- title: 'Molecular Subtype Prediction for Breast Cancer Using H&E Specialized Backbone' abstract: ' Identifying the molecular markers to categorize breast cancer is one of the key steps in determining the prognosis and treatment strategy. The standard clinical practice is to do this analysis based on multiple immunohistochemistry (IHC) stainings for each biomarker, which is expensive and inconsistent when lacking resources. In this work, we investigated the predictiveness of morphological characteristics of Hematoxylin and Eosin (H&E) stained tissues for molecular subtype analysis, as an initial step for direct treatment response prediction based on H&E whole slide images (WSI). Transfer learning using backbones pre-trained on natural images is a common practice to deal with the challenge of lack of large and precisely annotated datasets. However, using pre-training on natural images is not optimal for clinical images. To deal with this challenge and leverage large pools of unlabeled data, we propose to use a specialized backbone pre-trained on H&E WSI in a self-supervised setting, i.e. without any labels. Our experiments show that this backbone is capable of learning discriminating morphological characteristics from H&E images which are well predictive of the molecular subtypes in weakly supervised settings. Also, the network performs better in terms of generalization to unseen data from new scanner types, despite the relatively small size of the dataset used for pre-training the backbone. ' volume: 156 URL: https://proceedings.mlr.press/v156/abbasi-sureshjani21a.html PDF: https://proceedings.mlr.press/v156/abbasi-sureshjani21a/abbasi-sureshjani21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-abbasi-sureshjani21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Samaneh family: Abbasi-Sureshjani - given: Anıl family: Yüce - given: Simon family: Schönenberger - given: Maris family: Skujevskis - given: Uwe family: Schalles - given: Fabien family: Gaire - given: Konstanty family: Korski editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 1-9 id: abbasi-sureshjani21a issued: date-parts: - 2021 - 9 - 16 firstpage: 1 lastpage: 9 published: 2021-09-16 00:00:00 +0000 - title: 'A Novel Cell Map Representation for Weakly Supervised Prediction of ER & PR Status from H&E WSIs' abstract: 'Digital pathology opens new pathways for computational algorithms to play a significant role in the prognosis, diagnosis, and analysis of cancer. However, handling large whole slide images (WSIs) is a vital challenge that these algorithms encounter. In this paper, we propose a novel technique that creates a compressed representation of histology images. This representation is composed of cellular maps and compresses the WSIs while keeping relevant information at hand including the spatial relationships between cells. The compression technique is used to predict the status of ER & PR expressions from H&E WSIs. Our results show that the proposed compression technique can improve the prediction performance by 11-26%.' volume: 156 URL: https://proceedings.mlr.press/v156/alghamdia21a.html PDF: https://proceedings.mlr.press/v156/alghamdia21a/alghamdia21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-alghamdia21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Hammam M. family: AlGhamdiă - given: Navid Alemi family: Koohbanani - given: Nasir family: Rajpoot - given: Shan E. Ahmed family: Raza editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 10-19 id: alghamdia21a issued: date-parts: - 2021 - 9 - 16 firstpage: 10 lastpage: 19 published: 2021-09-16 00:00:00 +0000 - title: 'Improving Mask R-CNN for Nuclei Instance Segmentation in Hematoxylin & Eosin-Stained Histological Images' abstract: 'Digital pathology is an emerging topic in the analysis of pathologic tissue samples. It includes providing the tools towards more automated workflows to derive clinically relevant information. Digitization and storage of whole slide images have become commonplace and allow modern image analysis methods to be used. In recent years, computer-based segmentation of cell nuclei has gathered considerable attention in the development of highly specialized algorithms. Currently, most of these algorithms are based on performing semantic segmentation of all cell nuclei and separating overlapping instances in a post-processing step. Recently, instance-aware segmentation methods such as Mask R-CNN have been proposed to enable unified instance detection and segmentation, even in overlapping cases. In this work, we propose a modified Mask R-CNN-based approach by incorporating distance maps of instances and hematoxylin-stain intensities as extra input channels to the model. Moreover, we explore the impact of three well-known inference strategies, namely test-time augmentation, ensembling, and knowledge transfer through pre-training on the segmentation performance. We perform extensive ablation experiments across multiple runs to quantitatively define the most optimal inference strategy in the proposed Mask R-CNN algorithm. Our results show that average instance segmentation improvements of up to 3.5% and 4.1% based on Aggregate Jaccard Index and Panoptic Quality score can be obtained, respectively, using the proposed techniques in comparison to a standard Mask R-CNN model. Our findings confirm the effectiveness of aggregating information at the network input stage and optimizing inference workflows using minimal effort. Implemented modifications and codes are publicly available through a GitHub repository under: https://github.com/bbanc/Improved-Mask-R-CNN-for-nuclei-segmentation' volume: 156 URL: https://proceedings.mlr.press/v156/bancher21a.html PDF: https://proceedings.mlr.press/v156/bancher21a/bancher21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-bancher21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Benjamin family: Bancher - given: Amirreza family: Mahbod - given: Isabella family: Ellinger - given: Rupert family: Ecker - given: Georg family: Dorffner editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 20-35 id: bancher21a issued: date-parts: - 2021 - 9 - 16 firstpage: 20 lastpage: 35 published: 2021-09-16 00:00:00 +0000 - title: 'Random Multi-Channel Image Synthesis for Multiplexed Immunofluorescence Imaging' abstract: 'Multiplex immunofluorescence (MxIF) is an emerging imaging technique that produces the high sensitivity and specificity of single-cell mapping. With a tenet of “seeing is believing”, MxIF enables iterative staining and imaging extensive antibodies, which provides comprehensive biomarkers to segment and group different cells on a single tissue section. However, considerable depletion of the scarce tissue is inevitable from extensive rounds of staining and bleaching (“missing tissue”). Moreover, the immunofluorescence (IF) imaging can globally fail for particular rounds (“missing stain”). In this work, we focus on the “missing stain” issue. It would be appealing to develop digital image synthesis approaches to restore missing stain images without losing more tissue physically. Herein, we aim to develop image synthesis approaches for eleven MxIF structural molecular markers (i.e., epithelial and stromal) on real samples. We propose a novel multi-channel high-resolution image synthesis approach, called pixN2N-HD, to tackle possible missing stain scenarios via a high-resolution generative adversarial network (GAN). Our contribution is three-fold: (1) a single deep network framework is proposed to tackle missing stain in MxIF; (2) the proposed “N-to-N” strategy reduces theoretical four years of computational time to 20 hours when covering all possible missing stains scenarios, with up to five missing stains (e.g., “(N-1)-to-1”, “(N-2)-to-2”); and (3) this work is the first comprehensive experimental study of investigating cross-stain synthesis in MxIF. Our results elucidate a promising direction of advancing MxIF imaging with deep image synthesis. ' volume: 156 URL: https://proceedings.mlr.press/v156/bao21a.html PDF: https://proceedings.mlr.press/v156/bao21a/bao21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-bao21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Shunxing family: Bao - given: Yucheng family: Tang - given: Ho Hin family: Lee - given: Riqiang family: Gao - given: Sophie family: Chiron - given: Ilwoo family: Lyu - given: Lori A. family: Coburn - given: Keith T. family: Wilson - given: Joseph T. family: Roland - given: Bennett A. family: Landman - given: Yuankai family: Huo editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 36-46 id: bao21a issued: date-parts: - 2021 - 9 - 16 firstpage: 36 lastpage: 46 published: 2021-09-16 00:00:00 +0000 - title: 'Multi-scale Regional Attention Deeplab3+: Multiple Myeloma Plasma Cells Segmentation in Microscopic Images' abstract: 'Multiple myeloma cancer is a type of blood cancer that happens when the growth of abnormal plasma cells becomes out of control in the bone marrow. There are various ways to diagnose multiple myeloma in bone marrow, such as a complete blood count test (CBC) or counting myeloma plasma cells in aspirate slide images using manual visualization or image processing techniques. In this work, an automatic deep learning method for detecting and segmentation multiple myeloma plasma cells has been explored. To this end, a two- stage deep learning method is designed. In the first stage, the nucleus detection network is utilized to extract each instance of a cell of interest. The extracted instance is then fed to the multi-scale function to generate a multi-scale representation. The objective of the multi-scale function is to capture the shape variation and reduce the effect of object scale on the cytoplasm segmentation network. The generated scales are then fed into a pyramid of cytoplasm networks to learn the segmentation map in various scales. On top of the cytoplasm segmentation network, we included a scale aggregation function to refine and generate a final prediction. The proposed approach has been evaluated on the SegPC2021 grand challenge and ranked second on the final test phase among all teams.' volume: 156 URL: https://proceedings.mlr.press/v156/afshin21a.html PDF: https://proceedings.mlr.press/v156/afshin21a/afshin21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-afshin21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Bozorgpour family: Afshin - given: Azad family: Reza - given: Showkatian family: Eman - given: Sulaiman family: Alaa editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 47-56 id: afshin21a issued: date-parts: - 2021 - 9 - 16 firstpage: 47 lastpage: 56 published: 2021-09-16 00:00:00 +0000 - title: 'End-to-end Multiple Instance Learning for Whole-Slide Cytopathology of Urothelial Carcinoma' abstract: 'As a non-invasive approach, cytopathology of urine sediment is a highly promising approach to diagnosing urothelial carcinoma. However, computational assessment of the cytopathological status of a sample raises the challenge of identifying few cancerous cells among thousands of cells in a microscopic whole-slide image. To address this challenge, we propose an end-to-end trainable multiple instance learning approach that combines the attention mechanism and hard negative mining to classify hematoxylin and eosin stained patient-level whole-slide images of urine sediment cells. The singular cells are extracted by a simple foreground detection algorithm. With feature embeddings computed for each image patch in a bag by a convolutional neural network, the attention mechanism serves as the pooling operator, enabling a bag-level prediction while still giving an interpretable score for each image patch. This enables the identification of key instances and potential regions of interest that trigger a patient-level decision. Our results show that the proposed system can