Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images

So Hyun Byun, Julip Jung, Helen Hong, Yong Sub Song, Hyungjin Kim, Chang Min Park

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

Abstract

The malignancy rate of GGN is different according to the presence and the size of a solid component. Thus, it is important to differentiate part-solid GGN with a variable sized solid component from pure GGN. In this paper, we propose a method of classifying the GGNs according to presence or size of solid component using multiple 2.5- dimensional deep CNNs. First, to consider not only intensity but also texture, and shape information, we propose an enhanced input image using image augmentation and removing background. Second, we proposed GGN-Net which can classify GGNs into three classes using multiple input images in chest CT images. Finally, we comparatively evaluate the classification performance according to different type of input images. In experiments, the accuracy of the proposed method using multiple input images was the highest at 82.76% and it was 10.35%, 13.79%, and 6.90% higher than that of using three single input image such as intensity-based, texture- and shape-enhanced input images, respectively.

Original languageEnglish
Title of host publicationInternational Forum on Medical Imaging in Asia 2019
EditorsHiroshi Fujita, Feng Lin, Jong Hyo Kim
PublisherSPIE
ISBN (Electronic)9781510627758
DOIs
StatePublished - 1 Jan 2019
EventInternational Forum on Medical Imaging in Asia 2019 - Singapore, Singapore
Duration: 7 Jan 20199 Jan 2019

Publication series

NameProceedings of SPIE - The International Society for Optical Engineering
Volume11050
ISSN (Print)0277-786X
ISSN (Electronic)1996-756X

Conference

ConferenceInternational Forum on Medical Imaging in Asia 2019
CountrySingapore
CitySingapore
Period7/01/199/01/19

Fingerprint

CT Image
Nodule
chest
nodules
Neural Networks
Neural networks
Glass
glass
Textures
Texture
textures
Augmentation
Differentiate
classifying
Classify
Experiments
Evaluate
augmentation
Experiment

Keywords

  • Chest CT image
  • deep convolutional neural network
  • ground-glass nodule

Cite this

Byun, S. H., Jung, J., Hong, H., Song, Y. S., Kim, H., & Park, C. M. (2019). Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images. In H. Fujita, F. Lin, & J. H. Kim (Eds.), International Forum on Medical Imaging in Asia 2019 [110500O] (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 11050). SPIE. https://doi.org/10.1117/12.2523715
Byun, So Hyun ; Jung, Julip ; Hong, Helen ; Song, Yong Sub ; Kim, Hyungjin ; Park, Chang Min. / Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images. International Forum on Medical Imaging in Asia 2019. editor / Hiroshi Fujita ; Feng Lin ; Jong Hyo Kim. SPIE, 2019. (Proceedings of SPIE - The International Society for Optical Engineering).
@inproceedings{80ecc1bda14e4ce4af067d6ab7fc8139,
title = "Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images",
abstract = "The malignancy rate of GGN is different according to the presence and the size of a solid component. Thus, it is important to differentiate part-solid GGN with a variable sized solid component from pure GGN. In this paper, we propose a method of classifying the GGNs according to presence or size of solid component using multiple 2.5- dimensional deep CNNs. First, to consider not only intensity but also texture, and shape information, we propose an enhanced input image using image augmentation and removing background. Second, we proposed GGN-Net which can classify GGNs into three classes using multiple input images in chest CT images. Finally, we comparatively evaluate the classification performance according to different type of input images. In experiments, the accuracy of the proposed method using multiple input images was the highest at 82.76{\%} and it was 10.35{\%}, 13.79{\%}, and 6.90{\%} higher than that of using three single input image such as intensity-based, texture- and shape-enhanced input images, respectively.",
keywords = "Chest CT image, deep convolutional neural network, ground-glass nodule",
author = "Byun, {So Hyun} and Julip Jung and Helen Hong and Song, {Yong Sub} and Hyungjin Kim and Park, {Chang Min}",
year = "2019",
month = "1",
day = "1",
doi = "10.1117/12.2523715",
language = "English",
series = "Proceedings of SPIE - The International Society for Optical Engineering",
publisher = "SPIE",
editor = "Hiroshi Fujita and Feng Lin and Kim, {Jong Hyo}",
booktitle = "International Forum on Medical Imaging in Asia 2019",

