1
|
Jain Y, Godwin LL, Joshi S, Mandarapu S, Le T, Lindskog C, Lundberg E, Börner K. Segmenting functional tissue units across human organs using community-driven development of generalizable machine learning algorithms. BIORXIV : THE PREPRINT SERVER FOR BIOLOGY 2023:2023.01.05.522764. [PMID: 36711953 PMCID: PMC9881902 DOI: 10.1101/2023.01.05.522764] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Grants] [Track Full Text] [Figures] [Subscribe] [Scholar Register] [Indexed: 01/07/2023]
Abstract
The development of a reference atlas of the healthy human body requires automated image segmentation of major anatomical structures across multiple organs based on spatial bioimages generated from various sources with differences in sample preparation. We present the setup and results of the "Hacking the Human Body" machine learning algorithm development competition hosted by the Human Biomolecular Atlas (HuBMAP) and the Human Protein Atlas (HPA) teams on the Kaggle platform. We showcase how 1,175 teams from 78 countries engaged in community- driven, open-science code development that resulted in machine learning models which successfully segment anatomical structures across five organs using histology images from two consortia and that will be productized in the HuBMAP data portal to process large datasets at scale in support of Human Reference Atlas construction. We discuss the benchmark data created for the competition, major challenges faced by the participants, and the winning models and strategies.
Collapse
Affiliation(s)
- Yashvardhan Jain
- Department of Intelligent Systems Engineering, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA
| | - Leah L. Godwin
- Department of Intelligent Systems Engineering, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA
| | - Sripad Joshi
- Department of Intelligent Systems Engineering, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA
| | - Shriya Mandarapu
- Department of Intelligent Systems Engineering, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA
| | - Trang Le
- Science for Life Laboratory, School of Engineering Sciences in Chemistry, Biotechnology and Health, KTH - Royal Institute of Technology, Stockholm, Sweden
- Department of Bioengineering, Stanford University, Stanford, CA 94305, USA
| | - Cecilia Lindskog
- Department of Immunology, Genetics and Pathology, Division of Cancer Precision Medicine, Uppsala University, Uppsala, Sweden
| | - Emma Lundberg
- Science for Life Laboratory, School of Engineering Sciences in Chemistry, Biotechnology and Health, KTH - Royal Institute of Technology, Stockholm, Sweden
- Department of Bioengineering, Stanford University, Stanford, CA 94305, USA
- Department of Pathology, Stanford University, Stanford, CA 94305, USA
- Chan Zuckerberg Biohub, San Francisco, CA 94305, USA
| | - Katy Börner
- Department of Intelligent Systems Engineering, Luddy School of Informatics, Computing, and Engineering, Indiana University, Bloomington, IN 47408, USA
| |
Collapse
|
2
|
Lutnick B, Manthey D, Becker JU, Zuckerman JE, Rodrigues L, Jen KY, Sarder P. A tool for federated training of segmentation models on whole slide images. J Pathol Inform 2022; 13:100101. [PMID: 35910077 PMCID: PMC9326476 DOI: 10.1016/j.jpi.2022.100101] [Citation(s) in RCA: 2] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/20/2022] [Revised: 04/06/2022] [Accepted: 04/09/2022] [Indexed: 11/22/2022] Open
Abstract
The largest bottleneck to the development of convolutional neural network (CNN) models in the computational pathology domain is the collection and curation of diverse training datasets. Training CNNs requires large cohorts of image data, and model generalizability is dependent on training data heterogeneity. Including data from multiple centers enhances the generalizability of CNN-based models, but this is hindered by the logistical challenges of sharing medical data. In this paper, we explore the feasibility of training our recently developed cloud-based segmentation tool (Histo-Cloud) using federated learning. Using a dataset of renal tissue biopsies we show that federated training to segment interstitial fibrosis and tubular atrophy (IFTA) using datasets from three institutions is not found to be different from a training by pooling the data on one server when tested on a fourth (holdout) institution's data. Further, training a model to segment glomeruli for a federated dataset (split by staining) demonstrates similar performance.
Collapse
Affiliation(s)
- Brendon Lutnick
- Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, NY, USA
| | | | - Jan U. Becker
- Institute of Pathology, University Hospital Cologne, Cologne, Germany
| | - Jonathan E. Zuckerman
- Department of Pathology and Laboratory Medicine, University of California at Los Angeles, Los Angeles, CA, USA
| | - Luis Rodrigues
- University Clinic of Nephrology, Faculty of Medicine, University of Coimbra, Portugal
| | - Kuang-Yu Jen
- University of California, Davis School of Medicine, Sacramento, CA, USA
| | - Pinaki Sarder
- Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, NY, USA
| |
Collapse
|
3
|
Lutnick B, Murali LK, Ginley B, Rosenberg AZ, Sarder P. Histo-Fetch - On-the-Fly Processing of Gigapixel Whole Slide Images Simplifies and Speeds Neural Network Training. J Pathol Inform 2022; 13:7. [PMID: 35136674 PMCID: PMC8794032 DOI: 10.4103/jpi.jpi_59_20] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/14/2020] [Revised: 01/23/2021] [Accepted: 11/11/2021] [Indexed: 02/04/2023] Open
Abstract
BACKGROUND Training convolutional neural networks using pathology whole slide images (WSIs) is traditionally prefaced by the extraction of a training dataset of image patches. While effective, for large datasets of WSIs, this dataset preparation is inefficient. METHODS We created a custom pipeline (histo-fetch) to efficiently extract random patches and labels from pathology WSIs for input to a neural network on-the-fly. We prefetch these patches as needed during network training, avoiding the need for WSI preparation such as chopping/tiling. RESULTS & CONCLUSIONS We demonstrate the utility of this pipeline to perform artificial stain transfer and image generation using the popular networks CycleGAN and ProGAN, respectively. For a large WSI dataset, histo-fetch is 98.6% faster to start training and used 7535x less disk space.
