters for accurate classification of nuclear size using a newly derived reference size and final image classification based on a pathologists-derived ground truth. Our image classification module achieves a classification accuracy of up to 92% on canine lymphoma data. Compared to the accuracy ranging from 66.67 to 84% achieved using measurements provided by three individual pathologists, our algorithm provides a higher accuracy level and reproducible results. Our workflow also demonstrates a high transferability to feline lymphoma, as shown by its accuracy of up to 84.21%, even though our workflow was not optimized for feline lymphoma images. By determining the nuclear size distribution in tumor areas, our workflow can assist pathologists in subtyping lymphoma based on the nuclei size and potentially improve reproducibility. Our proposed approach is modular and comprehensible, thus allowing adaptation for specific tasks and increasing the users' trust in computer-assisted image classification.

Cross-species oncogenomics offers insight into human muscle-invasive bladder cancer.

BACKGROUND: In humans, muscle-invasive bladder cancer (MIBC) is highly aggressive and associated with a poor prognosis. With a high mutation load and large number of altered genes, strategies to delineate key driver events are necessary. Dogs and cats develop urothelial carcinoma (UC) with histological and clinical similarities to human MIBC. Cattle that graze on bracken fern also develop UC, associated with exposure to the carcinogen ptaquiloside. These species may represent relevant animal models of spontaneous and carcinogen-induced UC that can provide insight into human MIBC. RESULTS: Whole-exome sequencing of domestic canine (n = 87) and feline (n = 23) UC, and comparative analysis with human MIBC reveals a lower mutation rate in animal cases and the absence of APOBEC mutational signatures. A convergence of driver genes (ARID1A, KDM6A, TP53, FAT1, and NRAS) is discovered, along with common focally amplified and deleted genes involved in regulation of the cell cycle and chromatin remodelling. We identify mismatch repair deficiency in a subset of canine and feline UCs with biallelic inactivation of MSH2. Bovine UC (n = 8) is distinctly different; we identify novel mutational signatures which are recapitulated in vitro in human urinary bladder UC cells treated with bracken fern extracts or purified ptaquiloside. CONCLUSION: Canine and feline urinary bladder UC represent relevant models of MIBC in humans, and cross-species analysis can identify evolutionarily conserved driver genes. We characterize mutational signatures in bovine UC associated with bracken fern and ptaquiloside exposure, a human-linked cancer exposure. Our work demonstrates the relevance of cross-species comparative analysis in understanding both human and animal UC.

A systematic safety pipeline for selection of T-cell receptors to enter clinical use.

Cancer immunotherapy using T cell receptor-engineered T cells (TCR-Ts) represents a promising treatment option. However, technologies for pre-clinical safety assessment are incomplete or inaccessible to most laboratories. Here, TCR-T off-target reactivity was assessed in five steps: (1) Mapping target amino acids necessary for TCR-T recognition, followed by (2) a computational search for, and (3) reactivity screening against, candidate cross-reactive peptides in the human proteome. Natural processing and presentation of recognized peptides was evaluated using (4) short mRNAs, and (5) full-length proteins. TCR-Ts were screened for recognition of unintended HLA alleles, and as proxy for off-target reactivity in vivo, a syngeneic, HLA-A*02:01-transgenic mouse model was used. Validation demonstrated importance of studying recognition of full-length candidate off-targets, and that the clinically applied 1G4 TCR has a hitherto unknown reactivity to unintended HLA alleles, relevant for patient selection. This widely applicable strategy should facilitate evaluation of candidate therapeutic TCRs and inform clinical decision-making.

A comprehensive multi-domain dataset for mitotic figure detection.

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.

Automated diagnosis of 7 canine skin tumors using machine learning on H&E-stained whole slide images.

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.

SFRP1 Expression is Inversely Associated With Metastasis Formation in Canine Mammary Tumours.

BACKGROUND: Canine mammary tumours (CMTs) are the most frequent tumours in intact female dogs and show strong similarities with human breast cancer. In contrast to the human disease there are no standardised diagnostic or prognostic biomarkers available to guide treatment. We recently identified a prognostic 18-gene RNA signature that could stratify human breast cancer patients into groups with significantly different risk of distant metastasis formation. Here, we assessed whether expression patterns of these RNAs were also associated with canine tumour progression. METHOD: A sequential forward feature selection process was performed on a previously published microarray dataset of 27 CMTs with and without lymph node (LN) metastases to identify RNAs with significantly differential expression to identify prognostic genes within the 18-gene signature. Using an independent set of 33 newly identified archival CMTs, we compared expression of the identified prognostic subset on RNA and protein basis using RT-qPCR and immunohistochemistry on FFPE-tissue sections. RESULTS: While the 18-gene signature as a whole did not have any prognostic power, a subset of three RNAs: Col13a1, Spock2, and Sfrp1, together completely separated CMTs with and without LN metastasis in the microarray set. However, in the new independent set assessed by RT-qPCR, only the Wnt-antagonist Sfrp1 showed significantly increased mRNA abundance in CMTs without LN metastases on its own (p = 0.013) in logistic regression analysis. This correlated with stronger SFRP1 protein staining intensity of the myoepithelium and/or stroma (p < 0.001). SFRP1 staining, as well as ß-catenin membrane staining, was significantly associated with negative LN status (p = 0.010 and 0.014 respectively). However, SFRP1 did not correlate with ß-catenin membrane staining (p = 0.14). CONCLUSION: The study identified SFRP1 as a potential biomarker for metastasis formation in CMTs, but lack of SFRP1 was not associated with reduced membrane-localisation of ß-catenin in CMTs.

