Papers
Topics
Authors
Recent
Gemini 2.5 Flash
Gemini 2.5 Flash
153 tokens/sec
GPT-4o
7 tokens/sec
Gemini 2.5 Pro Pro
45 tokens/sec
o3 Pro
4 tokens/sec
GPT-4.1 Pro
38 tokens/sec
DeepSeek R1 via Azure Pro
28 tokens/sec
2000 character limit reached

Autonomous Catheterization with Open-source Simulator and Expert Trajectory (2401.09059v2)

Published 17 Jan 2024 in cs.RO and cs.CV

Abstract: Endovascular robots have been actively developed in both academia and industry. However, progress toward autonomous catheterization is often hampered by the widespread use of closed-source simulators and physical phantoms. Additionally, the acquisition of large-scale datasets for training machine learning algorithms with endovascular robots is usually infeasible due to expensive medical procedures. In this chapter, we introduce CathSim, the first open-source simulator for endovascular intervention to address these limitations. CathSim emphasizes real-time performance to enable rapid development and testing of learning algorithms. We validate CathSim against the real robot and show that our simulator can successfully mimic the behavior of the real robot. Based on CathSim, we develop a multimodal expert navigation network and demonstrate its effectiveness in downstream endovascular navigation tasks. The intensive experimental results suggest that CathSim has the potential to significantly accelerate research in the autonomous catheterization field. Our project is publicly available at https://github.com/airvlab/cathsim.

