Recent approaches for surgical activity localization rely on motion features derived from the optical flow (OF). However, although they consider state-of-the-art CNNs when computing the OF, they typically resort to pre-trained implementations which are domain-unaware. We address this problem in two ways: (i) Using the pre-trained OF-CNN of recent localization approach, we analyze the impact of video properties such as reflections, motion and blur on the quality of the OF from neurosurgical data. (ii) Based on this analysis, we design a specifically tailored synthetic training dataset which allows us to customize the pre-trained OF-CNN for surgical activity localization. Our evaluation clearly shows the benefit of this customization approach. It not only leads to an improved accuracy of the OF itself but, even more importantly, also to an improved performance for the actual localization task.
%0 Conference Paper
%1 Philipp_2022
%A Philipp, Markus
%A Bacher, Neal
%A Sauer, Stefan
%A Mathis-Ullrich, Franziska
%A Bruhn, Andrés
%B Proceedings of the IEEE International Symposium on Biomedical Imaging (ISBI)
%D 2022
%I IEEE
%K 2022 b04 from:christinawarren sfbtrr161
%P 1–5
%R 10.1109/ISBI52829.2022.9761704
%T From Chairs To Brains: Customizing Optical Flow For Surgical Activity Localization
%U https://ieeexplore.ieee.org/document/9761704
%X Recent approaches for surgical activity localization rely on motion features derived from the optical flow (OF). However, although they consider state-of-the-art CNNs when computing the OF, they typically resort to pre-trained implementations which are domain-unaware. We address this problem in two ways: (i) Using the pre-trained OF-CNN of recent localization approach, we analyze the impact of video properties such as reflections, motion and blur on the quality of the OF from neurosurgical data. (ii) Based on this analysis, we design a specifically tailored synthetic training dataset which allows us to customize the pre-trained OF-CNN for surgical activity localization. Our evaluation clearly shows the benefit of this customization approach. It not only leads to an improved accuracy of the OF itself but, even more importantly, also to an improved performance for the actual localization task.
@inproceedings{Philipp_2022,
abstract = {Recent approaches for surgical activity localization rely on motion features derived from the optical flow (OF). However, although they consider state-of-the-art CNNs when computing the OF, they typically resort to pre-trained implementations which are domain-unaware. We address this problem in two ways: (i) Using the pre-trained OF-CNN of recent localization approach, we analyze the impact of video properties such as reflections, motion and blur on the quality of the OF from neurosurgical data. (ii) Based on this analysis, we design a specifically tailored synthetic training dataset which allows us to customize the pre-trained OF-CNN for surgical activity localization. Our evaluation clearly shows the benefit of this customization approach. It not only leads to an improved accuracy of the OF itself but, even more importantly, also to an improved performance for the actual localization task.},
added-at = {2022-05-17T09:28:43.000+0200},
author = {Philipp, Markus and Bacher, Neal and Sauer, Stefan and Mathis-Ullrich, Franziska and Bruhn, Andrés},
biburl = {https://puma.ub.uni-stuttgart.de/bibtex/2ae97eaa996db1533f95bed3151511f53/sfbtrr161},
booktitle = {Proceedings of the IEEE International Symposium on Biomedical Imaging (ISBI)},
doi = {10.1109/ISBI52829.2022.9761704},
interhash = {9f7a29d0b176e2a878bf612b5554af04},
intrahash = {ae97eaa996db1533f95bed3151511f53},
keywords = {2022 b04 from:christinawarren sfbtrr161},
month = {3},
pages = {1–5},
publisher = {IEEE},
timestamp = {2022-08-12T07:53:13.000+0200},
title = {From Chairs To Brains: Customizing Optical Flow For Surgical Activity Localization},
url = {https://ieeexplore.ieee.org/document/9761704},
year = 2022
}