Viable and necrotic tumor assessment from whole slide images of osteosarcoma using machine-learning and deep-learning models

Harish Babu Arunachalam, Rashika Mishra, Ovidiu Daescu, Kevin B Cederberg, Dinesh Rakheja, Anita L Sengupta, David Leonard, Rami Robert Hallac, Patrick J Leavey

Research output: Contribution to journalArticle

Abstract

Pathological estimation of tumor necrosis after chemotherapy is essential for patients with osteosarcoma. This study reports the first fully automated tool to assess viable and necrotic tumor in osteosarcoma, employing advances in histopathology digitization and automated learning. We selected 40 digitized whole slide images representing the heterogeneity of osteosarcoma and chemotherapy response. With the goal of labeling the diverse regions of the digitized tissue into viable tumor, necrotic tumor, and non-tumor, we trained 13 machine-learning models and selected the top performing one (a Support Vector Machine) based on reported accuracy. We also developed a deep-learning architecture and trained it on the same data set. We computed the receiver-operator characteristic for discrimination of non-tumor from tumor followed by conditional discrimination of necrotic from viable tumor and found our models performing exceptionally well. We then used the trained models to identify regions of interest on image-tiles generated from test whole slide images. The classification output is visualized as a tumor-prediction map, displaying the extent of viable and necrotic tumor in the slide image. Thus, we lay the foundation for a complete tumor assessment pipeline from original histology images to tumor-prediction map generation. The proposed pipeline can also be adopted for other types of tumor.

Original languageEnglish (US)
Article numbere0210706
JournalPloS one
Volume14
Issue number4
DOIs
StatePublished - Apr 1 2019

Fingerprint

osteosarcoma
artificial intelligence
Osteosarcoma
Learning systems
Tumors
learning
Learning
neoplasms
Neoplasms
Chemotherapy
drug therapy
Deep learning
Machine Learning
Pipelines
Drug Therapy
Histology
prediction
tiles
Analog to digital conversion
Tile

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)

Cite this

Viable and necrotic tumor assessment from whole slide images of osteosarcoma using machine-learning and deep-learning models. / Arunachalam, Harish Babu; Mishra, Rashika; Daescu, Ovidiu; Cederberg, Kevin B; Rakheja, Dinesh; Sengupta, Anita L; Leonard, David; Hallac, Rami Robert; Leavey, Patrick J.

In: PloS one, Vol. 14, No. 4, e0210706, 01.04.2019.

Research output: Contribution to journalArticle

@article{1bddce70c96647828b07f987bf52413f,
title = "Viable and necrotic tumor assessment from whole slide images of osteosarcoma using machine-learning and deep-learning models",
abstract = "Pathological estimation of tumor necrosis after chemotherapy is essential for patients with osteosarcoma. This study reports the first fully automated tool to assess viable and necrotic tumor in osteosarcoma, employing advances in histopathology digitization and automated learning. We selected 40 digitized whole slide images representing the heterogeneity of osteosarcoma and chemotherapy response. With the goal of labeling the diverse regions of the digitized tissue into viable tumor, necrotic tumor, and non-tumor, we trained 13 machine-learning models and selected the top performing one (a Support Vector Machine) based on reported accuracy. We also developed a deep-learning architecture and trained it on the same data set. We computed the receiver-operator characteristic for discrimination of non-tumor from tumor followed by conditional discrimination of necrotic from viable tumor and found our models performing exceptionally well. We then used the trained models to identify regions of interest on image-tiles generated from test whole slide images. The classification output is visualized as a tumor-prediction map, displaying the extent of viable and necrotic tumor in the slide image. Thus, we lay the foundation for a complete tumor assessment pipeline from original histology images to tumor-prediction map generation. The proposed pipeline can also be adopted for other types of tumor.",
author = "Arunachalam, {Harish Babu} and Rashika Mishra and Ovidiu Daescu and Cederberg, {Kevin B} and Dinesh Rakheja and Sengupta, {Anita L} and David Leonard and Hallac, {Rami Robert} and Leavey, {Patrick J}",
year = "2019",
month = "4",
day = "1",
doi = "10.1371/journal.pone.0210706",
language = "English (US)",
volume = "14",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "4",

}

TY - JOUR

T1 - Viable and necrotic tumor assessment from whole slide images of osteosarcoma using machine-learning and deep-learning models

AU - Arunachalam, Harish Babu

AU - Mishra, Rashika

AU - Daescu, Ovidiu

AU - Cederberg, Kevin B

AU - Rakheja, Dinesh

AU - Sengupta, Anita L

AU - Leonard, David

AU - Hallac, Rami Robert

AU - Leavey, Patrick J

PY - 2019/4/1

Y1 - 2019/4/1

N2 - Pathological estimation of tumor necrosis after chemotherapy is essential for patients with osteosarcoma. This study reports the first fully automated tool to assess viable and necrotic tumor in osteosarcoma, employing advances in histopathology digitization and automated learning. We selected 40 digitized whole slide images representing the heterogeneity of osteosarcoma and chemotherapy response. With the goal of labeling the diverse regions of the digitized tissue into viable tumor, necrotic tumor, and non-tumor, we trained 13 machine-learning models and selected the top performing one (a Support Vector Machine) based on reported accuracy. We also developed a deep-learning architecture and trained it on the same data set. We computed the receiver-operator characteristic for discrimination of non-tumor from tumor followed by conditional discrimination of necrotic from viable tumor and found our models performing exceptionally well. We then used the trained models to identify regions of interest on image-tiles generated from test whole slide images. The classification output is visualized as a tumor-prediction map, displaying the extent of viable and necrotic tumor in the slide image. Thus, we lay the foundation for a complete tumor assessment pipeline from original histology images to tumor-prediction map generation. The proposed pipeline can also be adopted for other types of tumor.

AB - Pathological estimation of tumor necrosis after chemotherapy is essential for patients with osteosarcoma. This study reports the first fully automated tool to assess viable and necrotic tumor in osteosarcoma, employing advances in histopathology digitization and automated learning. We selected 40 digitized whole slide images representing the heterogeneity of osteosarcoma and chemotherapy response. With the goal of labeling the diverse regions of the digitized tissue into viable tumor, necrotic tumor, and non-tumor, we trained 13 machine-learning models and selected the top performing one (a Support Vector Machine) based on reported accuracy. We also developed a deep-learning architecture and trained it on the same data set. We computed the receiver-operator characteristic for discrimination of non-tumor from tumor followed by conditional discrimination of necrotic from viable tumor and found our models performing exceptionally well. We then used the trained models to identify regions of interest on image-tiles generated from test whole slide images. The classification output is visualized as a tumor-prediction map, displaying the extent of viable and necrotic tumor in the slide image. Thus, we lay the foundation for a complete tumor assessment pipeline from original histology images to tumor-prediction map generation. The proposed pipeline can also be adopted for other types of tumor.

UR - http://www.scopus.com/inward/record.url?scp=85064441641&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064441641&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0210706

DO - 10.1371/journal.pone.0210706

M3 - Article

VL - 14

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 4

M1 - e0210706

ER -