differentiate between normal and cancerous urothelial cells, thus enabling the non-invasive diagnosis of urothelial carcinoma in patients using urine sediment analysis.' volume: 156 URL: https://proceedings.mlr.press/v156/butke21a.html PDF: https://proceedings.mlr.press/v156/butke21a/butke21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-butke21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Joshua family: Butke - given: Tatjana family: Frick - given: Florian family: Roghmann - given: Samir F family: El-Mashtoly - given: Klaus family: Gerwert - given: Axel family: Mosig editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 57-68 id: butke21a issued: date-parts: - 2021 - 9 - 16 firstpage: 57 lastpage: 68 published: 2021-09-16 00:00:00 +0000 - title: 'Automatic and explainable grading of meningiomas from histopathology images' abstract: 'Meningioma is one of the most prevalent brain tumors in adults. To determine its malignancy, it is graded by a pathologist into three grades according to WHO standards. This grade plays a decisive role in treatment, and yet may be subject to inter-rater discordance. In this work, we present and compare three approaches towards fully automatic meningioma grading from histology whole slide images. All approaches are following a two-stage paradigm, where we first identify a region of interest based on the detection of mitotic figures in the slide using a state-of-the-art object detection deep learning network. This region of highest mitotic rate is considered characteristic for biological tumor behavior. In the second stage, we calculate a score corresponding to tumor malignancy based on information contained in this region using three different settings. In a first approach, image patches are sampled from this region and regression is based on morphological features encoded by a ResNet-based network. We compare this to learning a logistic regression from the determined mitotic count, an approach which is easily traceable and explainable. Lastly, we combine both approaches in a single network. We trained the pipeline on 951 slides from 341 patients and evaluated them on a separate set of 141 slides from 43 patients. All approaches yield a high correlation to the WHO grade. The logistic regression and the combined approach had the best results in our experiments, yielding correct predictions in $32$ and $33$ of all cases, respectively, with the image-based approach only predicting $25$ cases correctly. Spearman’s correlation was $0.7163$, $0.7926$ and $0.7900$ respectively. It might be counter-intuitive at first that morphological features provided by the image patches do not improve model performance. Yet, this mirrors the criteria of the grading scheme, where mitotic count is the only unequivocal parameter.' volume: 156 URL: https://proceedings.mlr.press/v156/ganz21a.html PDF: https://proceedings.mlr.press/v156/ganz21a/ganz21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-ganz21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Jonathan family: Ganz - given: Tobias family: Kirsch - given: Lucas family: Hoffmann - given: Christof A. family: Bertram - given: Christoph family: Hoffmann - given: Andreas family: Maier - given: Katharina family: Breininger - given: Ingmar family: Blümcke - given: Samir family: Jabari - given: Marc family: Aubreville editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 69-80 id: ganz21a issued: date-parts: - 2021 - 9 - 16 firstpage: 69 lastpage: 80 published: 2021-09-16 00:00:00 +0000 - title: 'Deep Learning for interpretable end-to-end survival (E-ESurv) prediction in gastrointestinal cancer histopathology' abstract: 'This paper demonstrates and validates EE-Surv, a powerful yet algorithmically simple method to predict survival directly from whole slide images which we validate in colorectal and gastric cancer, two clinically relevant and markedly different tumor types.' volume: 156 URL: https://proceedings.mlr.press/v156/ghaffari-laleh21a.html PDF: https://proceedings.mlr.press/v156/ghaffari-laleh21a/ghaffari-laleh21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-ghaffari-laleh21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Narmin family: Ghaffari Laleh - given: Amelie family: Echle - given: Hannah Sophie family: Muti - given: Katherine Jane family: Hewitt - given: Schulz family: Volkmar - given: Jakob Nikolas family: Kather editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 81-93 id: ghaffari-laleh21a issued: date-parts: - 2021 - 9 - 16 firstpage: 81 lastpage: 93 published: 2021-09-16 00:00:00 +0000 - title: 'Automated Quantification Of Blood Microvessels In Hematoxylin And Eosin Whole Slide Images' abstract: 'Tumour cells require resources to survive and proliferate. In order to be provided with a supportive micro-environment rich with resources to sustain optimal growth, tumour cells tend to reside in close proximity to a network of blood vessels. Quantification of blood microvessel density can be a useful measure to investigate the importance of resource limitation in tumours for prognostication and assigning treatment and mode of drug delivery. Currently, immunohistochemistry (IHC) with specific antibodies and the subsequent detection of its binding in the tumour tissue are used to identify microvessels. The automated quantification of blood microvessels in Hematoxylin and Eosin (H&E) stained images is not widely studied because microvessels are very complex and heterogeneous. In addition, their manual identification is tedious, time-consuming and subjective. We investigate whether the vasculature in H&E can be robustly identified in whole slide sections that would ultimately avoid the need for IHC and manual annotations. We propose an artificial intelligence model based on Generative Adversarial Networks (GAN) that, from an input H&E image, can generate a synthetic Erythroblast Transformation specific related gene (ERG) stained