Deep feature learning for pulmonary nodule classification in a lung CT

Bum Chae Kim, Yu Sub Sung, Heung-Il Suk

Research output: Chapter in Book/Report/Conference proceedingConference contribution

24 Citations (Scopus)

Abstract

In this paper, we propose a novel method of identifying pulmonary nodules in a lung CT. Specifically, we devise a deep neural network by which we extract abstract information inherent in raw hand-crafted imaging features. We then combine the deep learned representations with the original raw imaging features into a long feature vector. By taking the combined feature vectors, we train a classifier, preceded by a feature selection via t-test. To validate the effectiveness of the proposed method, we performed experiments on our in-house dataset of 20 subjects; 3,598 pulmonary nodules (malignant: 178, benign: 3,420), which were manually segmented by a radiologist. In our experiments, we achieved the maximal accuracy of 95.5%, sensitivity of 94.4%, and AUC of 0.987, outperforming the competing method.

Original languageEnglish
Title of host publication4th International Winter Conference on Brain-Computer Interface, BCI 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781467378413
DOIs
Publication statusPublished - 2016 Apr 20
Event4th International Winter Conference on Brain-Computer Interface, BCI 2016 - Gangwon Province, Korea, Republic of
Duration: 2016 Feb 222016 Feb 24

Publication series

Name4th International Winter Conference on Brain-Computer Interface, BCI 2016

Other

Other4th International Winter Conference on Brain-Computer Interface, BCI 2016
CountryKorea, Republic of
CityGangwon Province
Period16/2/2216/2/24

Fingerprint

Imaging techniques
Feature extraction
Classifiers
Experiments
Deep neural networks

Keywords

  • Deep learning
  • Lung cancer
  • Pulmonary nodule classification
  • Stacked denoising autoencoder

ASJC Scopus subject areas

  • Human-Computer Interaction

Cite this

Kim, B. C., Sung, Y. S., & Suk, H-I. (2016). Deep feature learning for pulmonary nodule classification in a lung CT. In 4th International Winter Conference on Brain-Computer Interface, BCI 2016 [7457462] (4th International Winter Conference on Brain-Computer Interface, BCI 2016). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IWW-BCI.2016.7457462

Deep feature learning for pulmonary nodule classification in a lung CT. / Kim, Bum Chae; Sung, Yu Sub; Suk, Heung-Il.

4th International Winter Conference on Brain-Computer Interface, BCI 2016. Institute of Electrical and Electronics Engineers Inc., 2016. 7457462 (4th International Winter Conference on Brain-Computer Interface, BCI 2016).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Kim, BC, Sung, YS & Suk, H-I 2016, Deep feature learning for pulmonary nodule classification in a lung CT. in 4th International Winter Conference on Brain-Computer Interface, BCI 2016., 7457462, 4th International Winter Conference on Brain-Computer Interface, BCI 2016, Institute of Electrical and Electronics Engineers Inc., 4th International Winter Conference on Brain-Computer Interface, BCI 2016, Gangwon Province, Korea, Republic of, 16/2/22. https://doi.org/10.1109/IWW-BCI.2016.7457462
Kim BC, Sung YS, Suk H-I. Deep feature learning for pulmonary nodule classification in a lung CT. In 4th International Winter Conference on Brain-Computer Interface, BCI 2016. Institute of Electrical and Electronics Engineers Inc. 2016. 7457462. (4th International Winter Conference on Brain-Computer Interface, BCI 2016). https://doi.org/10.1109/IWW-BCI.2016.7457462
Kim, Bum Chae ; Sung, Yu Sub ; Suk, Heung-Il. / Deep feature learning for pulmonary nodule classification in a lung CT. 4th International Winter Conference on Brain-Computer Interface, BCI 2016. Institute of Electrical and Electronics Engineers Inc., 2016. (4th International Winter Conference on Brain-Computer Interface, BCI 2016).
@inproceedings{9390fb8499ff4ae79761362cce17cde8,
title = "Deep feature learning for pulmonary nodule classification in a lung CT",
abstract = "In this paper, we propose a novel method of identifying pulmonary nodules in a lung CT. Specifically, we devise a deep neural network by which we extract abstract information inherent in raw hand-crafted imaging features. We then combine the deep learned representations with the original raw imaging features into a long feature vector. By taking the combined feature vectors, we train a classifier, preceded by a feature selection via t-test. To validate the effectiveness of the proposed method, we performed experiments on our in-house dataset of 20 subjects; 3,598 pulmonary nodules (malignant: 178, benign: 3,420), which were manually segmented by a radiologist. In our experiments, we achieved the maximal accuracy of 95.5{\%}, sensitivity of 94.4{\%}, and AUC of 0.987, outperforming the competing method.",
keywords = "Deep learning, Lung cancer, Pulmonary nodule classification, Stacked denoising autoencoder",
author = "Kim, {Bum Chae} and Sung, {Yu Sub} and Heung-Il Suk",
year = "2016",
month = "4",
day = "20",
doi = "10.1109/IWW-BCI.2016.7457462",
language = "English",
series = "4th International Winter Conference on Brain-Computer Interface, BCI 2016",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "4th International Winter Conference on Brain-Computer Interface, BCI 2016",

}

TY - GEN

T1 - Deep feature learning for pulmonary nodule classification in a lung CT

AU - Kim, Bum Chae

AU - Sung, Yu Sub

AU - Suk, Heung-Il

PY - 2016/4/20

Y1 - 2016/4/20

N2 - In this paper, we propose a novel method of identifying pulmonary nodules in a lung CT. Specifically, we devise a deep neural network by which we extract abstract information inherent in raw hand-crafted imaging features. We then combine the deep learned representations with the original raw imaging features into a long feature vector. By taking the combined feature vectors, we train a classifier, preceded by a feature selection via t-test. To validate the effectiveness of the proposed method, we performed experiments on our in-house dataset of 20 subjects; 3,598 pulmonary nodules (malignant: 178, benign: 3,420), which were manually segmented by a radiologist. In our experiments, we achieved the maximal accuracy of 95.5%, sensitivity of 94.4%, and AUC of 0.987, outperforming the competing method.

AB - In this paper, we propose a novel method of identifying pulmonary nodules in a lung CT. Specifically, we devise a deep neural network by which we extract abstract information inherent in raw hand-crafted imaging features. We then combine the deep learned representations with the original raw imaging features into a long feature vector. By taking the combined feature vectors, we train a classifier, preceded by a feature selection via t-test. To validate the effectiveness of the proposed method, we performed experiments on our in-house dataset of 20 subjects; 3,598 pulmonary nodules (malignant: 178, benign: 3,420), which were manually segmented by a radiologist. In our experiments, we achieved the maximal accuracy of 95.5%, sensitivity of 94.4%, and AUC of 0.987, outperforming the competing method.

KW - Deep learning

KW - Lung cancer

KW - Pulmonary nodule classification

KW - Stacked denoising autoencoder

UR - http://www.scopus.com/inward/record.url?scp=84969175371&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84969175371&partnerID=8YFLogxK

U2 - 10.1109/IWW-BCI.2016.7457462

DO - 10.1109/IWW-BCI.2016.7457462

M3 - Conference contribution

AN - SCOPUS:84969175371

T3 - 4th International Winter Conference on Brain-Computer Interface, BCI 2016

BT - 4th International Winter Conference on Brain-Computer Interface, BCI 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -