TY - GEN
T1 - Guided Unsupervised Desmoking of Laparoscopic Images Using Cycle-Desmoke
AU - Vishal, V.
AU - Sharma, Neeraj
AU - Singh, Munendra
N1 - Publisher Copyright:
© 2019, Springer Nature Switzerland AG.
PY - 2019/1/1
Y1 - 2019/1/1
N2 - The generation of smoke in laparoscopic surgery due to laser ablation and cauterization causes deterioration in the visual quality of the operative field. In order to reduce the effect of smoke, the present paper proposes an end-to-end network, called Cycle-Desmoke. The network enhances the CycleGAN framework by adoption of a new generator architecture and addition of new Guided-Unsharp Upsample loss in combination to adversarial and cycle-consistency loss. The Atrous Convolution Feature Extraction Module present in the encoder blocks of the generator helps distinguishing smoke by capturing features at multiple scales by the use of kernels with different receptive fields. Further, the use of Guided-Unsharp Upsample loss supervises the upsampling process of the feature maps and helps improve the contrast of the desmoked image. The network performs robust unsupervised Image-to-Image Translation from smoke domain to smoke-free domain. The public Cholec80 dataset is used to evaluate the performance of the proposed method. Quantitative and qualitative comparative analysis of the proposed method over the state-of-the-methods reveals the effectiveness of the method at the task of smoke removal and enhancement of the image.
AB - The generation of smoke in laparoscopic surgery due to laser ablation and cauterization causes deterioration in the visual quality of the operative field. In order to reduce the effect of smoke, the present paper proposes an end-to-end network, called Cycle-Desmoke. The network enhances the CycleGAN framework by adoption of a new generator architecture and addition of new Guided-Unsharp Upsample loss in combination to adversarial and cycle-consistency loss. The Atrous Convolution Feature Extraction Module present in the encoder blocks of the generator helps distinguishing smoke by capturing features at multiple scales by the use of kernels with different receptive fields. Further, the use of Guided-Unsharp Upsample loss supervises the upsampling process of the feature maps and helps improve the contrast of the desmoked image. The network performs robust unsupervised Image-to-Image Translation from smoke domain to smoke-free domain. The public Cholec80 dataset is used to evaluate the performance of the proposed method. Quantitative and qualitative comparative analysis of the proposed method over the state-of-the-methods reveals the effectiveness of the method at the task of smoke removal and enhancement of the image.
UR - https://www.scopus.com/pages/publications/85075569490
UR - https://www.scopus.com/pages/publications/85075569490#tab=citedBy
U2 - 10.1007/978-3-030-32695-1_3
DO - 10.1007/978-3-030-32695-1_3
M3 - Conference contribution
AN - SCOPUS:85075569490
SN - 9783030326944
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 21
EP - 28
BT - OR 2.0 Context-Aware Operating Theaters and Machine Learning in Clinical Neuroimaging - 2nd International Workshop, OR 2.0 2019, and 2nd International Workshop, MLCN 2019, Held in Conjunction with MICCAI 2019, Proceedings
A2 - Zhou, Luping
A2 - Sarikaya, Duygu
A2 - Kia, Seyed Mostafa
A2 - Speidel, Stefanie
A2 - Malpani, Anand
A2 - Hashimoto, Daniel
A2 - Habes, Mohamad
A2 - Löfstedt, Tommy
A2 - Ritter, Kerstin
A2 - Wang, Hongzhi
PB - Springer Paris
T2 - 2nd International Workshop on Context-Aware Surgical Theaters, OR 2.0 2019, and the 2nd International Workshop on Machine Learning in Clinical Neuroimaging, MLCN 2019, held in conjunction with the 22nd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2019
Y2 - 17 October 2019 through 17 October 2019
ER -