講演抄録/キーワード |
講演名 |
2019-01-22 14:50
[ショートペーパー]Towards Annotating Less Medical Images:
-- PGGAN-based MR Image Augmentation for Brain Tumor Detection -- ○Changhee Han(UTokyo)・Hideaki Hayashi(Kyushu Univ.)・Leonardo Rundo(Univ. Cambridge)・Ryosuke Araki(Chubu Univ.)・Yudai Nagano(UTokyo)・Yujiro Furukawa(Kanto Rosai Hosp.)・Giancarlo Mauri(Univ. Milano-Bicocca)・Hideki Nakayama(UTokyo) MI2018-82 |
抄録 |
(和) |
How can we tackle the lack of available annotated medical image data through Data Augmentation (DA) techniques for accurate computer-assisted diagnosis? 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). Especially, we exploit Progressive Growing of GANs (PGGANs) to generate original-sized 256 × 256 brain MR images. Our results show that this novel PGGAN-based medical DA method can achieve better performance, when combined with classical DA and GAN-based refinement, in convolutional neural network-based tumor detection and also in other medical imaging tasks. |
(英) |
How can we tackle the lack of available annotated medical image data through Data Augmentation (DA) techniques for accurate computer-assisted diagnosis? 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). Especially, we exploit Progressive Growing of GANs (PGGANs) to generate original-sized 256 × 256 brain MR images. Our results show that this novel PGGAN-based medical DA method can achieve better performance, when combined with classical DA and GAN-based refinement, in convolutional neural network-based tumor detection and also in other medical imaging tasks. |
キーワード |
(和) |
Data Augmentation / Generative Adversarial Networks / Deep Learning / / / / / |
(英) |
Data Augmentation / Generative Adversarial Networks / Deep Learning / / / / / |
文献情報 |
信学技報, vol. 118, no. 412, MI2018-82, pp. 93-94, 2019年1月. |
資料番号 |
MI2018-82 |
発行日 |
2019-01-15 (MI) |
ISSN |
Online edition: ISSN 2432-6380 |
著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します.(許諾番号:10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
PDFダウンロード |
MI2018-82 |