TY - GEN
T1 - Self-supervised learning for dense depth estimation in monocular endoscopy
AU - Liu, Xingtong
AU - Sinha, Ayushi
AU - Unberath, Mathias
AU - Ishii, Masaru
AU - Hager, Gregory D.
AU - Taylor, Russell H.
AU - Reiter, Austin
N1 - Funding Information:
Acknowledgement. The work reported in this paper was funded in part by NIH R01-EB015530, in part by a research contract from Galen Robotics, and in part by Johns Hopkins University internal funds.
Publisher Copyright:
© Springer Nature Switzerland AG 2018.
PY - 2018
Y1 - 2018
N2 - We present a self-supervised approach to training convolutional neural networks for dense depth estimation from monocular endoscopy data without a priori modeling of anatomy or shading. Our method only requires sequential data from monocular endoscopic videos and a multi-view stereo reconstruction method, e.g. structure from motion, that supervises learning in a sparse but accurate manner. Consequently, our method requires neither manual interaction, such as scaling or labeling, nor patient CT in the training and application phases. We demonstrate the performance of our method on sinus endoscopy data from two patients and validate depth prediction quantitatively using corresponding patient CT scans where we found submillimeter residual errors. (Link to the supplementary video: https://camp.lcsr.jhu.edu/miccai-2018-demonstration-videos/).
AB - We present a self-supervised approach to training convolutional neural networks for dense depth estimation from monocular endoscopy data without a priori modeling of anatomy or shading. Our method only requires sequential data from monocular endoscopic videos and a multi-view stereo reconstruction method, e.g. structure from motion, that supervises learning in a sparse but accurate manner. Consequently, our method requires neither manual interaction, such as scaling or labeling, nor patient CT in the training and application phases. We demonstrate the performance of our method on sinus endoscopy data from two patients and validate depth prediction quantitatively using corresponding patient CT scans where we found submillimeter residual errors. (Link to the supplementary video: https://camp.lcsr.jhu.edu/miccai-2018-demonstration-videos/).
UR - http://www.scopus.com/inward/record.url?scp=85054850438&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85054850438&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-01201-4_15
DO - 10.1007/978-3-030-01201-4_15
M3 - Conference contribution
AN - SCOPUS:85054850438
SN - 9783030012007
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 128
EP - 138
BT - OR 2.0 Context-Aware Operating Theaters, Computer Assisted Robotic Endoscopy, Clinical Image-Based Procedures, and Skin Image Analysis - 1st International Workshop, OR 2.0 2018 5th International Workshop, CARE 2018, 7th International Workshop, CLIP 2018, 3rd International Workshop, ISIC 2018 Held in Conjunction with MICCAI 2018
A2 - Malpani, Anand
A2 - Zenati, Marco A.
A2 - Oyarzun Laura, Cristina
A2 - Celebi, M. Emre
A2 - Sarikaya, Duygu
A2 - Codella, Noel C.
A2 - Halpern, Allan
A2 - Erdt, Marius
A2 - Maier-Hein, Lena
A2 - Xiongbiao, Luo
A2 - Wesarg, Stefan
A2 - Stoyanov, Danail
A2 - Taylor, Zeike
A2 - Drechsler, Klaus
A2 - Dana, Kristin
A2 - Martel, Anne
A2 - Shekhar, Raj
A2 - De Ribaupierre, Sandrine
A2 - Reichl, Tobias
A2 - McLeod, Jonathan
A2 - González Ballester, Miguel Angel
A2 - Collins, Toby
A2 - Linguraru, Marius George
PB - Springer Verlag
T2 - 1st International Workshop on OR 2.0 Context-Aware Operating Theaters, OR 2.0 2018, 5th International Workshop on Computer Assisted Robotic Endoscopy, CARE 2018, 7th International Workshop on Clinical Image-Based Procedures, CLIP 2018, and 1st International Workshop on Skin Image Analysis, ISIC 2018, held in conjunction with the 21st International Conference on Medical Imaging and Computer-Assisted Intervention, MICCAI 2018
Y2 - 16 September 2018 through 20 September 2018
ER -