Project Details
Projekt
Intelligent design assistance for personalized medical surgery based on concentric tube continuum robots: Part II
Applicants
Professorin Dr. Kathrin Flaßkamp, until 5/2025; Professor Dr.-Ing. Jürgen Maas; Professor Dr. Joachim Oertel; Professor Dr.-Ing. Thomas Sattel
Subject Area
Mechanics
Term
since 2022
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 501928699
Continuum robotics is now attracting increasing attention from the medical world, particularly for minimally invasive applications. Stereotactic neurosurgery for surgical interventions in the brain is one example of this. Due to the complexity and variety in the design of surgical instruments, in this case cannulas, surgical application scenarios and the required high precision of interventions, AI-based design assistants are proving to be expedient. In the first funding period of the DFG-SPP 2353, meta-heuristics from multi-objective optimisation and learning methods for the combined instrument design and control problem were researched. In addition, learning methods for physics-inspired learning with experimental data and expert knowledge for precise behaviour mapping and classification of the instruments were investigated and used. In the first phase of the second funding period, the AI-based design methods will be expanded. Until now, the trajectory in the brain has been planned and specified manually by the surgeon. The aim is to extend the optimal instrument and control design to the combined and simultaneous optimal design of instruments, instrument trajectories and control. Consideration of the trajectory in the design process is the decisive step from a surgical perspective. The data-driven classification of instrument designs with support vector machines according to their stability behaviour must also be incorporated. Extended experimental data acquisition is used to refine the instrument models for sufficiently precise movement predictions using various learning methods. For the first time, a complete surgical workflow with continuum robots is to be carried out in a clinical environment in order to integrate the knowledge gained. The second section deals with the instrumentation of a continuum robot with sensors and control algorithms. In order to achieve the precision of path and target point guidance, experimental data-driven learning methods for motion control and regulation are to be investigated. Model predictive methods and reinforcement learning methods can be considered. The final section of the research project is dedicated to the development and experimental validation of a holistic AI-based design assistant. The methods/processes and modules researched and developed to date will now be brought together. In this research phase, surgeons will be involved in the development by incorporating the surgical procedures and process flows as well as the evaluation criteria and decision-making via graduated Pareto fronts and their visualisation. The testing, calibration and validation of the AI-based design assistant is then the final step carried out jointly by engineers and surgeons.
DFG Programme
Priority Programmes
