Bias reduction using combined stain normalization and augmentation for AI-based classification of histological images.

Artificial intelligence (AI)-assisted diagnosis is an ongoing revolution in pathology. However, a frequent drawback of AI models is their propension to make decisions based rather on bias in training dataset than on concrete biological features, thus weakening pathologists' trust in these tools. Technically, it is well known that microscopic images are altered by tissue processing and staining procedures, being one of the main sources of bias in machine learning for digital pathology. So as to deal with it, many teams have written about color normalization and augmentation methods. However, only a few of them have monitored their effects on bias reduction and model generalizability. In our study, two methods for stain augmentation (AugmentHE) and fast normalization (HEnorm) have been created and their effect on bias reduction has been monitored. Actually, they have also been compared to previously described strategies. To that end, a multicenter dataset created for breast cancer histological grading has been used. Thanks to it, classification models have been trained in a single center before assessing its performance in other centers images. This setting led to extensively monitor bias reduction while providing accurate insight of both augmentation and normalization methods. AugmentHE provided an 81% increase in color dispersion compared to geometric augmentations only. In addition, every classification model that involved AugmentHE presented a significant increase in the area under receiving operator characteristic curve (AUC) over the widely used RGB shift. More precisely, AugmentHE-based models showed at least 0.14 AUC increase over RGB shift-based models. Regarding normalization, HEnorm appeared to be up to 78x faster than conventional methods. It also provided satisfying results in terms of bias reduction. Altogether, our pipeline composed of AugmentHE and HEnorm improved AUC on biased data by up to 21.7% compared to usual augmentations. Conventional normalization methods coupled with AugmentHE yielded similar results while being much slower. In conclusion, we have validated an open-source tool that can be used in any deep learning-based digital pathology project on H&E whole slide images (WSI) that efficiently reduces stain-induced bias and later on might help increase pathologists' confidence when using AI-based products.

Evolutionary design of explainable algorithms for biomedical image segmentation.

An unresolved issue in contemporary biomedicine is the overwhelming number and diversity of complex images that require annotation, analysis and interpretation. Recent advances in Deep Learning have revolutionized the field of computer vision, creating algorithms that compete with human experts in image segmentation tasks. However, these frameworks require large human-annotated datasets for training and the resulting "black box" models are difficult to interpret. In this study, we introduce Kartezio, a modular Cartesian Genetic Programming-based computational strategy that generates fully transparent and easily interpretable image processing pipelines by iteratively assembling and parameterizing computer vision functions. The pipelines thus generated exhibit comparable precision to state-of-the-art Deep Learning approaches on instance segmentation tasks, while requiring drastically smaller training datasets. This Few-Shot Learning method confers tremendous flexibility, speed, and functionality to this approach. We then deploy Kartezio to solve a series of semantic and instance segmentation problems, and demonstrate its utility across diverse images ranging from multiplexed tissue histopathology images to high resolution microscopy images. While the flexibility, robustness and practical utility of Kartezio make this fully explicable evolutionary designer a potential game-changer in the field of biomedical image processing, Kartezio remains complementary and potentially auxiliary to mainstream Deep Learning approaches.

Multigenerational Effects of a Complex Human-Relevant Exposure during Folliculogenesis and Preimplantation Embryo Development: The FEDEXPO Study.

Animal toxicological studies often fail to mimic the complexity of the human exposome, associating low doses, combined molecules and long-term exposure. Since the reproductive potential of a woman begins in the fetal ovary, the literature regarding the disruption of its reproductive health by environmental toxicants remains limited. Studies draw attention to follicle development, a major determinant for the quality of the oocyte, and the preimplantation embryo, as both of them are targets for epigenetic reprogramming. The "Folliculogenesis and Embryo Development EXPOsure to a mixture of toxicants: evaluation in the rabbit model" (FEDEXPO) project emerged from consideration of these limitations and aims to evaluate in the rabbit model the impacts of an exposure to a mixture of known and suspected endocrine disrupting chemicals (EDCs) during two specific windows, including folliculogenesis and preimplantation embryo development. The mixture combines eight environmental toxicants, namely perfluorooctanesulfonic acid (PFOS), perfluorooctanoic acid (PFOA), dichlorodiphenyldichloroethylene (DDE), hexachlorobenzene (HCB), ß-hexachlorocyclohexane (ß-HCH), 2,2'4,4'-tetrabromodiphenyl ether (BDE-47), di(2-ethylhexyl) phthalate (DEHP) and bisphenol S (BPS), at relevant exposure levels for reproductive-aged women based on biomonitoring data. The project will be organized in order to assess the consequences of this exposure on the ovarian function of the directly exposed F0 females and monitor the development and health of the F1 offspring from the preimplantation stage. Emphasis will be made on the reproductive health of the offspring. Lastly, this multigenerational study will also tackle potential mechanisms for the inheritance of health disruption via the oocyte or the preimplantation embryo.