image, highlighting vessel structures. We also trained a spatially constrained Convolutional Neural Network (CNN) to identify single cells on ERG stained whole slide images, and found good concordance between detected cells in synthetic and real ERG. This pipeline was evaluated on 2002 image patches of size 2000x2000 pixels, sampled from 9 whole slide images. We achieved the mean $R^2$ of 0.70$\pm$0.14 in our testing set. This pipeline can pave the way to study proximity of tumour cells to blood vessels. This approach has the potential to reduce the use of IHC and tissues and enable large quantitative studies.' volume: 156 URL: https://proceedings.mlr.press/v156/hamidinekoo21a.html PDF: https://proceedings.mlr.press/v156/hamidinekoo21a/hamidinekoo21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-hamidinekoo21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Azam family: Hamidinekoo - given: Anna family: Kelsey - given: Nicholas family: Trahearn - given: Joanna family: Selfe - given: Janet family: Shipley - given: Yinyin family: Yuan editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 94-104 id: hamidinekoo21a issued: date-parts: - 2021 - 9 - 16 firstpage: 94 lastpage: 104 published: 2021-09-16 00:00:00 +0000 - title: 'Detecting genetic alterations in BRAF and NTRK as oncogenic drivers in digital pathology images: towards model generalization within and across multiple thyroid cohorts' abstract: 'In this paper, we describe the machine learning problem of identifying different types of tumors based on digital pathology images. Given a set of Hematoxylin and Eosin (H&E) stained images of thyroid tumors, we train deep learning models to detect two known molecular oncogenic drivers: \textit{BRAF} mutations and \textit{NTRK} gene fusions. We implement an attention-based multiple instance learning (MIL) classifier and we assess its generalization within and across three independent cohorts. We find that the model can detect both oncogenic drivers with the MIL approach, however the problem remains challenging: our exhaustive evaluation scenarios exemplify unknown data drifts and batch effects in digital pathology as the model performance decreases when processing images from an unseen cohort. These findings highlight the necessity of rich and diverse datasets for training and evaluation as well as methods for domain-agnostic learning.' volume: 156 URL: https://proceedings.mlr.press/v156/hohne21a.html PDF: https://proceedings.mlr.press/v156/hohne21a/hohne21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-hohne21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Johannes family: Höhne - given: Jacob prefix: de family: Zoete - given: Arndt family: A. Schmitz - given: Tricia family: Bal - given: Emmanuelle prefix: di family: Tomaso - given: Matthias family: Lenga editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 105-116 id: hohne21a issued: date-parts: - 2021 - 9 - 16 firstpage: 105 lastpage: 116 published: 2021-09-16 00:00:00 +0000 - title: 'HistoCartography: A Toolkit for Graph Analytics in Digital Pathology' abstract: 'Advances in entity-graph analysis of histopathology images have brought in a new paradigm to describe tissue composition, and learn the tissue structure-to-function relationship. Entity-graphs offer flexible and scalable representations to characterize tissue organization, while allowing the incorporation of prior pathological knowledge to further support model explainability. However, their analysis requires prerequisites for image-to-graph translation and knowledge of state-of-the-art algorithms applied to graph-structured data, which can potentially hinder their adoption. In this work, we aim to alleviate these issues by developing HistoCartography, a standardized python API with necessary preprocessing, machine learning and explainability tools to facilitate graph-analytics in computational pathology. Further, we have benchmarked the computational time and performance on multiple datasets across different imaging types and histopathology tasks to highlight the applicability of the API for building computational pathology workflows. HistoCartography is available at https://github.com/histocartography/histocartography.' volume: 156 URL: https://proceedings.mlr.press/v156/jaume21a.html PDF: https://proceedings.mlr.press/v156/jaume21a/jaume21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-jaume21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Guillaume family: Jaume - given: Pushpak family: Pati - given: Valentin family: Anklin - given: Antonio family: Foncubierta - given: Maria family: Gabrani editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 117-128 id: jaume21a issued: date-parts: - 2021 - 9 - 16 firstpage: 117 lastpage: 128 published: 2021-09-16 00:00:00 +0000 - title: 'SparseConvMIL: Sparse Convolutional Context-Aware Multiple Instance Learning for Whole Slide Image Classification' abstract: 'Multiple instance learning (MIL) is the preferred approach for whole slide image classification. However, most MIL approaches do not exploit the interdependencies of tiles extracted from a whole slide image, which could provide valuable cues for classification. This paper presents a novel MIL approach that exploits the spatial relationship of tiles for classifying whole slide images. To do so, a sparse map is built from tiles embeddings, and is then classified by a sparse-input CNN. It obtained state-of-the-art performance over popular MIL approaches on the classification of cancer subtype involving 10,000 whole slide images. Our results suggest that the proposed approach might (i) improve the representation learning of instances and (ii) exploit the context of instance embeddings to enhance the classification performance. The code of this work is made open-source.' volume: 156 URL: https://proceedings.mlr.press/v156/lerousseau21a.html PDF: