Machine Learning-based Adaptive Robot-assisted Training for Surgical Robotics
Funding source: Intuitive Surgical Technology research grant
Funding period: 2017
The main hypothesis of this proposal is that training phases of a trainee can be recognized through performance measures and that a proportional physical guidance can measurably improve the outcome of the training procedure. The project comprises two principal aims:
Aim 1: Implementing trainee progress assessment through machine learning methods
The methodology for assessing surgical skills is gradually shifting from subjective scoring of an expert, which may be a variably biased opinion using vague and subjective criteria, towards a more quantitative analysis. This project aims at assessing various skills of a tele-operated surgical system’s operator through a Machine Learning (ML) powered agent. The assessment is in the context of training and thus the agent will be learning-phase-aware. For a set of predefined tasks, the ML agent will measure the performance of a trainee and produce an estimate of his/her progress with respect to the average learning curve of the specific task that can be used as an effective feedback to the trainee or in adapting accordingly the training program.
Aim 2: Design and evaluation of adaptive physical assistive methods in training
The second aim of the project is to investigate the effects of providing physical guidance in primary phases of tele-operated surgical training. This robot-assisted training guides the trainee through motion, force, torque, or vibration cues applied by the master device and is adapted to users’ progress modelled by the ML agent. Such an assistive method is hypothesized to facilitate the initial stages of training by preventing cognitive overload. These methods will be studied through multiple experiments using a tele-operation setup with statistically significant populations of subjects. The results of the study will reveal potential benefits or harms of motion guidance in skills training in terms of both immediate improvements and long-term retention.
Funding source: RIA H2020-ICT-2016
Grant number:H2020-ICT-2016- 732515
Funding period: 2017 – 2020
Funding period: 2016 – 2020
EDEN2020 will provide a step change in the modelling, planning and delivery of diagnostic sensors and therapies to the brain via flexible surgical access, with an initial focus on cancer therapy. It will engineer a family of steerable catheters for chronic disease management that can be robotically deployed and kept in situ for extended periods. The system will feature enhanced autonomy, surgeon cooperation, targeting proficiency and fault tolerance with a suite of technologies that are commensurate to the unique challenges of neurosurgery. Amongst these, the system will be able to sense and perceive intraoperative, continuously deforming, brain anatomy at unmatched accuracy, precision and update rates, and deploy a range of diagnostic optical sensors with the potential to revolutionise today’s approach to brain disease management. By modelling and predicting drug diffusion within the brain with unprecedented fidelity, EDEN2020 will contribute to the wider clinical challenge of extending and enhancing the quality of life of cancer patients–with the ability to plan therapies around delicate tissue structures and with unparalleled delivery accuracy.
EDEN2020 is strengthened by a significant industrial presence, which is embedded within the entire R&D process to enforce best practices and maximise translation and the exploitation of project outputs. As it aspires to impact the state of the art and consolidate the position of European industrial robotics, it will directly support the Europe 2020 Strategy.
For more information, please visit EDEN2020 website.