Pre- and Postnatal Dietary Exposure to a Pesticide Cocktail Disrupts Ovarian Functions in 8-Week-Old Female Mice.

Female infertility has a multifactorial origin, and exposure to contaminants, including pesticides, with endocrine-disrupting properties is considered to be involved in this reproductive disorder, especially when it occurs during early life. Pesticides are present in various facets of the environment, and consumers are exposed to a combination of multiple pesticide residues through food intake. The consequences of such exposure with respect to female fertility are not well known. Therefore, we aimed to assess the impact of pre- and postnatal dietary exposure to a pesticide mixture on folliculogenesis, a crucial process in female reproduction. Mice were exposed to the acceptable daily intake levels of six pesticides in a mixture (boscalid, captan, chlorpyrifos, thiacloprid, thiophanate and ziram) from foetal development until 8 weeks old. Female offspring presented with decreased body weight at weaning, which was maintained at 8 weeks old. This was accompanied by an abnormal ovarian ultrastructure, a drastic decrease in the number of corpora lutea and progesterone levels and an increase in ovary cell proliferation. In conclusion, this study shows that this pesticide mixture that can be commonly found in fruits in Europe, causing endocrine disruption in female mice with pre- and postnatal exposure by disturbing folliculogenesis, mainly in the luteinisation process.

Ultrarapid lytic granule release from CTLs activates Ca2+-dependent synaptic resistance pathways in melanoma cells.

Human cytotoxic T lymphocytes (CTLs) exhibit ultrarapid lytic granule secretion, but whether melanoma cells mobilize defense mechanisms with commensurate rapidity remains unknown. We used single-cell time-lapse microscopy to offer high spatiotemporal resolution analyses of subcellular events in melanoma cells upon CTL attack. Target cell perforation initiated an intracellular Ca2+ wave that propagated outward from the synapse within milliseconds and triggered lysosomal mobilization to the synapse, facilitating membrane repair and conferring resistance to CTL induced cytotoxicity. Inhibition of Ca2+ flux and silencing of synaptotagmin VII limited synaptic lysosomal exposure and enhanced cytotoxicity. Multiplexed immunohistochemistry of patient melanoma nodules combined with automated image analysis showed that melanoma cells facing CD8+ CTLs in the tumor periphery or peritumoral area exhibited significant lysosomal enrichment. Our results identified synaptic Ca2+ entry as the definitive trigger for lysosomal deployment to the synapse upon CTL attack and highlighted an unpredicted defensive topology of lysosome distribution in melanoma nodules.

Gene alterations in epigenetic modifiers and JAK-STAT signaling are frequent in breast implant-associated ALCL.

The oncogenic events involved in breast implant-associated anaplastic large cell lymphoma (BI-ALCL) remain elusive. To clarify this point, we have characterized the genomic landscape of 34 BI-ALCLs (15 tumor and 19 in situ subtypes) collected from 54 BI-ALCL patients diagnosed through the French Lymphopath network. Whole-exome sequencing (n = 22, with paired tumor/germline DNA) and/or targeted deep sequencing (n = 24) showed recurrent mutations of epigenetic modifiers in 74% of cases, involving notably KMT2C (26%), KMT2D (9%), CHD2 (15%), and CREBBP (15%). KMT2D and KMT2C mutations correlated with a loss of H3K4 mono- and trimethylation by immunohistochemistry. Twenty cases (59%) showed mutations in ≥1 member of the JAK/STAT pathway, including STAT3 (38%), JAK1 (18%), and STAT5B (3%), and in negative regulators, including SOCS3 (6%), SOCS1 (3%), and PTPN1 (3%). These mutations were more frequent in tumor-type samples than in situ samples (P = .038). All BI-ALCLs expressed pSTAT3, regardless of the mutational status of genes in the JAK/STAT pathway. Mutations in the EOMES gene (12%) involved in lymphocyte development, PI3K-AKT/mTOR (6%), and loss-of-function mutations in TP53 (12%) were also identified. Copy-number aberration (CNA) analysis identified recurrent alterations, including gains on chromosomes 2, 9p, 12p, and 21 and losses on 4q, 8p, 15, 16, and 20. Regions of CNA encompassed genes involved in the JAK/STAT pathway and epigenetic regulators. Our results show that the BI-ALCL genomic landscape is characterized by not only JAK/STAT activating mutations but also loss-of-function alterations of epigenetic modifiers.

