Standard
Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation. / Xu, Peidi; Lee, Blaire; Sosnovtseva, Olga; Sørensen, Charlotte Mehlin; Erleben, Kenny; Darkner, Sune.
Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. red. / Zhiyun Xue; Sameer Antani; Ghada Zamzmi; Feng Yang; Sivaramakrishnan Rajaraman; Zhaohui Liang; Sharon Xiaolei Huang; Marius George Linguraru. Springer, 2023. s. 191-201 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Bind 14307 LNCS).
Publikation: Bidrag til bog/antologi/rapport › Konferencebidrag i proceedings › Forskning › fagfællebedømt
Harvard
Xu, P, Lee, B
, Sosnovtseva, O, Sørensen, CM, Erleben, K & Darkner, S 2023,
Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation. i Z Xue, S Antani, G Zamzmi, F Yang, S Rajaraman, Z Liang, SX Huang & MG Linguraru (red),
Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. Springer, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), bind 14307 LNCS, s. 191-201, 2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023, Vancouver, Canada,
08/10/2023.
https://doi.org/10.1007/978-3-031-44917-8_18
APA
Xu, P., Lee, B.
, Sosnovtseva, O., Sørensen, C. M., Erleben, K., & Darkner, S. (2023).
Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation. I Z. Xue, S. Antani, G. Zamzmi, F. Yang, S. Rajaraman, Z. Liang, S. X. Huang, & M. G. Linguraru (red.),
Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings (s. 191-201). Springer. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) Bind 14307 LNCS
https://doi.org/10.1007/978-3-031-44917-8_18
Vancouver
Xu P, Lee B
, Sosnovtseva O, Sørensen CM, Erleben K, Darkner S.
Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation. I Xue Z, Antani S, Zamzmi G, Yang F, Rajaraman S, Liang Z, Huang SX, Linguraru MG, red., Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. Springer. 2023. s. 191-201. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Bind 14307 LNCS).
https://doi.org/10.1007/978-3-031-44917-8_18
Author
Xu, Peidi ; Lee, Blaire ; Sosnovtseva, Olga ; Sørensen, Charlotte Mehlin ; Erleben, Kenny ; Darkner, Sune. / Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation. Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings. red. / Zhiyun Xue ; Sameer Antani ; Ghada Zamzmi ; Feng Yang ; Sivaramakrishnan Rajaraman ; Zhaohui Liang ; Sharon Xiaolei Huang ; Marius George Linguraru. Springer, 2023. s. 191-201 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Bind 14307 LNCS).
Bibtex
@inproceedings{fc1a7497383548028ee7c44dc88dc459,
title = "Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation",
abstract = "Accurate analysis and modeling of renal functions require a precise segmentation of the renal blood vessels. Micro-CT scans provide image data at higher resolutions, making deeper vessels near the renal cortex visible. Although deep-learning-based methods have shown state-of-the-art performance in automatic blood vessel segmentations, they require a large amount of labeled training data. However, voxel-wise labeling in micro-CT scans is extremely time-consuming, given the huge volume sizes. To mitigate the problem, we simulate synthetic renal vascular trees physiologically while generating corresponding scans of the simulated trees by training a generative model on unlabeled scans. This enables the generative model to learn the mapping implicitly without the need for explicit functions to emulate the image acquisition process. We further propose an additional segmentation branch over the generative model trained on the generated scans. We demonstrate that the model can directly segment blood vessels on real scans and validate our method on both 3D micro-CT scans of rat kidneys and a proof-of-concept experiment on 2D retinal images. Code and 3D results are available at (https://github.com/diku-dk/RenalVesselSeg ).",
keywords = "Blood vessel, Domain adaptation, Generative model, Physiological simulation, Renal vasculature, Semantic segmentation",
author = "Peidi Xu and Blaire Lee and Olga Sosnovtseva and S{\o}rensen, {Charlotte Mehlin} and Kenny Erleben and Sune Darkner",
