RoboticCatheter
Project: Non-fluoroscopy-based robotic catheter manipulation and tracking based on machine learning, bioelectrical localization and robotics
Collaborating Departments: Computer Aided Medical Procedures & Augmented Reality (TUM); Department of Mechanical Engineering (Imperial)
Minimally invasive surgery for vascular disease diagnostics and treatment requires a means to steer surgical tools through complex vessel trees. The procedure is usually performed manually, under fluoroscopic guidance. Thus, it exposes the patient and staff to unnecessary radiation, which is a growing concern for the National Cancer Institute. The aim of this project is to explore non-fluoroscopy-based robotic catheter manipulation and tracking based on machine learning, Bioelectric localization, Ultrasound and robotics. We propose to simultaneously explore two systems. The first, an electrogenic sensorised catheter which will employ embedded electrodes to induce low voltage electric currents within the patient and measure the local electric impedance at different sections of the vascular tree. AI based reconstruction of the vasculature from preoperative CTA images will be used to generate a patient-specific model. Machine learning will also be used to map the online time-series impedance measurements to the signals simulated based on the model, to estimate the catheter location within the vascular tree. The second, a robotic ultrasound guided system which uses deep learning to track catheters in sequences of images to learn spatial and temporal information, thus localise the catheter using a vision-based approach. In addition, FBG-based shape sensing will be investigated, which allows to estimate the catheter body and tip shape, enhancing the capabilities of both systems. Together with the impedance or Ultrasound localization, this could provide both catheter position and pose at the same time. When fused together with preoperative CT/MR, the system will also be able to monitor the relative position of the catheter body with the vasculature’s inner wall, leading to a system that significantly reduces the risks of perforation, dissection, embolisation and thrombosis. The implemented tracking system will be incorporated within a state-of-the-art robotic steering system for cooperative catheter insertion available at Imperial. The system will feature multiple degrees of freedom to accurately control the insertion process, haptic feedback for an improved user experience, and the option for manual or robotic insertion. The result will be a novel, radiation-free system for endovascular surgery.
At TUM, Heiko Maier firstly focused on extending the concept of Bioelectric Navigation. During this phase we investigated Deep Learning variants of Dynamic Time Warping (DTW), and also applied them to segmentation in an exemplary application [1]. Furthermore, we proposed an extended electrical sensing concept which adds sensing of distance and travelled direction to Bioelectric Navigation [2]. Finally, in a collaboration project with Ardit Ramadani, another Doctoral Researcher at TUM, we investigated the combination of Bioelectric Navigation with electromagnetic tracking [3].
Since Alex Ranne, at Imperial College, is planning to first focus on Deep Learning and FBG tracking, we have further extended the existing collaboration to create a working group between the Doctoral Researchers Alex Ranne, Heiko Maier and Yordanka Velikova, another TUM researcher who works on US imaging. This extended collaboration originated during a visit of Yordanka Velikova, supported by the TUM Global Incentive Fund (Imperial Collaboration), at Imperial College. Since obtaining intraoperative ultrasound has been challenging, Alex and Yordanka has developed a physics based endovascular simulator pipeline, and are currently exploring the extension of this work using softbody simulators [4]. The objective here is to use US and transformer structures to track catheters. Later on, we plan to study fusion concepts of Bioelectric Navigation, FBG, US and electromagnetic tracking and produce a series of scientific work towards radiation-free catheter tracking for vascular procedures.
Ramadani, A.*, Maier, H.*, Bourier, F.,Meierhofer, C., Ewert, P., Schunkert, H. and Navab, N. (2023). Feature-based Electromagnetic Tracking Registration Using Bioelectric Sensing (Accepted for Publication in: RA-L (IEEE Robotics and Automation Letters). - [3]
A. Ramadani, H. Maier, F. Bourier, C. Meierhofer, P. Ewert, H. Schunkert and N. Navab (2022): Bioelectric Registration of Electromagnetic Tracking and Preoperative Volume Data.
Submitted to IEEE Robotics and Automation Letters. arxiv.2206.10616 [physics.med-ph].
Conferences:
- H. Maier, S. Schunkert, N. Navab (2023): Extending Bioelectric Navigation for Displacement and Direction Detection, in: IPCAI 2023 (The 14th International Conference on Information Processing in Computer-Assisted Interventions). - [2]
- H. Maier, S. Faghihroohi, N. Navab(2021): A Line to Align: Deep Dynamic Time Warping for Retinal OCT Segmentation, in: Medical Image Computing and Computer Assisted Intervention – MICCAI 2021. Lecture Notes in Computer Science(), vol 12901. Springer, Cham. - [1]
Studyproject/Masterthesis:
- Breschi, S., Ranne, A.T.J, Rodriguez y Baena, F., de Momi, E. (2023). A soft body simulator for human robot collaborative endovascular procedures. Masters thesis, Politecnico Di Milano. - [4]
Team
Principal Investigator (Imperial)
Professor Ferdinando Rodriguez y Baena,
Professor of Medical Robotics | Imperial
Co-Director of Hamlyn Centre
Principal Investigator (TUM)
Professor Nassir Navab
Chair for Computer Aided Medical Procedures & Augmented Reality
Doctoral Candidate (Imperial)
Alex Ranne