Lipid nanoparticle-encapsulated, chemically modified anti-adenoviral siRNAs inhibit hepatic adenovirus infection in immunosuppressed Syrian hamsters.

RNA interference has demonstrated its potential as an antiviral therapy for treatment of human adenovirus (hAd) infections. The only existing viral vector-based system for delivery of anti-adenoviral artificial microRNAs available for in vivo use, however, has proven to be inefficient in therapeutic applications. In this study, we investigated the potential of stabilized small interfering RNA (siRNA) encapsulated in lipid nanoparticles (LNPs) for treatment of hepatic hAd serotype 5 (hAd5) infection in an hAd infection model using immunosuppressed Syrian hamsters. The siRNA sipTPmod directed against the adenoviral pre-terminal protein (pTP) and containing 2'-O-methyl modifications as well as phosphorothioate linkages effectively inhibited hAd5 infection in vitro. In light of this success, sipTPmod was encapsulated in LNPs containing the cationic lipid XL-10, which enables hepatocyte-specific siRNA transfer, and injected intravenously into hAd5-infected immunosuppressed Syrian hamsters. This resulted in a significant reduction of liver hAd5 titers, a trend toward reduced liver injury and inflammation, and reduction of viral titers in the blood and spleen compared with hAd5-infected animals that received a non-silencing siRNA. These effects were demonstrated in animals infected with low and moderate doses of hAd5. These data demonstrate that hepatic hAd5 infection can be successfully treated with anti-adenoviral sipTPmod encapsulated in LNPs.

Comparative Study of Digital Squamous Cell Carcinoma in Giant, Standard, and Miniature Schnauzers.

In schnauzers, a breed predisposition to squamous cell carcinoma of the digit (dSCC) is well known. The aim of this study was to compare the clinical and macroscopic findings of dSCCs in giant (GSs), standard (SSs), and miniature schnauzers (MSs). METHODS: Pathology reports of 478 dSCCs from 417 schnauzers (227 GSs, 174 SSs, and 16 MSs) were retrospectively evaluated. RESULTS: The MSs were older than the SSs and GSs (p ≤ 0.01). The male GSs were predisposed to dSCC (p < 0.05). In the GSs, the nodular dSCCs were larger than in the MSs (p ≤ 0.05) and SSs (p ≤ 0.001). The digital SCCs were mostly diagnosed at the forelimbs, especially at digits 1, 2, and 5. At the hindlimbs, the affected toes differed between the GSs and SSs. Multiple dSCCs were more common in SSs than in GSs (p = 0.003). If dSCC was the cause of death, the survival time was shorter than in dogs dying from other diseases (p = 0.004). Metastases occurred in 20% of the cases and led to a significantly shorter survival time in both the GSs and SSs (p < 0.001). CONCLUSIONS: The results showed various differences in the dSCC depending on the size variant of the schnauzer.

Immunohistochemical and Molecular Genetic Analysis of Canine Digital Mast Cell Tumours.

Grading, immunohistochemistry and c-kit mutation status are criteria for assessing the prognosis and therapeutic options of canine cutaneous mast cell tumours (MCTs). As a subset, canine digital MCTs have rarely been explored in this context. Therefore, in this retrospective study, 68 paraffin-embedded canine digital MCTs were analysed, and histological grading was assessed according to Patnaik and Kiupel. The immunohistochemical markers KIT and Ki67 were used, as well as polymerase chain reaction (PCR) for mutational screening in c-kit exons 8, 9, 11 and 14. Patnaik grading resulted in 22.1% grade I, 67.6% grade II and 10.3% grade III tumours. Some 86.8% of the digital MCTs were Kiupel low-grade. Aberrant KIT staining patterns II and III were found in 58.8%, and a count of more than 23 Ki67-positive cells in 52.3% of the cases. Both parameters were significantly associated with an internal tandem duplication (ITD) in c-kit exon 11 (12.7%). French Bulldogs, which tend to form well-differentiated cutaneous MCTs, had a higher proportion of digital high-grade MCTs and ITD in c-kit exon 11 compared with mongrels. Due to its retrospective nature, this study did not allow for an analysis of survival data. Nevertheless, it may contribute to the targeted characterisation of digital MCTs.

Pan-tumor T-lymphocyte detection using deep neural networks: Recommendations for transfer learning in immunohistochemistry.

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.

