Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer
Reliable entity subtyping is paramount for therapy stratification in lung cancer. Morphological evaluation remains the basis for entity subtyping and directs the application of additional methods such as immunohistochemistry (IHC). The decision of whether to perform IHC for subtyping is subjective,...
Main Authors: | , , , , , , , , , , , , , , , , , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2020-06-01
|
Series: | Cancers |
Subjects: | |
Online Access: | https://www.mdpi.com/2072-6694/12/6/1604 |
id |
doaj-d77f75e060c541c98e178178e8b928c4 |
---|---|
record_format |
Article |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Mark Kriegsmann Christian Haag Cleo-Aron Weis Georg Steinbuss Arne Warth Christiane Zgorzelski Thomas Muley Hauke Winter Martin E. Eichhorn Florian Eichhorn Joerg Kriegsmann Petros Christopolous Michael Thomas Mathias Witzens-Harig Peter Sinn Moritz von Winterfeld Claus Peter Heussel Felix J. F. Herth Frederick Klauschen Albrecht Stenzinger Katharina Kriegsmann |
spellingShingle |
Mark Kriegsmann Christian Haag Cleo-Aron Weis Georg Steinbuss Arne Warth Christiane Zgorzelski Thomas Muley Hauke Winter Martin E. Eichhorn Florian Eichhorn Joerg Kriegsmann Petros Christopolous Michael Thomas Mathias Witzens-Harig Peter Sinn Moritz von Winterfeld Claus Peter Heussel Felix J. F. Herth Frederick Klauschen Albrecht Stenzinger Katharina Kriegsmann Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer Cancers Artificial intelligence deep learning lung cancer histology non-small cell lung cancer small cell lung cancer |
author_facet |
Mark Kriegsmann Christian Haag Cleo-Aron Weis Georg Steinbuss Arne Warth Christiane Zgorzelski Thomas Muley Hauke Winter Martin E. Eichhorn Florian Eichhorn Joerg Kriegsmann Petros Christopolous Michael Thomas Mathias Witzens-Harig Peter Sinn Moritz von Winterfeld Claus Peter Heussel Felix J. F. Herth Frederick Klauschen Albrecht Stenzinger Katharina Kriegsmann |
author_sort |
Mark Kriegsmann |
title |
Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer |
title_short |
Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer |
title_full |
Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer |
title_fullStr |
Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer |
title_full_unstemmed |
Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung Cancer |
title_sort |
deep learning for the classification of small-cell and non-small-cell lung cancer |
publisher |
MDPI AG |
series |
Cancers |
issn |
2072-6694 |
publishDate |
2020-06-01 |
description |
Reliable entity subtyping is paramount for therapy stratification in lung cancer. Morphological evaluation remains the basis for entity subtyping and directs the application of additional methods such as immunohistochemistry (IHC). The decision of whether to perform IHC for subtyping is subjective, and access to IHC is not available worldwide. Thus, the application of additional methods to support morphological entity subtyping is desirable. Therefore, the ability of convolutional neuronal networks (CNNs) to classify the most common lung cancer subtypes, pulmonary adenocarcinoma (ADC), pulmonary squamous cell carcinoma (SqCC), and small-cell lung cancer (SCLC), was evaluated. A cohort of 80 ADC, 80 SqCC, 80 SCLC, and 30 skeletal muscle specimens was assembled; slides were scanned; tumor areas were annotated; image patches were extracted; and cases were randomly assigned to a training, validation or test set. Multiple CNN architectures (VGG16, InceptionV3, and InceptionResNetV2) were trained and optimized to classify the four entities. A quality control (QC) metric was established. An optimized InceptionV3 CNN architecture yielded the highest classification accuracy and was used for the classification of the test set. Image patch and patient-based CNN classification results were 95% and 100% in the test set after the application of strict QC. Misclassified cases mainly included ADC and SqCC. The QC metric identified cases that needed further IHC for definite entity subtyping. The study highlights the potential and limitations of CNN image classification models for tumor differentiation. |
topic |
Artificial intelligence deep learning lung cancer histology non-small cell lung cancer small cell lung cancer |
url |
https://www.mdpi.com/2072-6694/12/6/1604 |
work_keys_str_mv |
