Infinite Brain MR Images: PGGAN-Based Data Augmentation for Tumor Detection

Changhee Han, Leonardo Rundo, Ryosuke Araki, Yujiro Furukawa, Giancarlo Mauri, Hideki Nakayama, Hideaki Hayashi

研究成果: 著書/レポートタイプへの貢献

4 引用 (Scopus)

抄録

Due to the lack of available annotated medical images, accurate computer-assisted diagnosis requires intensive data augmentation (DA) techniques, such as geometric/intensity transformations of original images; however, those transformed images intrinsically have a similar distribution to the original ones, leading to limited performance improvement. To fill the data lack in the real image distribution, we synthesize brain contrast-enhanced magnetic resonance (MR) images—realistic but completely different from the original ones—using generative adversarial networks (GANs). This study exploits progressive growing of GANs (PGGANs), a multistage generative training method, to generate original-sized 256 × 256 MR images for convolutional neural network-based brain tumor detection, which is challenging via conventional GANs; difficulties arise due to unstable GAN training with high resolution and a variety of tumors in size, location, shape, and contrast. Our preliminary results show that this novel PGGAN-based DA method can achieve a promising performance improvement, when combined with classical DA, in tumor detection and also in other medical imaging tasks.

元の言語英語
ホスト出版物のタイトルSmart Innovation, Systems and Technologies
出版者Springer Science and Business Media Deutschland GmbH
ページ291-303
ページ数13
DOI
出版物ステータス出版済み - 1 1 2020

出版物シリーズ

名前Smart Innovation, Systems and Technologies
151
ISSN(印刷物)2190-3018
ISSN(電子版)2190-3026

Fingerprint

Magnetic resonance
Tumors
Brain
Medical imaging
Neural networks
Data augmentation
Tumor
Performance improvement

All Science Journal Classification (ASJC) codes

  • Decision Sciences(all)
  • Computer Science(all)

これを引用

Han, C., Rundo, L., Araki, R., Furukawa, Y., Mauri, G., Nakayama, H., & Hayashi, H. (2020). Infinite Brain MR Images: PGGAN-Based Data Augmentation for Tumor Detection. : Smart Innovation, Systems and Technologies (pp. 291-303). (Smart Innovation, Systems and Technologies; 巻数 151). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-13-8950-4_27

Infinite Brain MR Images : PGGAN-Based Data Augmentation for Tumor Detection. / Han, Changhee; Rundo, Leonardo; Araki, Ryosuke; Furukawa, Yujiro; Mauri, Giancarlo; Nakayama, Hideki; Hayashi, Hideaki.

Smart Innovation, Systems and Technologies. Springer Science and Business Media Deutschland GmbH, 2020. p. 291-303 (Smart Innovation, Systems and Technologies; 巻 151).