https://proceedings.mlr.press/v156/lerousseau21a/lerousseau21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-lerousseau21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Marvin family: Lerousseau - given: Maria family: Vakalopoulou - given: Eric family: Deutsch - given: Nikos family: Paragios editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 129-139 id: lerousseau21a issued: date-parts: - 2021 - 9 - 16 firstpage: 129 lastpage: 139 published: 2021-09-16 00:00:00 +0000 - title: 'Magnetic Resonance Imaging Virtual Histopathology from Weakly Paired Data' abstract: 'The pathological analysis of biopsy specimens is essential to cancer diagnosis, treatment selection and prognosis. However, biopsies are only taken from part of the tumor and cannot assess the full cellular extension. Such information is essential to delineate as accurately as possible the tumor volume on a three-dimensional basis. Furthermore, they require highly qualified personnel and are associated with significant risks. The aim of our work is to provide alternative means to gather clinical information related to histology through MR image translation towards virtual pathological content generation. Conventional approaches to address this objective exploit paired data that is cumbersome to achieve due to tissue collapse and deformation, different resolution scales and absence of plane correspondences. In this paper, we introduce a versatile, scalable and robust closed-loop dual synthesis concept that composes two generation mechanisms - cycle-consistent generative adversarial networks -, one exploring weakly paired data and a subsequent harnessing virtually generated paired correspondences. The clinical relevance and interest of our framework are demonstrated in prostate cancer patients. Qualitative clinical assessment and quantitative reconstruction measurements demonstrate the potential of our approach.' volume: 156 URL: https://proceedings.mlr.press/v156/leroy21a.html PDF: https://proceedings.mlr.press/v156/leroy21a/leroy21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-leroy21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Amaury family: Leroy - given: Kumar family: Shreshtha - given: Marvin family: Lerousseau - given: Théophraste family: Henry - given: Théo family: Estienne - given: Marion family: Classe - given: Nikos family: Paragios - given: Vincent family: Grégoire - given: Eric family: Deutsch editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 140-150 id: leroy21a issued: date-parts: - 2021 - 9 - 16 firstpage: 140 lastpage: 150 published: 2021-09-16 00:00:00 +0000 - title: 'Unsupervised Domain Adaptation for the Histopathological Cell Segmentation through Self-Ensembling' abstract: 'Histopathological images are generally considered as the golden standard for clinical diagnosis and cancer grading. Accurate segmentation of cells/nuclei from histopathological images is a critical step to obtain reliable morphological information for quantitative analysis. However, cell/nuclei segmentation relies heavily on well-annotated datasets, which are extremely labor-intensive, time-consuming, and expensive in practical applications. Meanwhile, one might want to fine-tune pretrained models on certain target datasets. But it is always difficult to collect enough target training images for proper fine-tuning. Therefore, there is a need for methods that can transfer learned information from one domain to another without additional target annotations. In this paper, we propose a novel framework for cell segmentation on the unlabeled images through the unsupervised domain adaptation with self-ensembling. It is achieved by applying generative adversarial networks (GANs) for the unsupervised domain adaptation of cell segmentation crossing different tissues. Images in the source and target domain can be differentiated through the learned discriminator. Meanwhile, we present a self-ensembling model to consider the source and the target domain together as a semi-supervised segmentation task to reduce the differences of outputs. Additionally, we introduce conditional random field (CRF) as post-processing to preserve the local consistency on the outputs. We validate our framework with unsupervised domain adaptation on three public cell segmentation datasets captured from different types of tissues, which achieved superior performance in comparison with state-of-the-art.' volume: 156 URL: https://proceedings.mlr.press/v156/li21a.html PDF: https://proceedings.mlr.press/v156/li21a/li21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-li21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Chaoqun family: Li - given: Yitian family: Zhou - given: Tangqi family: Shi - given: Yenan family: Wu - given: Meng family: Yang - given: Zhongyu family: Li editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 151-158 id: li21a issued: date-parts: - 2021 - 9 - 16 firstpage: 151 lastpage: 158 published: 2021-09-16 00:00:00 +0000 - title: 'SMILE: Sparse-Attention based Multiple Instance Contrastive Learning for Glioma Sub-Type Classification Using Pathological Images' abstract: 'Gliomas are the most prevalent malignant brain tumor in adults and can be classified into four typical sub-types based on histological features. Histological diagnosis by pathologists via microscopic visual inspection of pathological slides has been the gold standard for glioma grading, especially hematoxylin and eosin (H&E) sections. However, due to spatial heterogeneity and complex tumor micro-environment, it is difficult and time-consuming for pathologists to differentiate glioma sub-types. In this paper, we propose a Sparse-attention based Multiple Instance contrastive LEarning (SMILE) method for glioma sub-type classification. First, we use contrastive learning to extract meaningful representations from pathological images. Second, we propose the sparse-attention multiple instance learning aggregator to get sparse instance representations in a bag for