Convolutional neural network for histopathological analysis of osteosarcoma

Rashika Mishra, Ovidiu Daescu, Patrick Leavey, Dinesh Rakheja, Anita Sengupta

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

Pathologists often deal with high complexity and sometimes disagreement over osteosarcoma tumor classification due to cellular heterogeneity in the dataset. Segmentation and classification of histology tissue in H&E stained tumor image datasets is a challenging task because of intra-class variations, inter-class similarity, crowded context, and noisy data. In recent years, deep learning approaches have led to encouraging results in breast cancer and prostate cancer analysis. In this article, we propose convolutional neural network (CNN) as a tool to improve efficiency and accuracy of osteosarcoma tumor classification into tumor classes (viable tumor, necrosis) versus nontumor. The proposed CNN architecture contains eight learned layers: three sets of stacked two convolutional layers interspersed with max pooling layers for feature extraction and two fully connected layers with data augmentation strategies to boost performance. The use of a neural network results in higher accuracy of average 92% for the classification. We compare the proposed architecture with three existing and proven CNN architectures for image classification: AlexNet, LeNet, and VGGNet. We also provide a pipeline to calculate percentage necrosis in a given whole slide image. We conclude that the use of neural networks can assure both high accuracy and efficiency in osteosarcoma classification.

Original languageEnglish (US)
Pages (from-to)313-325
Number of pages13
JournalJournal of Computational Biology
Volume25
Issue number3
DOIs
StatePublished - Mar 1 2018

Fingerprint

Osteosarcoma
Tumors
Tumor
Neural Networks
Neural networks
Necrosis
Network Architecture
Network architecture
High Accuracy
Prostatic Neoplasms
Data Augmentation
Histology
Prostate Cancer
Pooling
Image classification
Image Classification
Efficiency
Noisy Data
Neoplasms
Breast Cancer

Keywords

  • convolutional neural network
  • histology image analysis
  • osteosarcoma

ASJC Scopus subject areas

  • Modeling and Simulation
  • Molecular Biology
  • Genetics
  • Computational Mathematics
  • Computational Theory and Mathematics

Cite this

Convolutional neural network for histopathological analysis of osteosarcoma. / Mishra, Rashika; Daescu, Ovidiu; Leavey, Patrick; Rakheja, Dinesh; Sengupta, Anita.

In: Journal of Computational Biology, Vol. 25, No. 3, 01.03.2018, p. 313-325.

Research output: Contribution to journalArticle

@article{fd7e2db6270e47d9802181d2e4e576b9,
title = "Convolutional neural network for histopathological analysis of osteosarcoma",
abstract = "Pathologists often deal with high complexity and sometimes disagreement over osteosarcoma tumor classification due to cellular heterogeneity in the dataset. Segmentation and classification of histology tissue in H&E stained tumor image datasets is a challenging task because of intra-class variations, inter-class similarity, crowded context, and noisy data. In recent years, deep learning approaches have led to encouraging results in breast cancer and prostate cancer analysis. In this article, we propose convolutional neural network (CNN) as a tool to improve efficiency and accuracy of osteosarcoma tumor classification into tumor classes (viable tumor, necrosis) versus nontumor. The proposed CNN architecture contains eight learned layers: three sets of stacked two convolutional layers interspersed with max pooling layers for feature extraction and two fully connected layers with data augmentation strategies to boost performance. The use of a neural network results in higher accuracy of average 92{\%} for the classification. We compare the proposed architecture with three existing and proven CNN architectures for image classification: AlexNet, LeNet, and VGGNet. We also provide a pipeline to calculate percentage necrosis in a given whole slide image. We conclude that the use of neural networks can assure both high accuracy and efficiency in osteosarcoma classification.",
keywords = "convolutional neural network, histology image analysis, osteosarcoma",
author = "Rashika Mishra and Ovidiu Daescu and Patrick Leavey and Dinesh Rakheja and Anita Sengupta",
year = "2018",
month = "3",
day = "1",
doi = "10.1089/cmb.2017.0153",
language = "English (US)",
volume = "25",
pages = "313--325",
journal = "Journal of Computational Biology",
issn = "1066-5277",
publisher = "Mary Ann Liebert Inc.",
number = "3",

}

TY - JOUR

T1 - Convolutional neural network for histopathological analysis of osteosarcoma

AU - Mishra, Rashika

AU - Daescu, Ovidiu

AU - Leavey, Patrick

AU - Rakheja, Dinesh

AU - Sengupta, Anita

PY - 2018/3/1

Y1 - 2018/3/1

N2 - Pathologists often deal with high complexity and sometimes disagreement over osteosarcoma tumor classification due to cellular heterogeneity in the dataset. Segmentation and classification of histology tissue in H&E stained tumor image datasets is a challenging task because of intra-class variations, inter-class similarity, crowded context, and noisy data. In recent years, deep learning approaches have led to encouraging results in breast cancer and prostate cancer analysis. In this article, we propose convolutional neural network (CNN) as a tool to improve efficiency and accuracy of osteosarcoma tumor classification into tumor classes (viable tumor, necrosis) versus nontumor. The proposed CNN architecture contains eight learned layers: three sets of stacked two convolutional layers interspersed with max pooling layers for feature extraction and two fully connected layers with data augmentation strategies to boost performance. The use of a neural network results in higher accuracy of average 92% for the classification. We compare the proposed architecture with three existing and proven CNN architectures for image classification: AlexNet, LeNet, and VGGNet. We also provide a pipeline to calculate percentage necrosis in a given whole slide image. We conclude that the use of neural networks can assure both high accuracy and efficiency in osteosarcoma classification.

AB - Pathologists often deal with high complexity and sometimes disagreement over osteosarcoma tumor classification due to cellular heterogeneity in the dataset. Segmentation and classification of histology tissue in H&E stained tumor image datasets is a challenging task because of intra-class variations, inter-class similarity, crowded context, and noisy data. In recent years, deep learning approaches have led to encouraging results in breast cancer and prostate cancer analysis. In this article, we propose convolutional neural network (CNN) as a tool to improve efficiency and accuracy of osteosarcoma tumor classification into tumor classes (viable tumor, necrosis) versus nontumor. The proposed CNN architecture contains eight learned layers: three sets of stacked two convolutional layers interspersed with max pooling layers for feature extraction and two fully connected layers with data augmentation strategies to boost performance. The use of a neural network results in higher accuracy of average 92% for the classification. We compare the proposed architecture with three existing and proven CNN architectures for image classification: AlexNet, LeNet, and VGGNet. We also provide a pipeline to calculate percentage necrosis in a given whole slide image. We conclude that the use of neural networks can assure both high accuracy and efficiency in osteosarcoma classification.

KW - convolutional neural network

KW - histology image analysis

KW - osteosarcoma

UR - http://www.scopus.com/inward/record.url?scp=85045457451&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85045457451&partnerID=8YFLogxK

U2 - 10.1089/cmb.2017.0153

DO - 10.1089/cmb.2017.0153

M3 - Article

C2 - 29083930

AN - SCOPUS:85045457451

VL - 25

SP - 313

EP - 325

JO - Journal of Computational Biology

JF - Journal of Computational Biology

SN - 1066-5277

IS - 3

ER -