TY - GEN
T1 - CNN in CT Image Segmentation
T2 - 17th IEEE International Symposium on Biomedical Imaging, ISBI 2020
AU - Song, Youyi
AU - Yu, Zhen
AU - Zhou, Teng
AU - Teoh, Jeremy Yuen Chun
AU - Lei, Baiying
AU - Choi, Kup Sze
AU - Qin, Jing
N1 - Funding Information:
The work described in this paper is supported by a grant from the Hong Kong Research Grants Council (No. PolyU 152035/17E).
Publisher Copyright:
© 2020 IEEE.
PY - 2020/4
Y1 - 2020/4
N2 - Exploiting more information from ground truth (GT) images now is a new research direction for further improving CNN's performance in CT image segmentation. Previous methods focus on devising the loss function for fulfilling such a purpose. However, it is rather difficult to devise a general and optimization-friendly loss function. We here present a novel and practical method that exploits GT images beyond the loss function. Our insight is that feature maps of two CNNs trained respectively on GT and CT images should be similar on some metric space, because they both are used to describe the same objects for the same purpose. We hence exploit GT images by enforcing such two CNNs' feature maps to be consistent. We assess the proposed method on two data sets, and compare its performance to several competitive methods. Extensive experimental results show that the proposed method is effective, outperforming all the compared methods.
AB - Exploiting more information from ground truth (GT) images now is a new research direction for further improving CNN's performance in CT image segmentation. Previous methods focus on devising the loss function for fulfilling such a purpose. However, it is rather difficult to devise a general and optimization-friendly loss function. We here present a novel and practical method that exploits GT images beyond the loss function. Our insight is that feature maps of two CNNs trained respectively on GT and CT images should be similar on some metric space, because they both are used to describe the same objects for the same purpose. We hence exploit GT images by enforcing such two CNNs' feature maps to be consistent. We assess the proposed method on two data sets, and compare its performance to several competitive methods. Extensive experimental results show that the proposed method is effective, outperforming all the compared methods.
KW - CNN
KW - CT image segmentation
KW - ground truth image exploitation
KW - network transfer
UR - http://www.scopus.com/inward/record.url?scp=85081093565&partnerID=8YFLogxK
U2 - 10.1109/ISBI45749.2020.9098488
DO - 10.1109/ISBI45749.2020.9098488
M3 - Conference article published in proceeding or book
AN - SCOPUS:85081093565
T3 - Proceedings - International Symposium on Biomedical Imaging
SP - 325
EP - 328
BT - ISBI 2020 - 2020 IEEE International Symposium on Biomedical Imaging
PB - IEEE Computer Society
Y2 - 3 April 2020 through 7 April 2020
ER -