Evaluating the Histologic Grade of Digital Squamous Cell Carcinomas in Dogs and Copy Number Variation of KIT Ligand-A Correlation Study.

Dark-haired dogs are predisposed to the development of digital squamous cell carcinoma (DSCC). This may potentially suggest an underlying genetic predisposition not yet completely elucidated. Some authors have suggested a potential correlation between the number of copies KIT Ligand (KITLG) and the predisposition of dogs to DSCC, containing a higher number of copies in those affected by the neoplasm. In this study, the aim was to evaluate a potential correlation between the number of copies of the KITLG and the histological grade of malignancy in dogs with DSCC. For this, 72 paraffin-embedded DSCCs with paired whole blood samples of 70 different dogs were included and grouped according to their haircoat color as follow: Group 0/unknown haircoat color (n = 11); Group 1.a/black non-Schnauzers (n = 15); group 1.b/black Schnauzers (n = 33); group 1.c/black and tan dogs (n = 7); group 2/tan animals (n = 4). The DSCCs were histologically graded. Additionally, KITLG Copy Number Variation (CNV) was determined by ddPCR. A significant correlation was observed between KITLG copy number and the histological grade and score value. This finding may suggest a possible factor for the development of canine DSCC, thus potentially having an impact on personalized veterinary oncological strategies and breeding programs.

IFNγ binding to extracellular matrix prevents fatal systemic toxicity.

Interferon-γ (IFNγ) is an important mediator of cellular immune responses, but high systemic levels of this cytokine are associated with immunopathology. IFNγ binds to its receptor (IFNγR) and to extracellular matrix (ECM) via four positively charged C-terminal amino acids (KRKR), the ECM-binding domain (EBD). Across evolution, IFNγ is not well conserved, but the EBD is highly conserved, suggesting a critical function. Here, we show that IFNγ lacking the EBD (IFNγΔKRKR) does not bind to ECM but still binds to the IFNγR and retains bioactivity. Overexpression of IFNγΔKRKR in tumors reduced local ECM binding, increased systemic levels and induced sickness behavior, weight loss and toxicity. To analyze the function of the EBD during infection, we generated IFNγΔKRKR mice lacking the EBD by using CRISPR-Cas9. Infection with lymphocytic choriomeningitis virus resulted in higher systemic IFNγΔKRKR levels, enhanced sickness behavior, weight loss and fatal toxicity. We conclude that local retention of IFNγ is a pivotal mechanism to protect the organism from systemic toxicity during prolonged immune stimulation.

Mitosis domain generalization in histopathology images - The MIDOG challenge.

The density of mitotic figures (MF) within tumor tissue is known to be highly correlated with tumor proliferation and thus is an important marker in tumor grading. Recognition of MF by pathologists is subject to a strong inter-rater bias, limiting its prognostic value. State-of-the-art deep learning methods can support experts but have been observed to strongly deteriorate when applied in a different clinical environment. The variability caused by using different whole slide scanners has been identified as one decisive component in the underlying domain shift. The goal of the MICCAI MIDOG 2021 challenge was the creation of scanner-agnostic MF detection algorithms. The challenge used a training set of 200 cases, split across four scanning systems. As test set, an additional 100 cases split across four scanning systems, including two previously unseen scanners, were provided. In this paper, we evaluate and compare the approaches that were submitted to the challenge and identify methodological factors contributing to better performance. The winning algorithm yielded an F1 score of 0.748 (CI95: 0.704-0.781), exceeding the performance of six experts on the same task.

Acute cytomegalovirus infection modulates the intestinal microbiota and targets intestinal epithelial cells.

Primary and recurrent cytomegalovirus (CMV) infections frequently cause CMV colitis in immunocompromised as well as inflammatory bowel disease (IBD) patients. Additionally, colitis occasionally occurs upon primary CMV infection in patients who are apparently immunocompetent. In both cases, the underlying pathophysiologic mechanisms are largely elusive - in part due to the lack of adequate access to specimens. We employed the mouse cytomegalovirus (MCMV) model to assess the association between CMV and colitis. During acute primary MCMV infection of immunocompetent mice, the gut microbial composition was affected as manifested by an altered ratio of the Firmicutes to Bacteroidetes phyla. Interestingly, these microbial changes coincided with high-titer MCMV replication in the colon, crypt hyperplasia, increased colonic pro-inflammatory cytokine levels, and a transient increase in the expression of the antimicrobial protein Regenerating islet-derived protein 3 gamma (Reg3γ). Further analyses revealed that murine and human intestinal epithelial cell lines, as well as primary intestinal crypt cells and organoids represent direct targets of CMV infection causing increased cell death. Accordingly, in vivo MCMV infection disrupted the intestinal epithelial barrier and increased apoptosis of intestinal epithelial cells. In summary, our data show that CMV transiently induces colitis in immunocompetent hosts by altering the intestinal homeostasis.

Pan-tumor CAnine cuTaneous Cancer Histology (CATCH) dataset.

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.