["Discovering Pathology", a serious game dedicated to the discovery of pathology for medical students]. / « Discovering Pathology ¼, un serious game dédié à la découverte de l'anatomopathologie pour les étudiants en médecine.

The serious game is a digital concept whose intention is to combine serious aspects with the playful springs of video games. Educational, learning and communication tool's, their production has been growing steadily since the 2000s. France has become the world's second largest producer of serious games, behind the United States of America. Gradually essential in health care, they invite themselves to universities to support medical and paramedical education. We aim to create a serious game designed to introduce anatomy and pathological cytology to medical students. The project is taking place in the University of Franche-Comté and the University Hospital of Besançon. The themes addressed refer to the program of French Pathologists College's. The game structure's makes to follow the progress of a sample within a laboratory and relies on the combined use of macroscopic images and digitized slides to build a diagnosis. By using computer support for video games, this type of teaching tool aims to challenge students and increase their motivation. This non-profit pilot game will be accessible to students of the University of Franche-Comté, on the internet, in January 2019. Developed in French and English, it will then be made available to other universities wishing to use this type of educational tool.

Impaired functional responses in follicular lymphoma CD8+TIM-3+ T lymphocytes following TCR engagement.

Upregulation of T cell immunoglobulin-3 (TIM-3) has been associated with negative regulation of the immune response in chronic infection and cancer, including lymphoma. Here, we investigated the possible correlation between TIM-3 expression by ex vivo cytotoxic T cells (CTL) from follicular lymphoma (FL) biopsies and their functional unresponsiveness that could limit the favorable impact of CTL on disease progression. We report a high percentage of CD8+TIM-3+T cells in lymph nodes of FL patients. When compared to their CD8+TIM-3- counterparts, CD8+TIM-3+ T cells exhibited defective cytokine production following TCR engagement. Furthermore, CD8+TIM-3+ T cells display ex vivo markers of lytic granule release and remain unresponsive to further TCR-induced activation of the lytic machinery. Although confocal microscopy showed that TIM-3 expression on CD8+ T cells correlated with minor alterations of immunological synapse, a selective reduction of ERK signaling in CD8+TIM-3+T cells was observed by phospho-flow analysis. Finally, short relapse-free survival despite rituximab(R)-chemotherapy was observed in patients with high content of TIM-3+ cells and a poor infiltrate of granzyme B+ T cells in FL lymph nodes. Together, our data indicate that, besides selective TCR early signaling defects, TIM-3 expression correlates with unresponsiveness of ex vivo CD8+ T cells in FL. They show that scores based on the combination of exhaustion and cytolytic markers in FL microenvironment might be instrumental to identify patients at early risk of relapses following R-chemotherapy.

Whole-slide imaging is a robust alternative to traditional fluorescent microscopy for fluorescence in situ hybridization imaging using break-apart DNA probes.

Fluorescence in situ hybridization is an indispensable technique used in routine pathology and for theranostic purposes. Because fluorescence in situ hybridization techniques require sophisticated microscopic workstations and long procedures of image acquisition with sometimes subjective and poorly reproducible results, we decided to test a whole-slide imaging system as an alternative approach. In this study, we used the latest generation of Pannoramic 250 Flash digital microscopes (P250 Flash digital microscopes; 3DHISTECH, Budapest, Hungary) to digitize fluorescence in situ hybridization slides of diffuse large B cells lymphoma cases for detecting MYC rearrangement. The P250 Flash digital microscope was found to be precise with better definition of split signals in cells containing MYC rearrangement with fewer truncated signals as compared to traditional fluorescence microscopy. This digital technique is easier thanks to the preview function, which allows almost immediate identification of the tumor area, and the panning and zooming functionalities as well as a shorter acquisition time. Moreover, fluorescence in situ hybridization analyses using the digital technique appeared to be more reproducible between pathologists. Finally, the digital technique also allowed prolonged conservation of photos. In conclusion, whole-slide imaging technologies represent rapid, robust, and highly sensitive methods for interpreting fluorescence in situ hybridization slides with break-apart probes. In addition, these techniques offer an easier way to interpret the signals and allow definitive storage of the images for pathology expert networks or e-learning databases.