Design and Development of a 3 DOF Force Sensor Using Fiber Bragg Gratings in Endoscopic Applications
par Ben Hassen, Ramzi
Président du jury Nonclercq, Antoine
Promoteur Delchambre, Alain
Publication Non publié, 2023-05-31
Président du jury Nonclercq, Antoine
Promoteur Delchambre, Alain
Publication Non publié, 2023-05-31
Thèse de doctorat
Résumé : | Minimally invasive procedures have flourished in the last decades to treat patients with less trauma.Therapeutic endoscopy is one these emerging techniques using flexible endoscopes through naturalorifices to go beyond the simple diagnostic and enables the treatment of digestive tract disorders likeearly gastrointestinal tumors by the mean of advanced endoscopic techniques such as EndoscopicSubmucosal Dissection (ESD). Despite renowned clinical benefits compared to other tumor resectiontechniques, ESD is a challenging procedure with a high level of expertise and dexterity needed tomaster it. Lack of expertise centers and training in Western countries makes the diffusion of ESD stilllimited. According to experts, a force-feedback technology directly integrated in the electrosurgicalknife would help trainees to flatten their learning curve, especially on animal models. This thesisproposes to solve the problem by integrating Fiber Bragg Gratings (FBGs) as 3 degrees-of-freedomoptical force sensors into the catheter of the electrosurgical knife aiming to measure Fx, Fy and Fz.Several designs were explored to finally implement 3 FBGs placed circumferentially to the sectionof the catheter. The use of nitinol tubes and a specific pasting method of the FBG led to bettersensitivity and resolution. A force calibration test bench was specifically designed to calibrate theforce sensor in 30 3D spatial directions that cover most of its use cases. Non-linear regression modelswere implemented to tackle the non-linearities between the wavelength shifts of the FBGs and theforces applied, inherent to prototyping errors and imprecisions and non-linearity of the material. Ahybrid model made of mono- and bi-layered neural networks for the prediction of Fx and Fy anda support vector regression for the prediction of Fz was built and verified by using training andtesting data set. The results are promising and satisfying most of the technical requirements. Atemperature compensation technique was tested and verified on a specific prototype with endoscopicdimensions. In parallel to the force sensor development, a force-feedback user interface was designedto display the three components of the force using polar coordinates for the lateral forces andconcentric circles for the axial one. A warning alarm was also implemented to prevent from criticalevent such as exceeding a force threshold. Further works will mainly consist in the implementation ofa temperature compensation technique directly in the design of the force sensor and the integrationof the force-feedback interface with the force sensor in real-time ESD training settings. |