Classification of lung nodules in CT scans using three-dimensional deep convolutional neural networks with a checkpoint ensemble method

Hwejin Jung, Bumsoo Kim, Inyeop Lee, Junhyun Lee, Jaewoo Kang

Research output: Contribution to journalArticle

3 Citations (Scopus)

Abstract

Background: Accurately detecting and examining lung nodules early is key in diagnosing lung cancers and thus one of the best ways to prevent lung cancer deaths. Radiologists spend countless hours detecting small spherical-shaped nodules in computed tomography (CT) images. In addition, even after detecting nodule candidates, a considerable amount of effort and time is required for them to determine whether they are real nodules. The aim of this paper is to introduce a high performance nodule classification method that uses three dimensional deep convolutional neural networks (DCNNs) and an ensemble method to distinguish nodules between non-nodules. Methods: In this paper, we use a three dimensional deep convolutional neural network (3D DCNN) with shortcut connections and a 3D DCNN with dense connections for lung nodule classification. The shortcut connections and dense connections successfully alleviate the gradient vanishing problem by allowing the gradient to pass quickly and directly. Connections help deep structured networks to obtain general as well as distinctive features of lung nodules. Moreover, we increased the dimension of DCNNs from two to three to capture 3D features. Compared with shallow 3D CNNs used in previous studies, deep 3D CNNs more effectively capture the features of spherical-shaped nodules. In addition, we use an alternative ensemble method called the checkpoint ensemble method to boost performance. Results: The performance of our nodule classification method is compared with that of the state-of-the-art methods which were used in the LUng Nodule Analysis 2016 Challenge. Our method achieves higher competition performance metric (CPM) scores than the state-of-the-art methods using deep learning. In the experimental setup ESB-ALL, the 3D DCNN with shortcut connections and the 3D DCNN with dense connections using the checkpoint ensemble method achieved the highest CPM score of 0.910. Conclusion: The result demonstrates that our method of using a 3D DCNN with shortcut connections, a 3D DCNN with dense connections, and the checkpoint ensemble method is effective for capturing 3D features of nodules and distinguishing nodules between non-nodules.

Original languageEnglish
Article number48
JournalBMC Medical Imaging
Volume18
Issue number1
DOIs
Publication statusPublished - 2018 Dec 3

Fingerprint

Tomography
Lung
Lung Neoplasms
Learning

Keywords

  • Convolutional neural network
  • Deep learning
  • Ensemble
  • Lung cancer
  • Lung nodule

ASJC Scopus subject areas

  • Radiology Nuclear Medicine and imaging

Cite this

Classification of lung nodules in CT scans using three-dimensional deep convolutional neural networks with a checkpoint ensemble method. / Jung, Hwejin; Kim, Bumsoo; Lee, Inyeop; Lee, Junhyun; Kang, Jaewoo.

In: BMC Medical Imaging, Vol. 18, No. 1, 48, 03.12.2018.

Research output: Contribution to journalArticle

@article{e69ad743e954445aa38ce31b1cf24544,
title = "Classification of lung nodules in CT scans using three-dimensional deep convolutional neural networks with a checkpoint ensemble method",
abstract = "Background: Accurately detecting and examining lung nodules early is key in diagnosing lung cancers and thus one of the best ways to prevent lung cancer deaths. Radiologists spend countless hours detecting small spherical-shaped nodules in computed tomography (CT) images. In addition, even after detecting nodule candidates, a considerable amount of effort and time is required for them to determine whether they are real nodules. The aim of this paper is to introduce a high performance nodule classification method that uses three dimensional deep convolutional neural networks (DCNNs) and an ensemble method to distinguish nodules between non-nodules. Methods: In this paper, we use a three dimensional deep convolutional neural network (3D DCNN) with shortcut connections and a 3D DCNN with dense connections for lung nodule classification. The shortcut connections and dense connections successfully alleviate the gradient vanishing problem by allowing the gradient to pass quickly and directly. Connections help deep structured networks to obtain general as well as distinctive features of lung nodules. Moreover, we increased the dimension of DCNNs from two to three to capture 3D features. Compared with shallow 3D CNNs used in previous studies, deep 3D CNNs more effectively capture the features of spherical-shaped nodules. In addition, we use an alternative ensemble method called the checkpoint ensemble method to boost performance. Results: The performance of our nodule classification method is compared with that of the state-of-the-art methods which were used in the LUng Nodule Analysis 2016 Challenge. Our method achieves higher competition performance metric (CPM) scores than the state-of-the-art methods using deep learning. In the experimental setup ESB-ALL, the 3D DCNN with shortcut connections and the 3D DCNN with dense connections using the checkpoint ensemble method achieved the highest CPM score of 0.910. Conclusion: The result demonstrates that our method of using a 3D DCNN with shortcut connections, a 3D DCNN with dense connections, and the checkpoint ensemble method is effective for capturing 3D features of nodules and distinguishing nodules between non-nodules.",
keywords = "Convolutional neural network, Deep learning, Ensemble, Lung cancer, Lung nodule",
author = "Hwejin Jung and Bumsoo Kim and Inyeop Lee and Junhyun Lee and Jaewoo Kang",
year = "2018",
month = "12",
day = "3",
doi = "10.1186/s12880-018-0286-0",
language = "English",
volume = "18",
journal = "BMC Medical Imaging",
issn = "1471-2342",
publisher = "BioMed Central",
number = "1",

