2018³â ÄÄÇ»ÅÍÁ¾ÇÕÇмú´ëȸ
Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
ÀûÀÀÀû U-NetÀ» ÀÌ¿ëÇÑ X-·¹ÀÌ ¿µ»óÀÇ »À ¿µ¿ª ºÐÇÒ |
¿µ¹®Á¦¸ñ(English Title) |
Bone Area Segmentation in X-Ray images using an Adaptable U-Net |
ÀúÀÚ(Author) |
È£³áÈØ
Æ®¶õµò½ã
ÁÖÁ¾¹Î
¾çÇüÁ¤
±è¼öÇü
Á¤¼ºÅÃ
ÁÖ»óµ·
Ngoc-Huynh Ho
Dinh-Son Tran
Jong-Min Joo
Hyung-Jeong Yang
Soo-Hyung Kim
Sung-Taek Jung
Sang-Don Joo
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 45 NO. 01 PP. 0849 ~ 0851 (2018. 06) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Segmentation of human bone from X-ray images has evolved as a tool for the diagnosis of bone tumor. Segmentation of a snapshot is the hardest and most complex task because there are proprietary objects and a huge variation between them using a common framework. In this paper, we have introduced a fully automatic and robust method, entitled the adaptable U-Net (aU-Net) applied in X-ray bone segmentation. This method is based on convolution neural networks (CNNs), which achieved state-of-the-art performance for automatic medical image segmentation. The experimental result show that 1) the aU-Net is more robust to segment previously unseen objects than the traditional methods, 2) the aU-Net outperforms to other conventional methods in terms of the image segmentation loss.
|
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|