研究成果: 著書/レポートタイプへの貢献

Han, C, Rundo, L, Araki, R, Furukawa, Y, Mauri, G, Nakayama, H & Hayashi, H 2020, Infinite Brain MR Images: PGGAN-Based Data Augmentation for Tumor Detection. : Smart Innovation, Systems and Technologies. Smart Innovation, Systems and Technologies, 巻. 151, Springer Science and Business Media Deutschland GmbH, pp. 291-303. https://doi.org/10.1007/978-981-13-8950-4_27
Han C, Rundo L, Araki R, Furukawa Y, Mauri G, Nakayama H その他. Infinite Brain MR Images: PGGAN-Based Data Augmentation for Tumor Detection. : Smart Innovation, Systems and Technologies. Springer Science and Business Media Deutschland GmbH. 2020. p. 291-303. (Smart Innovation, Systems and Technologies). https://doi.org/10.1007/978-981-13-8950-4_27
Han, Changhee ; Rundo, Leonardo ; Araki, Ryosuke ; Furukawa, Yujiro ; Mauri, Giancarlo ; Nakayama, Hideki ; Hayashi, Hideaki. / Infinite Brain MR Images : PGGAN-Based Data Augmentation for Tumor Detection. Smart Innovation, Systems and Technologies. Springer Science and Business Media Deutschland GmbH, 2020. pp. 291-303 (Smart Innovation, Systems and Technologies).
@inbook{c869e55708bb4ac2810a06cdc173193f,
title = "Infinite Brain MR Images: PGGAN-Based Data Augmentation for Tumor Detection",
abstract = "Due to the lack of available annotated medical images, accurate computer-assisted diagnosis requires intensive data augmentation (DA) techniques, such as geometric/intensity transformations of original images; however, those transformed images intrinsically have a similar distribution to the original ones, leading to limited performance improvement. To fill the data lack in the real image distribution, we synthesize brain contrast-enhanced magnetic resonance (MR) images—realistic but completely different from the original ones—using generative adversarial networks (GANs). This study exploits progressive growing of GANs (PGGANs), a multistage generative training method, to generate original-sized 256 × 256 MR images for convolutional neural network-based brain tumor detection, which is challenging via conventional GANs; difficulties arise due to unstable GAN training with high resolution and a variety of tumors in size, location, shape, and contrast. Our preliminary results show that this novel PGGAN-based DA method can achieve a promising performance improvement, when combined with classical DA, in tumor detection and also in other medical imaging tasks.",
author = "Changhee Han and Leonardo Rundo and Ryosuke Araki and Yujiro Furukawa and Giancarlo Mauri and Hideki Nakayama and Hideaki Hayashi",
year = "2020",
month = "1",
day = "1",
doi = "10.1007/978-981-13-8950-4_27",
language = "English",
series = "Smart Innovation, Systems and Technologies",
publisher = "Springer Science and Business Media Deutschland GmbH",
pages = "291--303",
booktitle = "Smart Innovation, Systems and Technologies",
address = "Germany",

}

TY - CHAP

T1 - Infinite Brain MR Images

T2 - PGGAN-Based Data Augmentation for Tumor Detection

AU - Han, Changhee

AU - Rundo, Leonardo

AU - Araki, Ryosuke

AU - Furukawa, Yujiro

AU - Mauri, Giancarlo

AU - Nakayama, Hideki

AU - Hayashi, Hideaki

PY - 2020/1/1

Y1 - 2020/1/1

N2 - Due to the lack of available annotated medical images, accurate computer-assisted diagnosis requires intensive data augmentation (DA) techniques, such as geometric/intensity transformations of original images; however, those transformed images intrinsically have a similar distribution to the original ones, leading to limited performance improvement. To fill the data lack in the real image distribution, we synthesize brain contrast-enhanced magnetic resonance (MR) images—realistic but completely different from the original ones—using generative adversarial networks (GANs). This study exploits progressive growing of GANs (PGGANs), a multistage generative training method, to generate original-sized 256 × 256 MR images for convolutional neural network-based brain tumor detection, which is challenging via conventional GANs; difficulties arise due to unstable GAN training with high resolution and a variety of tumors in size, location, shape, and contrast. Our preliminary results show that this novel PGGAN-based DA method can achieve a promising performance improvement, when combined with classical DA, in tumor detection and also in other medical imaging tasks.

AB - Due to the lack of available annotated medical images, accurate computer-assisted diagnosis requires intensive data augmentation (DA) techniques, such as geometric/intensity transformations of original images; however, those transformed images intrinsically have a similar distribution to the original ones, leading to limited performance improvement. To fill the data lack in the real image distribution, we synthesize brain contrast-enhanced magnetic resonance (MR) images—realistic but completely different from the original ones—using generative adversarial networks (GANs). This study exploits progressive growing of GANs (PGGANs), a multistage generative training method, to generate original-sized 256 × 256 MR images for convolutional neural network-based brain tumor detection, which is challenging via conventional GANs; difficulties arise due to unstable GAN training with high resolution and a variety of tumors in size, location, shape, and contrast. Our preliminary results show that this novel PGGAN-based DA method can achieve a promising performance improvement, when combined with classical DA, in tumor detection and also in other medical imaging tasks.

UR - http://www.scopus.com/inward/record.url?scp=85073148386&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85073148386&partnerID=8YFLogxK

U2 - 10.1007/978-981-13-8950-4_27

DO - 10.1007/978-981-13-8950-4_27

M3 - Chapter

AN - SCOPUS:85073148386

T3 - Smart Innovation, Systems and Technologies

SP - 291

EP - 303

BT - Smart Innovation, Systems and Technologies

PB - Springer Science and Business Media Deutschland GmbH

ER -