AT markkriegsmann deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT christianhaag deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT cleoaronweis deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT georgsteinbuss deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT arnewarth deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT christianezgorzelski deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT thomasmuley deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT haukewinter deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT martineeichhorn deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT florianeichhorn deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT joergkriegsmann deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT petroschristopolous deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT michaelthomas deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT mathiaswitzensharig deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT petersinn deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT moritzvonwinterfeld deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT clauspeterheussel deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT felixjfherth deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT frederickklauschen deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT albrechtstenzinger deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer AT katharinakriegsmann deeplearningfortheclassificationofsmallcellandnonsmallcelllungcancer |
_version_ |
1724730536940797952 |
spelling |
doaj-d77f75e060c541c98e178178e8b928c42020-11-25T02:52:21ZengMDPI AGCancers2072-66942020-06-01121604160410.3390/cancers12061604Deep Learning for the Classification of Small-Cell and Non-Small-Cell Lung CancerMark Kriegsmann0Christian Haag1Cleo-Aron Weis2Georg Steinbuss3Arne Warth4Christiane Zgorzelski5Thomas Muley6Hauke Winter7Martin E. Eichhorn8Florian Eichhorn9Joerg Kriegsmann10Petros Christopolous11Michael Thomas12Mathias Witzens-Harig13Peter Sinn14Moritz von Winterfeld15Claus Peter Heussel16Felix J. F. Herth17Frederick Klauschen18Albrecht Stenzinger19Katharina Kriegsmann20Institute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyInstitute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyInstitute of Pathology, University Medical Centre Mannheim, Heidelberg University, 68782 Mannheim, GermanyInstitute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyInstitute of Pathology, Cytopathology, and Molecular Pathology, UEGP MVZ Gießen/Wetzlar/LimburgInstitute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyMolecular Pathology Trier, 54296 Trier, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyMedical Faculty Heidelberg University, 69120 Heidelberg, GermanyInstitute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyInstitute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyTranslational Lung Research Centre Heidelberg, Member of the German Centre for Lung Research (DZL), 69120 Heidelberg, GermanyInstitute of Pathology, University Hospital Charité, 10117 Berlin, GermanyInstitute of Pathology, Heidelberg University, 69120 Heidelberg, GermanyDepartment Hematology, Oncology and Rheumatology, Heidelberg University, 69120 Heidelberg, GermanyReliable entity subtyping is paramount for therapy stratification in lung cancer. Morphological evaluation remains the basis for entity subtyping and directs the application of additional methods such as immunohistochemistry (IHC). The decision of whether to perform IHC for subtyping is subjective, and access to IHC is not available worldwide. Thus, the application of additional methods to support morphological entity subtyping is desirable. Therefore, the ability of convolutional neuronal networks (CNNs) to classify the most common lung cancer subtypes, pulmonary adenocarcinoma (ADC), pulmonary squamous cell carcinoma (SqCC), and small-cell lung cancer (SCLC), was evaluated. A cohort of 80 ADC, 80 SqCC, 80 SCLC, and 30 skeletal muscle specimens was assembled; slides were scanned; tumor areas were annotated; image patches were extracted; and cases were randomly assigned to a training, validation or test set. Multiple CNN architectures (VGG16, InceptionV3, and InceptionResNetV2) were trained and optimized to classify the four entities. A quality control (QC) metric was established. An optimized InceptionV3 CNN architecture yielded the highest classification accuracy and was used for the classification of the test set. Image patch and patient-based CNN classification results were 95% and 100% in the test set after the application of strict QC. Misclassified cases mainly included ADC and SqCC. The QC metric identified cases that needed further IHC for definite entity subtyping. The study highlights the potential and limitations of CNN image classification models for tumor differentiation.https://www.mdpi.com/2072-6694/12/6/1604Artificial intelligencedeep learninglung cancerhistologynon-small cell lung cancersmall cell lung cancer |