Definition Search Book Streamline Icon: https://streamlinehq.com
References (75)
  1. Toward a versatile robotic platform for fluoroscopy and mri-guided endovascular interventions: A pre-clinical study. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2019.
  2. Abien Fred Agarap. Deep learning using rectified linear units (relu). arXiv preprint arXiv:1803.08375, 2018.
  3. On evaluation of embodied navigation agents. arXiv preprint arXiv:1807.06757, 2018.
  4. Deep q reinforcement learning for autonomous navigation of surgical snake robot in confined spaces. In Proceedings of the The Hamlyn Symposium on Medical Robotics, London, UK, pp.  23–26, 2019.
  5. Deep Reinforcement Learning for the Navigation of Neurovascular Catheters. Current Directions in Biomedical Engineering, 5(1):5–8, September 2019. ISSN 2364-5504.
  6. Continuum robots for medical applications: A survey. IEEE Transactions on Robotics, 31(6):1261–1280, 2015.
  7. Trajectory optimization of robot-assisted endovascular catheterization with reinforcement learning. In 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp.  3875–3881. IEEE, 2018a.
  8. Learning-based endovascular navigation through the use of non-rigid registration for collaborative robotic catheterization. International Journal of Computer Assisted Radiology and Surgery, 13, 04 2018b.
  9. Collaborative robot-assisted endovascular catheterization with generative adversarial imitation learning. In 2020 IEEE International conference on robotics and automation (ICRA), pp.  2414–2420. IEEE, 2020.
  10. Image processing based autonomous guidewire navigation in percutaneous coronary intervention. In 2021 IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia), pp.  1–6. IEEE, 2021.
  11. In-vivo validation of a novel robotic platform for endovascular intervention. IEEE Transactions on Biomedical Engineering, 2022.
  12. Towards interactive planning of coil embolization in brain aneurysms. In Medical Image Computing and Computer Assisted Intervention-MICCAI 2009, volume 5761, pp.  377–385. Springer Berlin/Heidelberg, 2009.
  13. Edsger W Dijkstra. A note on two problems in connexion with graphs. Numerische mathematik, 1(1):269–271, 1959.
  14. CARLA: An open urban driving simulator. In Proceedings of the 1st Annual Conference on Robot Learning, pp.  1–16, 2017.
  15. A decade retrospective of medical robotics research from 2010 to 2020. Science Robotics, 6(60):eabi8017, 2021.
  16. Sofa: A multi-model framework for interactive physical simulation. In Soft Tissue Biomechanical Modeling for Computer Assisted Surgery. 2012. URL https://hal.inria.fr/hal-00681539.
  17. Roy Featherstone. Rigid body dynamics algorithms. Springer, 2014.
  18. Guided cost learning: Deep inverse optimal control via policy optimization. In International conference on machine learning, pp.  49–58. PMLR, 2016.
  19. Using magnetic fields to navigate and simultaneously localize catheters in endoluminal environments. IEEE Robotics and Automation Letters, 7(3):7217–7223, 2022.
  20. Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor. In International conference on machine learning, 2018.
  21. Generative adversarial imitation learning. Advances in neural information processing systems, 29, 2016.
  22. Learning to steer by mimicking features from heterogeneous auxiliary networks. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, pp.  8433–8440, 2019.
  23. Detecting the sensing area of a laparoscopic probe in minimally invasive cancer surgery. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pp.  260–270. Springer, 2023.
  24. Imitation learning: A survey of learning methods. ACM Computing Surveys (CSUR), 50(2):1–35, 2017.
  25. Complications of robotic surgery. Surgical Clinics, 100(2):461–468, 2020.
  26. An image-based guidewire navigation method for robot-assisted intravascular interventions. In 2011 Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp.  6680–6685. IEEE, 2011.
  27. 3d guidewire shape reconstruction from monoplane fluoroscopic images. arXiv preprint arXiv:2311.11209, 2023.
  28. 3d collision detection: a survey. Computers & Graphics, 25(2):269–285, 2001.
  29. Translating simulation images to x-ray images via multi-scale semantic matching. arXiv preprint arXiv:2304.07693, 2023.
  30. Autonomous guidewire navigation in a two dimensional vascular phantom. Current Directions in Biomedical Engineering, 6(1), 2020.
  31. Towards autonomous eye surgery by combining deep imitation learning with optimal control. In Conference on Robot Learning, pp.  2347–2358. PMLR, 2021.
  32. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014.
  33. Deep reinforcement learning for autonomous driving: A survey. IEEE Transactions on Intelligent Transportation Systems, 23(6):4909–4926, 2021.
  34. Shape-sensitive loss for catheter and guidewire segmentation. arXiv preprint arXiv:2311.11205, 2023.
  35. An mr-safe endovascular robotic platform: Design, control, and ex-vivo evaluation. IEEE Transactions on Biomedical Engineering, 68(10):3110–3121, 2021.
  36. Deep reinforcement learning for guidewire navigation in coronary artery phantom. IEEE Access, 2021.
  37. 3d perception based imitation learning under limited demonstration for laparoscope control in robotic surgery. In 2022 International Conference on Robotics and Automation (ICRA), pp.  7664–7670. IEEE, 2022.
  38. Rensis Likert. A technique for the measurement of attitudes. Archives of psychology, 1932.
  39. Continuous control with deep reinforcement learning. arXiv preprint arXiv:1509.02971, 2015.
  40. Feasibility and safety of robotic peripheral vascular interventions: results of the rapid trial. Cardiovascular Interventions, 9(19):2058–2064, 2016.
  41. A simple and efficient approach for 3d mesh approximate convex decomposition. In 2009 16th IEEE international conference on image processing (ICIP), 2009.
