Thèse soutenue

Utilisation d'un robot sous-actionné pour la détection du cancer du sein
FR  |  
EN
Accès à la thèse
Auteur / Autrice : Anja Marx
Direction : Guillaume Morel
Type : Thèse de doctorat
Discipline(s) : Robotique
Date : Soutenance en 2013
Etablissement(s) : Paris 6

Mots clés

FR

Mots clés contrôlés

Résumé

FR  |  
EN

This research is situated in the emerging field of co-manipulation systems where robot and user perform a task in a collaborative way. It is applied to the medical context of breast cancer diagnosis where the standard procedure today is the succession of an initial mammography (MX) examination and a supplementary Ultrasound (U/S) scan. The surgeon's task is to localize the target lesion defined in the MX images using 2D U/S. One difficulty of this procedure results from the fact that breast geometry changes for both examinations due to different patient's positions. A second difficulty is the mental correlation of two different image types. MX provides a 3D view, whereas U/S only displays a cross section of the object. The proposed system facilitates this combined examination by keeping the breast geometry and by adding a U/S probe guidance robot to the mammography system. A 6DOF parallel co-manipulation system is set up where the robot and user simultaneously impact on the probe. A robot control is developed for active task assistance. Its relevance is evaluated in vitro and showed a significant increase in examination quality when using robot guidance compared to the standard examination. The novel aspect treated in this thesis is under-actuated co-manipulation, where the robot has less DOF than the task requires. The initial idea is that, although the robot cannot perform the task autonomously, it may bring partial assistance that still improves the movement. This involves adaptation in terms of robot control and system architecture. It is shown in this work that, in case of under-actuation, simple reduction from a fully-actuated robot control to the remaining robot DOF is not sufficient to guarantee system stability. System architecture needs to be adapted accordingly. To quantify examination improvements when using under-actuated guidance, different under-actuated robot controls have been compared. The main outcome of this thesis is that even an under-actuated robot system increases task precision significantly while decreasing execution time.