label prediction. We validate the proposed SMILE method using a glioma dataset from The Cancer Genome Atlas (TCGA). Experimental results show superior performance of our method over competing ones. Ablation study further demonstrates the effectiveness of our design of SMILE.' volume: 156 URL: https://proceedings.mlr.press/v156/lu21a.html PDF: https://proceedings.mlr.press/v156/lu21a/lu21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-lu21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Mengkang family: Lu - given: Yongsheng family: Pan - given: Dong family: Nie - given: Feihong family: Liu - given: Feng family: Shi - given: Yong family: Xia - given: Dinggang family: Shen editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 159-169 id: lu21a issued: date-parts: - 2021 - 9 - 16 firstpage: 159 lastpage: 169 published: 2021-09-16 00:00:00 +0000 - title: 'Multi-Scale Task Multiple Instance Learning for the Classification of Digital Pathology Images with Global Annotations' abstract: 'Whole slide images (WSIs) are high-resolution digitized images of tissue samples, stored including different magnification levels. WSIs datasets often include only global annotations, available thanks to pathology reports. Global annotations refer to global findings in the high-resolution image and do not include information about the location of the regions of interest or the magnification levels used to identify a finding. This fact can limit the training of machine learning models, as WSIs are usually very large and each magnification level includes different information about the tissue. This paper presents a Multi-Scale Task Multiple Instance Learning (MuSTMIL) method, allowing to better exploit data paired with global labels and to combine contextual and detailed information identified at several magnification levels. The method is based on a multiple instance learning framework and on a multi-task network, that combines features from several magnification levels and produces multiple predictions (a global one and one for each magnification level involved). MuSTMIL is evaluated on colon cancer images, on binary and multilabel classification. MuSTMIL shows an improvement in performance in comparison to both single scale and another multi-scale multiple instance learning algorithm, demonstrating that MuSTMIL can help to better deal with global labels targeting full and multi-scale images.' volume: 156 URL: https://proceedings.mlr.press/v156/marini21a.html PDF: https://proceedings.mlr.press/v156/marini21a/marini21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-marini21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Niccolò family: Marini - given: Sebastian family: Otálora - given: Francesco family: Ciompi - given: Gianmaria family: Silvello - given: Stefano family: Marchesin - given: Simona family: Vatrano - given: Genziana family: Buttafuoco - given: Manfredo family: Atzori - given: Henning family: Müller editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 170-181 id: marini21a issued: date-parts: - 2021 - 9 - 16 firstpage: 170 lastpage: 181 published: 2021-09-16 00:00:00 +0000 - title: 'Robust Quad-Tree based Registration on Whole Slide Images' abstract: 'The registration of whole slide images (WSIs) provides the basis for many subsequent processing steps in digital pathology. For instance, the registration of immunohistochemistry (IHC) and hematoxylin & eosin (H&E)-stained WSIs is usually the first step in guiding IHC diagnostic procedures. Still, many registration methods operate poorly on WSIs. Reasons for this include the WSI size, fluctuating image quality or elastic tissue deformations. Multiple prior methods are further specialised towards a specific image modality, such as histology or cytology, or rely on a specific preparation protocol. To minimise these effects, we developed a robust WSI registration, which differs from previous methods by the following new aspect: We introduce a multi-scale approach based on a quad-tree (QT), with several termination criteria that makes the algorithm particularly insensitive to tissue artefacts and that further allows to estimate a piece-wise affine transformation. We validated our method on five scanner systems and 60 WSIs with different stainings. Our results outperformed any publicly available WSI registration method. The QT code, WSI landmarks and tools used to create the validation dataset are made publicly available. ' volume: 156 URL: https://proceedings.mlr.press/v156/marzahl21a.html PDF: https://proceedings.mlr.press/v156/marzahl21a/marzahl21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-marzahl21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Christian family: Marzahl - given: Frauke family: Wilm - given: Dressler Franz family: F. - given: Lars family: Tharun - given: Sven family: Perner - given: Christof A. family: Bertram - given: Christine family: Kröger - given: Jörn family: Voigt - given: Robert family: Klopfleisch - given: Andreas family: Maier - given: Marc family: Aubreville - given: Katharina family: Breininger editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 181-190 id: marzahl21a issued: date-parts: - 2021 - 9 - 16 firstpage: 181 lastpage: 190 published: 2021-09-16 00:00:00 +0000 - title: 'Self-supervised learning improves dMMR/MSI detection from histology slides across multiple cancers' abstract: 'Microsatellite instability (MSI) is a tumor phenotype whose diagnosis largely impacts patient care in colorectal cancers (CRC), and is associated with response to immunotherapy in all solid tumors. Deep learning models detecting MSI tumors directly from H&E stained slides have shown promise in improving diagnosis of MSI patients. Prior deep learning models for MSI detection have relied on neural networks pretrained on ImageNet dataset, which does not contain any medical image. In this study, we leverage recent advances in self-supervised learning by