NuCLS: A scalable crowdsourcing, deep learning approach and dataset for nucleus classification, localization and segmentation
Authors:
Mohamed Amgad,
Lamees A. Atteya,
Hagar Hussein,
Kareem Hosny Mohammed,
Ehab Hafiz,
Maha A. T. Elsebaie,
Ahmed M. Alhusseiny,
Mohamed Atef AlMoslemany,
Abdelmagid M. Elmatboly,
Philip A. Pappalardo,
Rokia Adel Sakr,
Pooya Mobadersany,
Ahmad Rachid,
Anas M. Saad,
Ahmad M. Alkashash,
Inas A. Ruhban,
Anas Alrefai,
Nada M. Elgazar,
Ali Abdulkarim,
Abo-Alela Farag,
Amira Etman,
Ahmed G. Elsaeed,
Yahya Alagha,
Yomna A. Amer,
Ahmed M. Raslan
, et al. (12 additional authors not shown)
Abstract:
High-resolution mapping of cells and tissue structures provides a foundation for developing interpretable machine-learning models for computational pathology. Deep learning algorithms can provide accurate mappings given large numbers of labeled instances for training and validation. Generating adequate volume of quality labels has emerged as a critical barrier in computational pathology given the…
▽ More
High-resolution mapping of cells and tissue structures provides a foundation for developing interpretable machine-learning models for computational pathology. Deep learning algorithms can provide accurate mappings given large numbers of labeled instances for training and validation. Generating adequate volume of quality labels has emerged as a critical barrier in computational pathology given the time and effort required from pathologists. In this paper we describe an approach for engaging crowds of medical students and pathologists that was used to produce a dataset of over 220,000 annotations of cell nuclei in breast cancers. We show how suggested annotations generated by a weak algorithm can improve the accuracy of annotations generated by non-experts and can yield useful data for training segmentation algorithms without laborious manual tracing. We systematically examine interrater agreement and describe modifications to the MaskRCNN model to improve cell mapping. We also describe a technique we call Decision Tree Approximation of Learned Embeddings (DTALE) that leverages nucleus segmentations and morphologic features to improve the transparency of nucleus classification models. The annotation data produced in this study are freely available for algorithm development and benchmarking at: https://sites.google.com/view/nucls.
△ Less
Submitted 17 February, 2021;
originally announced February 2021.
Polymer Modelling Predicts Chromosome Reorganisation in Senescence
Authors:
Michael Chiang,
Davide Michieletto,
Chris A. Brackley,
Nattaphon Rattanavirotkul,
Hisham Mohammed,
Davide Marenduzzo,
Tamir Chandra
Abstract:
Lamina-associated domains (LADs) cover a large part of the human genome and are thought to play a major role in shaping the nuclear architectural landscape. Here, we perform polymer simulations, microscopy and mass spectrometry to dissect the roles played by heterochromatin- and lamina-mediated interactions in nuclear organisation. Our model explains the conventional organisation of heterochromati…
▽ More
Lamina-associated domains (LADs) cover a large part of the human genome and are thought to play a major role in shaping the nuclear architectural landscape. Here, we perform polymer simulations, microscopy and mass spectrometry to dissect the roles played by heterochromatin- and lamina-mediated interactions in nuclear organisation. Our model explains the conventional organisation of heterochromatin and euchromatin in growing cells and the pathological organisation found in oncogene-induced senescence and progeria. We show that the experimentally observed changes in the locality of contacts in senescent and progeroid cells can be explained as arising due to phase transitions in the system. Within our simulations LADs are highly stochastic, as in experiments. Our model suggests that, once established, the senescent phenotype should be metastable even if lamina-mediated interactions were reinstated. Overall, our simulations uncover a generic physical mechanism that can regulate heterochromatin segregation and LAD formation in a wide range of mammalian nuclei.
△ Less
Submitted 19 October, 2020;
originally announced October 2020.