}

TY - JOUR

T1 - Classification of lung nodules in CT scans using three-dimensional deep convolutional neural networks with a checkpoint ensemble method

AU - Jung, Hwejin

AU - Kim, Bumsoo

AU - Lee, Inyeop

AU - Lee, Junhyun

AU - Kang, Jaewoo

PY - 2018/12/3

Y1 - 2018/12/3

N2 - Background: Accurately detecting and examining lung nodules early is key in diagnosing lung cancers and thus one of the best ways to prevent lung cancer deaths. Radiologists spend countless hours detecting small spherical-shaped nodules in computed tomography (CT) images. In addition, even after detecting nodule candidates, a considerable amount of effort and time is required for them to determine whether they are real nodules. The aim of this paper is to introduce a high performance nodule classification method that uses three dimensional deep convolutional neural networks (DCNNs) and an ensemble method to distinguish nodules between non-nodules. Methods: In this paper, we use a three dimensional deep convolutional neural network (3D DCNN) with shortcut connections and a 3D DCNN with dense connections for lung nodule classification. The shortcut connections and dense connections successfully alleviate the gradient vanishing problem by allowing the gradient to pass quickly and directly. Connections help deep structured networks to obtain general as well as distinctive features of lung nodules. Moreover, we increased the dimension of DCNNs from two to three to capture 3D features. Compared with shallow 3D CNNs used in previous studies, deep 3D CNNs more effectively capture the features of spherical-shaped nodules. In addition, we use an alternative ensemble method called the checkpoint ensemble method to boost performance. Results: The performance of our nodule classification method is compared with that of the state-of-the-art methods which were used in the LUng Nodule Analysis 2016 Challenge. Our method achieves higher competition performance metric (CPM) scores than the state-of-the-art methods using deep learning. In the experimental setup ESB-ALL, the 3D DCNN with shortcut connections and the 3D DCNN with dense connections using the checkpoint ensemble method achieved the highest CPM score of 0.910. Conclusion: The result demonstrates that our method of using a 3D DCNN with shortcut connections, a 3D DCNN with dense connections, and the checkpoint ensemble method is effective for capturing 3D features of nodules and distinguishing nodules between non-nodules.

AB - Background: Accurately detecting and examining lung nodules early is key in diagnosing lung cancers and thus one of the best ways to prevent lung cancer deaths. Radiologists spend countless hours detecting small spherical-shaped nodules in computed tomography (CT) images. In addition, even after detecting nodule candidates, a considerable amount of effort and time is required for them to determine whether they are real nodules. The aim of this paper is to introduce a high performance nodule classification method that uses three dimensional deep convolutional neural networks (DCNNs) and an ensemble method to distinguish nodules between non-nodules. Methods: In this paper, we use a three dimensional deep convolutional neural network (3D DCNN) with shortcut connections and a 3D DCNN with dense connections for lung nodule classification. The shortcut connections and dense connections successfully alleviate the gradient vanishing problem by allowing the gradient to pass quickly and directly. Connections help deep structured networks to obtain general as well as distinctive features of lung nodules. Moreover, we increased the dimension of DCNNs from two to three to capture 3D features. Compared with shallow 3D CNNs used in previous studies, deep 3D CNNs more effectively capture the features of spherical-shaped nodules. In addition, we use an alternative ensemble method called the checkpoint ensemble method to boost performance. Results: The performance of our nodule classification method is compared with that of the state-of-the-art methods which were used in the LUng Nodule Analysis 2016 Challenge. Our method achieves higher competition performance metric (CPM) scores than the state-of-the-art methods using deep learning. In the experimental setup ESB-ALL, the 3D DCNN with shortcut connections and the 3D DCNN with dense connections using the checkpoint ensemble method achieved the highest CPM score of 0.910. Conclusion: The result demonstrates that our method of using a 3D DCNN with shortcut connections, a 3D DCNN with dense connections, and the checkpoint ensemble method is effective for capturing 3D features of nodules and distinguishing nodules between non-nodules.

KW - Convolutional neural network

KW - Deep learning

KW - Ensemble

KW - Lung cancer

KW - Lung nodule

UR - http://www.scopus.com/inward/record.url?scp=85057798731&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85057798731&partnerID=8YFLogxK

U2 - 10.1186/s12880-018-0286-0

DO - 10.1186/s12880-018-0286-0

M3 - Article

C2 - 30509191

AN - SCOPUS:85057798731

VL - 18

JO - BMC Medical Imaging

JF - BMC Medical Imaging

SN - 1471-2342

IS - 1

M1 - 48

ER -