Collapse
Affiliation(s)
- Brendon Lutnick
- Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, NY, USA
| | | | - Brandon Ginley
- Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, NY, USA
| | - Avi Z. Rosenberg
- Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, MD, USA
| | - Pinaki Sarder
- Department of Pathology and Anatomical Sciences, SUNY Buffalo, Buffalo, NY, USA
- Department of Biomedical Engineering, SUNY Buffalo, Buffalo, NY, USA
| |
Collapse
|
4
|
Govind D, Becker JU, Miecznikowski J, Rosenberg AZ, Dang J, Tharaux PL, Yacoub R, Thaiss F, Hoyer PF, Manthey D, Lutnick B, Worral AM, Mohammad I, Walavalkar V, Tomaszewski JE, Jen KY, Sarder P. PodoSighter: A Cloud-Based Tool for Label-Free Podocyte Detection in Kidney Whole-Slide Images. J Am Soc Nephrol 2021; 32:2795-2813. [PMID: 34479966 PMCID: PMC8806084 DOI: 10.1681/asn.2021050630] [Citation(s) in RCA: 16] [Impact Index Per Article: 5.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/10/2021] [Accepted: 08/08/2021] [Indexed: 02/04/2023] Open
Abstract
BACKGROUND Podocyte depletion precedes progressive glomerular damage in several kidney diseases. However, the current standard of visual detection and quantification of podocyte nuclei from brightfield microscopy images is laborious and imprecise. METHODS We have developed PodoSighter, an online cloud-based tool, to automatically identify and quantify podocyte nuclei from giga-pixel brightfield whole-slide images (WSIs) using deep learning. Ground-truth to train the tool used immunohistochemically or immunofluorescence-labeled images from a multi-institutional cohort of 122 histologic sections from mouse, rat, and human kidneys. To demonstrate the generalizability of our tool in investigating podocyte loss in clinically relevant samples, we tested it in rodent models of glomerular diseases, including diabetic kidney disease, crescentic GN, and dose-dependent direct podocyte toxicity and depletion, and in human biopsies from steroid-resistant nephrotic syndrome and from human autopsy tissues. RESULTS The optimal model yielded high sensitivity/specificity of 0.80/0.80, 0.81/0.86, and 0.80/0.91, in mouse, rat, and human images, respectively, from periodic acid-Schiff-stained WSIs. Furthermore, the podocyte nuclear morphometrics extracted using PodoSighter were informative in identifying diseased glomeruli. We have made PodoSighter freely available to the general public as turnkey plugins in a cloud-based web application for end users. CONCLUSIONS Our study demonstrates an automated computational approach to detect and quantify podocyte nuclei in standard histologically stained WSIs, facilitating podocyte research, and enabling possible future clinical applications.
Collapse
Affiliation(s)
- Darshana Govind
- Department of Pathology and Anatomical Sciences, University at Buffalo, Buffalo, New York
| | - Jan U. Becker
- Institute of Pathology, University Hospital of Cologne, Cologne, Germany
| | | | - Avi Z. Rosenberg
- Department of Pathology, Johns Hopkins University School of Medicine, Baltimore, Maryland
| | | | | | - Rabi Yacoub
- Department of Internal Medicine, University at Buffalo, Buffalo, New York
| | - Friedrich Thaiss
- Third Medical Department of Clinical Medicine, University Hospital Hamburg Eppendorf, Hamburg, Germany
| | - Peter F. Hoyer
- Pediatric Nephrology, University Hospital Essen, Essen, Germany
| | | | - Brendon Lutnick
- Department of Pathology and Anatomical Sciences, University at Buffalo, Buffalo, New York
| | - Amber M. Worral
- Department of Pathology and Anatomical Sciences, University at Buffalo, Buffalo, New York
| | - Imtiaz Mohammad
- Department of Pathology and Anatomical Sciences, University at Buffalo, Buffalo, New York
| | - Vighnesh Walavalkar
- Department of Pathology, University of California San Francisco, San Francisco, California
| | - John E. Tomaszewski
- Department of Pathology and Anatomical Sciences, University at Buffalo, Buffalo, New York
| | - Kuang-Yu Jen
- Department of Pathology and Laboratory Medicine, University of California, Sacramento, California
| | - Pinaki Sarder
- Department of Pathology and Anatomical Sciences, University at Buffalo, Buffalo, New York
| |
Collapse
|