Deep learning quantification of percent steatosis in donor liver biopsy frozen sections
Background: Pathologist evaluation of donor liver biopsies provides information for accepting or discarding potential donor livers. Due to the urgent nature of the decision process, this is regularly performed using frozen sectioning at the time of biopsy. The percent steatosis in a donor liver biop...
Main Authors: | , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
Elsevier
2020-10-01
|
Series: | EBioMedicine |
Subjects: | |
Online Access: | http://www.sciencedirect.com/science/article/pii/S2352396420304059 |
id |
doaj-755e8a14cc0348e8b94c4e023403800c |
---|---|
record_format |
Article |
spelling |
doaj-755e8a14cc0348e8b94c4e023403800c2020-11-25T03:08:01ZengElsevierEBioMedicine2352-39642020-10-0160103029Deep learning quantification of percent steatosis in donor liver biopsy frozen sectionsLulu Sun0Jon N. Marsh1Matthew K. Matlock2Ling Chen3Joseph P. Gaut4Elizabeth M. Brunt5S. Joshua Swamidass6Ta-Chiang Liu7Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United StatesDepartment of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United States; Institue for Informatics (I2), Washington University School of Medicine, St. Louis, MO, United StatesDepartment of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United StatesDivision of Biostatistics, Washington University School of Medicine, St. Louis, MO, United StatesDepartment of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United StatesDepartment of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United StatesDepartment of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United States; Institue for Informatics (I2), Washington University School of Medicine, St. Louis, MO, United States; Corresponding authors: Ta-Chiang Liu (lead contact) and S. Joshua Swamidass, Department of Pathology and Immunology, 660 S. Euclid Ave, Box 8118, Saint Louis, MO, 63110.Department of Pathology and Immunology, Washington University School of Medicine, St. Louis, MO, United States; Lead contact; Corresponding authors: Ta-Chiang Liu (lead contact) and S. Joshua Swamidass, Department of Pathology and Immunology, 660 S. Euclid Ave, Box 8118, Saint Louis, MO, 63110.Background: Pathologist evaluation of donor liver biopsies provides information for accepting or discarding potential donor livers. Due to the urgent nature of the decision process, this is regularly performed using frozen sectioning at the time of biopsy. The percent steatosis in a donor liver biopsy correlates with transplant outcome, however there is significant inter- and intra-observer variability in quantifying steatosis, compounded by frozen section artifact. We hypothesized that a deep learning model could identify and quantify steatosis in donor liver biopsies. Methods: We developed a deep learning convolutional neural network that generates a steatosis probability map from an input whole slide image (WSI) of a hematoxylin and eosin-stained frozen section, and subsequently calculates the percent steatosis. Ninety-six WSI of frozen donor liver sections from our transplant pathology service were annotated for steatosis and used to train (n = 30 WSI) and test (n = 66 WSI) the deep learning model. Findings: The model had good correlation and agreement with the annotation in both the training set (r of 0.88, intraclass correlation coefficient [ICC] of 0.88) and novel input test sets (r = 0.85 and ICC=0.85). These measurements were superior to the estimates of the on-service pathologist at the time of initial evaluation (r = 0.52 and ICC=0.52 for the training set, and r = 0.74 and ICC=0.72 for the test set). Interpretation: Use of this deep learning algorithm could be incorporated into routine pathology workflows for fast, accurate, and reproducible donor liver evaluation. Funding: Mid-America Transplant Societyhttp://www.sciencedirect.com/science/article/pii/S2352396420304059Liver transplantationBiopsySteatosisDeep learningConvolutional neural networkImage analysis |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Lulu Sun Jon N. Marsh Matthew K. Matlock Ling Chen Joseph P. Gaut Elizabeth M. Brunt S. Joshua Swamidass Ta-Chiang Liu |
spellingShingle |
Lulu Sun Jon N. Marsh Matthew K. Matlock Ling Chen Joseph P. Gaut Elizabeth M. Brunt S. Joshua Swamidass Ta-Chiang Liu Deep learning quantification of percent steatosis in donor liver biopsy frozen sections EBioMedicine Liver transplantation Biopsy Steatosis Deep learning Convolutional neural network Image analysis |
author_facet |
Lulu Sun Jon N. Marsh Matthew K. Matlock Ling Chen Joseph P. Gaut Elizabeth M. Brunt S. Joshua Swamidass Ta-Chiang Liu |
author_sort |
Lulu Sun |
title |
Deep learning quantification of percent steatosis in donor liver biopsy frozen sections |
title_short |
Deep learning quantification of percent steatosis in donor liver biopsy frozen sections |
title_full |
Deep learning quantification of percent steatosis in donor liver biopsy frozen sections |
title_fullStr |
Deep learning quantification of percent steatosis in donor liver biopsy frozen sections |
title_full_unstemmed |
Deep learning quantification of percent steatosis in donor liver biopsy frozen sections |
title_sort |
deep learning quantification of percent steatosis in donor liver biopsy frozen sections |
publisher |
Elsevier |
series |
EBioMedicine |
issn |
2352-3964 |
publishDate |
2020-10-01 |
description |
Background: Pathologist evaluation of donor liver biopsies provides information for accepting or discarding potential donor livers. Due to the urgent nature of the decision process, this is regularly performed using frozen sectioning at the time of biopsy. The percent steatosis in a donor liver biopsy correlates with transplant outcome, however there is significant inter- and intra-observer variability in quantifying steatosis, compounded by frozen section artifact. We hypothesized that a deep learning model could identify and quantify steatosis in donor liver biopsies. Methods: We developed a deep learning convolutional neural network that generates a steatosis probability map from an input whole slide image (WSI) of a hematoxylin and eosin-stained frozen section, and subsequently calculates the percent steatosis. Ninety-six WSI of frozen donor liver sections from our transplant pathology service were annotated for steatosis and used to train (n = 30 WSI) and test (n = 66 WSI) the deep learning model. Findings: The model had good correlation and agreement with the annotation in both the training set (r of 0.88, intraclass correlation coefficient [ICC] of 0.88) and novel input test sets (r = 0.85 and ICC=0.85). These measurements were superior to the estimates of the on-service pathologist at the time of initial evaluation (r = 0.52 and ICC=0.52 for the training set, and r = 0.74 and ICC=0.72 for the test set). Interpretation: Use of this deep learning algorithm could be incorporated into routine pathology workflows for fast, accurate, and reproducible donor liver evaluation. Funding: Mid-America Transplant Society |
topic |
Liver transplantation Biopsy Steatosis Deep learning Convolutional neural network Image analysis |
url |
http://www.sciencedirect.com/science/article/pii/S2352396420304059 |
work_keys_str_mv |
AT lulusun deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT jonnmarsh deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT matthewkmatlock deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT lingchen deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT josephpgaut deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT elizabethmbrunt deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT sjoshuaswamidass deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections AT tachiangliu deeplearningquantificationofpercentsteatosisindonorliverbiopsyfrozensections |
_version_ |
1724667657270067200 |