  42. Human-level control through deep reinforcement learning. nature, 518(7540):529–533, 2015.
  43. Haptic guidance for robot-assisted endovascular procedures: implementation and evaluation on surgical simulator. In IROS. IEEE, 2019.
  44. The role of simulation in the development of endovascular surgical skills. Perspect Med Educ, 5(1):8–14, February 2016. ISSN 2212-277X.
  45. End-to-end real-time catheter segmentation with optical flow-guided warping during endovascular intervention. In 2020 IEEE International Conference on Robotics and Automation (ICRA), pp.  9967–9973. IEEE, 2020.
  46. Allison M Okamura. Haptic feedback in robot-assisted minimally invasive surgery. Current opinion in urology, 19(1):102, 2009.
  47. Towards characterization and adaptive compensation of backlash in a novel robotic catheter system for cardiovascular interventions. IEEE Transactions on Biomedical Circuits and Systems, 12(4):824–838, 2018.
  48. A novel sample-efficient deep reinforcement learning with episodic policy transfer for pid-based control in cardiac catheterization robots. arXiv preprint arXiv:2110.14941, 2021.
  49. AD Padsalgikar. Cardiovascular system: Structure, assessment, and diseases. Plastics in medical devices for cardiovascular applications. William Andrew Publishing, pp.  103–132, 2017.
  50. Robot-assisted techniques in vascular and endovascular surgery. Langenbeck’s Archives of Surgery, 407(5):1789–1795, 2022.
  51. Towards rebuild the interventionist’s intra-operative natural behavior: A fully sensorized endovascular robotic system design. In 2019 International Conference on Medical Imaging Physics and Engineering (ICMIPE), pp.  1–7. IEEE, 2019.
  52. Stable-baselines3: Reliable reinforcement learning implementations. Journal of Machine Learning Research, 22(268):1–8, 2021. URL http://jmlr.org/papers/v22/20-1364.html.
  53. Assessment of navigation cues with proximal force sensing during endovascular catheterization. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2012: 15th International Conference, Nice, France, October 1-5, 2012, Proceedings, Part II 15, pp.  560–567. Springer, 2012.
  54. Learning-based modeling of endovascular navigation for collaborative robotic catheterization. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2013: 16th International Conference, Nagoya, Japan, September 22-26, 2013, Proceedings, Part II 16, pp.  369–377. Springer, 2013.
  55. Hierarchical hmm based learning of navigation primitives for cooperative robotic endovascular catheterization. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2014: 17th International Conference, Boston, MA, USA, September 14-18, 2014, Proceedings, Part I 17, pp.  496–503. Springer, 2014.
  56. Objective assessment of endovascular navigation skills with force sensing. Annals of Biomedical Engineering, 45(5):1315–1327, May 2017. ISSN 1573-9686. URL https://doi.org/10.1007/s10439-017-1791-y.
  57. Dynamic movement primitives in robotics: A tutorial survey. arXiv preprint arXiv:2102.03861, 2021.
  58. Automated planning for robotic guidewire navigation in the coronary arteries. In 2022 IEEE 5th International Conference on Soft Robotics (RoboSoft), pp.  239–246. IEEE, 2022.
  59. Evidence for endovascular simulation training: a systematic review. European Journal of Vascular and Endovascular Surgery, 51(3):441–451, 2016.
  60. Shape sensing techniques for continuum robots in minimally invasive surgery: A survey. IEEE Transactions on Biomedical Engineering, 64(8):1665–1678, 2016.
  61. Basic endovascular skills trainer: A surgical simulator for the training of novice practitioners of endovascular procedures. In 2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2015.
  62. Learning-based catheter and guidewire-driven autonomous vascular intervention robotic system for reduced repulsive force. Journal of Computational Design and Engineering, 9(5):1549–1564, 2022.
  63. Socially compliant navigation through raw depth inputs with generative adversarial imitation learning. In 2018 IEEE international conference on robotics and automation (ICRA), pp.  1111–1117. IEEE, 2018.
  64. Interactive training system for interventional electrocardiology procedures. In Biomedical Simulation: 6th International Symposium, ISBMS 2014, Strasbourg, France, October 16-17, 2014. Proceedings 6, pp.  11–19. Springer, 2014.
  65. Deepmind control suite, 2018.
  66. Emanuel Todorov. A convex, smooth and invertible contact model for trajectory optimization. In 2011 IEEE International Conference on Robotics and Automation, pp.  1071–1076. IEEE, 2011.
  67. Mujoco: A physics engine for model-based control. In IROS, 2012.
  68. Light-weight deformable registration using adversarial learning with distilling knowledge. IEEE transactions on medical imaging, 41(6):1443–1453, 2022.
  69. The use of soft robotics in cardiovascular therapy. Expert Review of Cardiovascular Therapy, 15(10):767–774, 2017.
  70. A (near) real-time simulation method of aneurysm coil embolization. Aneurysm, 8(29):223–248, 2012.
  71. Medical robotics–regulatory, ethical, and legal considerations for increasing levels of autonomy, 2017.
  72. Automatic catheter detection in pediatric x-ray images using a scale-recurrent network and synthetic data. arXiv:1806.00921, 2018.
  73. Novel contact force sensor incorporated in irrigated radiofrequency ablation catheter predicts lesion size and incidence of steam pop and thrombus. Circulation: Arrhythmia and Electrophysiology, 1(5):354–362, 2008.
  74. Automatic control of cardiac ablation catheter with deep reinforcement learning method. Journal of Mechanical Science and Technology, 33:5415–5423, 2019.
  75. Sim-to-real transfer in deep reinforcement learning for robotics: a survey. In 2020 IEEE symposium series on computational intelligence (SSCI), pp.  737–744. IEEE, 2020.
Citations (4)
View on Semantic Scholar