training neural networks on histology images from the TCGA dataset using MoCo V2. We show that these networks consistently outperform their counterparts pretrained using ImageNet and obtain state-of-the-art results for MSI detection with AUCs of 0.92 and 0.83 for CRC and gastric tumors, respectively. These models generalize well on an external CRC cohort (0.97 AUC on PAIP) and improve transfer from one organ to another. Finally we show that predictive image regions exhibit meaningful histological patterns, and that the use of MoCo features highlighted more relevant patterns according to an expert pathologist.' volume: 156 URL: https://proceedings.mlr.press/v156/saillard21a.html PDF: https://proceedings.mlr.press/v156/saillard21a/saillard21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-saillard21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Charlie family: Saillard - given: Olivier family: Dehaene - given: Tanguy family: Marchand - given: Olivier family: Moindrot - given: Aurélie family: Kamoun - given: Benoit family: Schmauch - given: Simon family: Jegou editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 191-205 id: saillard21a issued: date-parts: - 2021 - 9 - 16 firstpage: 191 lastpage: 205 published: 2021-09-16 00:00:00 +0000 - title: 'Creating small but meaningful representations of digital pathology images' abstract: 'Representation learning is a popular application of deep learning where an object (e.g., an image) is converted into a lower-dimensional representation that still encodes relevant features of the original object. In digital pathology, however, this is more difficult because whole slide images (WSIs) are tiled before processing because they are too large to process at once. As a result, one WSI can be represented by thousands of representations - one for each tile. Common strategies to aggregate the "tile-level representations" to a "slide-level representation" rely on pooling operators or even attention networks, which all find some weighted average of the tile-level representations. In this work, we propose a novel approach to aggregate tile-level representations into a single slide-level representation. Our method is based on clustering representations from individual tiles that originate from a large pool of WSIs. Each cluster can be seen as encoding a specific feature that might occur in a tile. Then, the final slide-level representation is a function of the proportional cluster membership of all tiles from one WSI. We demonstrate that we can represent WSIs in parsimonious representations and that these aggregated slide-level representations allow for both WSI classification and, reversely, similar image search. ' volume: 156 URL: https://proceedings.mlr.press/v156/guerendel21a.html PDF: https://proceedings.mlr.press/v156/guerendel21a/guerendel21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-guerendel21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Corentin family: Gueréndel - given: Phil family: Arnold - given: Ben family: Torben-Nielsen editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 206-215 id: guerendel21a issued: date-parts: - 2021 - 9 - 16 firstpage: 206 lastpage: 215 published: 2021-09-16 00:00:00 +0000 - title: 'Attention-based Multiple Instance Learning with Mixed Supervision on the Camelyon16 Dataset' abstract: 'Since the standardization of Whole Slide Images (WSIs) digitization, the use of deep learning methods for the analysis of histological images has shown much potential. However, the sheer size of WSIs is a real challenge, as they are often up to 100,000 pixels wide and high at the highest resolution, and therefore cannot be processed directly by any model. Moreover, as the manual delineation of structures within WSIs is tedious, histological datasets often only contain slide-level labels, or a limited amount of delineated slides. In this context, multiple-instance learning (MIL) approaches have been proposed, considering WSIs as bags of smaller images, designated as tiles or patches. Among these methods, the attention-based MIL aims at learning the importance of each tile for the slide final classification while at the same time performing a clustering of those tiles. In this paper, we introduce the concept of mixed supervision within this framework, by exploiting tile-level labels in addition to slide-level labels to improve the classification of slides. More precisely, we show on the Camelyon16 dataset that even a small proportion of slides with pixel-wise annotations can improve their classification but also the localization of tumorous regions. This improves the consistency of the results between the tile and slide levels and the interpretability of the algorithm.' volume: 156 URL: https://proceedings.mlr.press/v156/tourniaire21a.html PDF: https://proceedings.mlr.press/v156/tourniaire21a/tourniaire21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-tourniaire21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Paul family: Tourniaire - given: Marius family: Ilie - given: Paul family: Hofman - given: Nicholas family: Ayache - given: Herve family: Delingette editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 216-226 id: tourniaire21a issued: date-parts: - 2021 - 9 - 16 firstpage: 216 lastpage: 226 published: 2021-09-16 00:00:00 +0000 - title: 'A Multi-scale Graph Network with Multi-head Attention for Histopathology Image Diagnosisn' abstract: ' Hematoxylin-eosin (H&E) staining plays an essential role in brain glioma diagnosis, but reading pathologic images and generating diagnostic reports can be a tedious and laborious work. Pathologists need to combine and navigate extremely large images with different scales and to quantify different aspects for subtyping. In this work, we propose an automatic diagnosis algorithm to identify cell types and severity of H&E slides, in order to classify five major subtypes of glioma from whole slide pathological