note = "Publisher Copyright: {\textcopyright} 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.; 2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023 ; Conference date: 08-10-2023 Through 08-10-2023",
year = "2023",
doi = "10.1007/978-3-031-44917-8_18",
language = "English",
isbn = "9783031471964",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer",
pages = "191--201",
editor = "Zhiyun Xue and Sameer Antani and Ghada Zamzmi and Feng Yang and Sivaramakrishnan Rajaraman and Zhaohui Liang and Huang, {Sharon Xiaolei} and Linguraru, {Marius George}",
booktitle = "Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings",
address = "Switzerland",
}
RIS
TY - GEN
T1 - Extremely Weakly-Supervised Blood Vessel Segmentation with Physiologically Based Synthesis and Domain Adaptation
AU - Xu, Peidi
AU - Lee, Blaire
AU - Sosnovtseva, Olga
AU - Sørensen, Charlotte Mehlin
AU - Erleben, Kenny
AU - Darkner, Sune
N1 - Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - Accurate analysis and modeling of renal functions require a precise segmentation of the renal blood vessels. Micro-CT scans provide image data at higher resolutions, making deeper vessels near the renal cortex visible. Although deep-learning-based methods have shown state-of-the-art performance in automatic blood vessel segmentations, they require a large amount of labeled training data. However, voxel-wise labeling in micro-CT scans is extremely time-consuming, given the huge volume sizes. To mitigate the problem, we simulate synthetic renal vascular trees physiologically while generating corresponding scans of the simulated trees by training a generative model on unlabeled scans. This enables the generative model to learn the mapping implicitly without the need for explicit functions to emulate the image acquisition process. We further propose an additional segmentation branch over the generative model trained on the generated scans. We demonstrate that the model can directly segment blood vessels on real scans and validate our method on both 3D micro-CT scans of rat kidneys and a proof-of-concept experiment on 2D retinal images. Code and 3D results are available at (https://github.com/diku-dk/RenalVesselSeg ).
AB - Accurate analysis and modeling of renal functions require a precise segmentation of the renal blood vessels. Micro-CT scans provide image data at higher resolutions, making deeper vessels near the renal cortex visible. Although deep-learning-based methods have shown state-of-the-art performance in automatic blood vessel segmentations, they require a large amount of labeled training data. However, voxel-wise labeling in micro-CT scans is extremely time-consuming, given the huge volume sizes. To mitigate the problem, we simulate synthetic renal vascular trees physiologically while generating corresponding scans of the simulated trees by training a generative model on unlabeled scans. This enables the generative model to learn the mapping implicitly without the need for explicit functions to emulate the image acquisition process. We further propose an additional segmentation branch over the generative model trained on the generated scans. We demonstrate that the model can directly segment blood vessels on real scans and validate our method on both 3D micro-CT scans of rat kidneys and a proof-of-concept experiment on 2D retinal images. Code and 3D results are available at (https://github.com/diku-dk/RenalVesselSeg ).
KW - Blood vessel
KW - Domain adaptation
KW - Generative model
KW - Physiological simulation
KW - Renal vasculature
KW - Semantic segmentation
U2 - 10.1007/978-3-031-44917-8_18
DO - 10.1007/978-3-031-44917-8_18
M3 - Article in proceedings
AN - SCOPUS:85174743869
SN - 9783031471964
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 191
EP - 201
BT - Medical Image Learning with Limited and Noisy Data - 2nd International Workshop, MILLanD 2023, Held in Conjunction with MICCAI 2023, Proceedings
A2 - Xue, Zhiyun
A2 - Antani, Sameer
A2 - Zamzmi, Ghada
A2 - Yang, Feng
A2 - Rajaraman, Sivaramakrishnan
A2 - Liang, Zhaohui
A2 - Huang, Sharon Xiaolei
A2 - Linguraru, Marius George
PB - Springer
T2 - 2nd Workshop on Medical Image Learning with Noisy and Limited Data, MILLanD 2023
Y2 - 8 October 2023 through 8 October 2023
ER -