Summary

  • The paper introduces CathSim, an open-source simulator facilitating realistic and safe testing of machine learning algorithms for autonomous catheterization.
  • It leverages an Expert Navigation Network using reinforcement learning (Soft Actor-Critic) to optimize trajectory planning in endovascular procedures.
  • The study highlights practical implications for reducing radiation exposure and increasing precision in robotic-assisted endovascular interventions.

Autonomous Catheterization with Open-source Simulator and Expert Trajectory

The quest for advancing robotic-assisted endovascular interventions is marked by significant obstacles, including the reliance on closed-source simulators and the scarcity of large-scale datasets necessary for machine learning applications. Addressing these challenges, the paper introduces "CathSim," an open-source simulator specifically designed to facilitate endovascular interventions with a focus on autonomous catheterization.

Context and Motivation

Endovascular procedures, due to their minimally invasive nature, offer substantial advantages over traditional surgeries. They involve the navigation of flexible tools such as guidewires and catheters to target regions within the vascular system for diagnostic and therapeutic purposes. The interventions typically utilize X-ray imaging, exposing surgeons to radiation hazards while demanding high precision in manual control—a task compounded by a lack of tactile feedback.

Robotic systems have emerged as a promising solution, enabling remote operation and potential automation. Nevertheless, the conventional master-slave robotic systems still depend heavily on human control, hindering the efficiency meant to be achieved through robot-assisted procedures. Therefore, the development of autonomous systems is crucial for further advancement.

CathSim: Development and Features

CathSim represents a considerable advancement towards autonomous catheterization by providing an open and accessible tool for the development and testing of machine learning algorithms in a realistic simulated environment. Built on the MuJoCo engine, CathSim offers high fidelity simulations of endovascular interventions with key features such as:

  • Anatomically Accurate Models: It includes high-resolution anatomical models like the aorta, sourced from Elastrat Sarl Ltd., Switzerland.
  • Multimodal Integration: Designed to support real-time force feedback and AR/VR interfaces, enabling comprehensive training and testing scenarios.
  • Modular Design: CathSim features a modular construction that allows for straightforward upgrades and extensions, crucial for diverse ML algorithm requirements.
  • Open-source Nature: Providing the research and medical community with unrestricted access enhances collaborative developments and benchmarking.

Expert Trajectory Network

A pivotal contribution of the paper is the development of a multimodal Expert Navigation Network (ENN). This network capitalizes on the simulation's capacities to generate extensive dataset-driven training environments that are otherwise limited or hazardous in real-world contexts. Various input modalities, such as segmentation and dynamic motion data, are integrated into a deep learning framework that extends beyond what a human surgeon typically utilizes during procedures.

The ENN operates on state-of-the-art reinforcement learning paradigms, particularly leveraging the Soft Actor-Critic (SAC) algorithm for policy optimization, and demonstrates notable performance improvements compared to human and baseline automated approaches. It highlights potential applications in reinforcing ML imitation learning strategies, offering streamlined transition from simulation to real-world scenarios.

Implications and Future Directions

CathSim, as the first open-source simulator in this domain, holds the potential to significantly catalyze progress towards feasible autonomous catheterization systems. The platform invites the ML community to contribute and expand research horizons, potentially addressing pressing challenges such as soft tissue deformation strategies not yet covered. Moreover, while results indicate high efficacy of CathSim-generated trajectories, translating these to actual robotic systems under genuine constraints is a nontrivial frontier requiring further exploration.

The paper underscores the ethical and regulatory barriers intrinsic to transitioning ML-based endovascular interventions from simulation to human trials. Continuous engagement with regulatory bodies and ethical frameworks will be imperative as autonomous systems inch closer to clinical application.

File Document Download Save Streamline Icon: https://streamlinehq.com PDF File Document Download Save Streamline Icon: https://streamlinehq.com Markdown
Github Logo Streamline Icon: https://streamlinehq.com

GitHub

  1. GitHub - airvlab/cathsim (101 stars)