images. The proposed method is featured by a pyramid graph structure and an attention-based multi-instance learning strategy. We claim that our method not only improve the classification accuracy by utilizing multi-scale information, but also help to identify high risk patches. We summarized patches from multiple resolutions into a graph structure. The nodes of the pyramid graph are feature vectors extracted from image patches, and these vectors are connected by their spatial adjacency. We then fed the graph into the proposed model with self-attention and graph convolutions. Here, we used a multi-head self-attention architecture, where same self-attention blocks are stacked in parallel. As proven in Transformer networks, multiple attention maps herein capture comprehensive activation patterns from different subspace representation. Using the proposed method, the results show a 70% accuracy for glioma subtyping. The multiresolution attention maps generated from the proposed method could help locate proliferations and necrosis in the whole pathologic slide.' volume: 156 URL: https://proceedings.mlr.press/v156/xing21a.html PDF: https://proceedings.mlr.press/v156/xing21a/xing21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-xing21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Xiaodan family: Xing - given: Yixin family: Ma - given: Lei family: Jin - given: Tianyang family: Sun - given: Zhong family: Xue - given: Feng family: Shi - given: Jinsong family: Wu - given: Dinggang family: Shen editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 227-235 id: xing21a issued: date-parts: - 2021 - 9 - 16 firstpage: 227 lastpage: 235 published: 2021-09-16 00:00:00 +0000 - title: 'An Automatic Nuclei Image Segmentation Based on Multi-Scale Split-Attention U-Net' abstract: 'Nuclei segmentation is an important step in the task of medical image analysis. Nowadays, deep learning techniques based on Convolutional Neural Networks (CNNs) have become prevalent methods in nuclei segmentation. In this paper, we propose a network called Multi-scale Split-Attention U-Net (MSAU-Net) for further improving the performance of cell segmentation. MSAU-Net is based on U-Net architecture and the original blocks used to down-sampling and up-sampling paths are replaced with Multi-scale Split-Attention blocks for capturing independent semantic information of nuclei images. A public microscopy image dataset from 2018 Data Science Bowl grand challenge is selected to train and evaluate MSAU-Net. By running trained models on the test set, our model reaches average Intersection over Union (IoU) of 0.851, which is better than other prominent models, especially 4.8 percent higher than the original U-Net. For other evaluation metrics including accuracy, precision, recall and F1-score, MSAU-Net shows better performance in the most of indicators. The outstanding result reveals that our proposed model presents a promising nuclei segmentation method for the microscopy image analysis.' volume: 156 URL: https://proceedings.mlr.press/v156/xu21a.html PDF: https://proceedings.mlr.press/v156/xu21a/xu21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-xu21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Qing family: Xu - given: Wenting family: Duan editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 236-245 id: xu21a issued: date-parts: - 2021 - 9 - 16 firstpage: 236 lastpage: 245 published: 2021-09-16 00:00:00 +0000 - title: 'Symmetric Dense Inception Network for Simultaneous Cell Detection and Classification in Multiplex Immunohistochemistry Images' abstract: 'Deep-learning based automatic analysis of the multiplex immunohistochemistry (mIHC) enables distinct cell populations to be localized on a large scale, providing insights into disease biology and therapeutic targets. However, standard deep-learning pipelines performed cell detection and classification as two-stage tasks, which is computationally inefficient and faces challenges to incorporate neighbouring tissue context for determining the cell identity. To overcome these limitations and to obtain a more accurate mapping of cell phenotypes, we presented a symmetric dense inception neural network for detecting and classifying cells in mIHC slides simultaneously. The model was applied with a novel stop-gradient strategy and a loss function accounted for class imbalance. When evaluated on an ovarian cancer dataset containing 6 cell types, the model achieved an F1 score of 0.835 in cell detection, and a weighted F1-score of 0.867 in cell classification, which outperformed separate models trained on individual tasks by 1.9% and 3.8% respectively. Taken together, the proposed method boosts the learning efficiency and prediction accuracy of cell detection and classification by simultaneously learning from both tasks.' volume: 156 URL: https://proceedings.mlr.press/v156/zhang21a.html PDF: https://proceedings.mlr.press/v156/zhang21a/zhang21a.pdf edit: https://github.com/mlresearch//v156/edit/gh-pages/_posts/2021-09-16-zhang21a.md series: 'Proceedings of Machine Learning Research' container-title: 'Proceedings of the MICCAI Workshop on Computational Pathology' publisher: 'PMLR' author: - given: Hanyun family: Zhang - given: Tami family: Grunewald - given: Ayse U. family: Akarca - given: Jonathan A. family: Ledermann - given: Teresa family: Marafioti - given: Yinyin family: Yuan editor: - given: Manfredo family: Atzori - given: Nikolay family: Burlutskiy - given: Francesco family: Ciompi - given: Zhang family: Li - given: Fayyaz family: Minhas - given: Henning family: Müller - given: Tingying family: Peng - given: Nasir family: Rajpoot - given: Ben family: Torben-Nielsen - given: Jeroen prefix: van der family: Laak - given: Mitko family: Veta - given: Yinyin family: Yuan - given: Inti family: Zlobec page: 246-257 id: zhang21a issued: date-parts: - 2021 - 9 - 16 firstpage: 246 lastpage: 257 published: 2021-09-16 00:00:00 +0000