}

Byun, SH, Jung, J, Hong, H, Song, YS, Kim, H & Park, CM 2019, Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images. in H Fujita, F Lin & JH Kim (eds), International Forum on Medical Imaging in Asia 2019., 110500O, Proceedings of SPIE - The International Society for Optical Engineering, vol. 11050, SPIE, International Forum on Medical Imaging in Asia 2019, Singapore, Singapore, 7/01/19. https://doi.org/10.1117/12.2523715

Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images. / Byun, So Hyun; Jung, Julip; Hong, Helen; Song, Yong Sub; Kim, Hyungjin; Park, Chang Min.

International Forum on Medical Imaging in Asia 2019. ed. / Hiroshi Fujita; Feng Lin; Jong Hyo Kim. SPIE, 2019. 110500O (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 11050).

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

TY - GEN

T1 - Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images

AU - Byun, So Hyun

AU - Jung, Julip

AU - Hong, Helen

AU - Song, Yong Sub

AU - Kim, Hyungjin

AU - Park, Chang Min

PY - 2019/1/1

Y1 - 2019/1/1

N2 - The malignancy rate of GGN is different according to the presence and the size of a solid component. Thus, it is important to differentiate part-solid GGN with a variable sized solid component from pure GGN. In this paper, we propose a method of classifying the GGNs according to presence or size of solid component using multiple 2.5- dimensional deep CNNs. First, to consider not only intensity but also texture, and shape information, we propose an enhanced input image using image augmentation and removing background. Second, we proposed GGN-Net which can classify GGNs into three classes using multiple input images in chest CT images. Finally, we comparatively evaluate the classification performance according to different type of input images. In experiments, the accuracy of the proposed method using multiple input images was the highest at 82.76% and it was 10.35%, 13.79%, and 6.90% higher than that of using three single input image such as intensity-based, texture- and shape-enhanced input images, respectively.

AB - The malignancy rate of GGN is different according to the presence and the size of a solid component. Thus, it is important to differentiate part-solid GGN with a variable sized solid component from pure GGN. In this paper, we propose a method of classifying the GGNs according to presence or size of solid component using multiple 2.5- dimensional deep CNNs. First, to consider not only intensity but also texture, and shape information, we propose an enhanced input image using image augmentation and removing background. Second, we proposed GGN-Net which can classify GGNs into three classes using multiple input images in chest CT images. Finally, we comparatively evaluate the classification performance according to different type of input images. In experiments, the accuracy of the proposed method using multiple input images was the highest at 82.76% and it was 10.35%, 13.79%, and 6.90% higher than that of using three single input image such as intensity-based, texture- and shape-enhanced input images, respectively.

KW - Chest CT image

KW - deep convolutional neural network

KW - ground-glass nodule

UR - http://www.scopus.com/inward/record.url?scp=85063907526&partnerID=8YFLogxK

U2 - 10.1117/12.2523715

DO - 10.1117/12.2523715

M3 - Conference contribution

T3 - Proceedings of SPIE - The International Society for Optical Engineering

BT - International Forum on Medical Imaging in Asia 2019

A2 - Fujita, Hiroshi

A2 - Lin, Feng

A2 - Kim, Jong Hyo

PB - SPIE

ER -

Byun SH, Jung J, Hong H, Song YS, Kim H, Park CM. Ground-glass nodule classification with multiple 2.5-dimensional deep convolutional neural networks in chest CT images. In Fujita H, Lin F, Kim JH, editors, International Forum on Medical Imaging in Asia 2019. SPIE. 2019. 110500O. (Proceedings of SPIE - The International Society for Optical Engineering). https://doi.org/10.1117/12.2523715