IROS2019国际学术会议论文集 0495

Abstract Vascular diseases are the most common precursors to ischemic heart disease and stroke which are two of the leading causes of death worldwide Endovascular intervention is a minimally invasive surgical approach to treat such diseases Compared to open surgery it has the advantages of faster recovery reduced need for general anesthesia reduced blood loss and significantly lower mortality Endovascular procedures require high surgical skills to minimize contacts between the manipulated instruments catheters and guidewires and the vessel wall which represent one of the major risks for the patient Robotic assistance can potentially improve the precision and stability of instruments manipulation One key limitation of current commercial robotic platforms is the lack of haptic feedback preventing their acceptance and limiting the clinical usability This paper proposes to bring the benefit of haptic feedback to robot assisted endovascular intervention Here we hypothesize that the introduction of 3D haptic guidance during robot assisted endovascular procedure can further improve the surgical performance and safety while overcoming the limitations of currently available technology The proposed 3D haptic guidance allows the surgeon to sense the vasculature while controlling a catheter through a robotic haptic manipulator Validation of the system is performed through end user experiments with vascular surgeons on a bespoke surgical simulator The obtained results demonstrate that 3D haptic guidance has the potential of improving effectiveness precision and safety of endovascular intervention Furthermore vascular surgeons found the proposed technology safe and overall easy to use indicating its potential on real surgical procedures I INTRODUCTION Thanks to the technological progression since the 1980s Minimally Invasive Surgery MIS has become an established approach across many surgical specialties 1 Despite its advantages over traditional open surgery quicker patient recovery reduced tissue disruption and hospitalization costs MIS procedures can be ergonomically difficult to perform due to the use of rigid instruments limited sensory feedback misalignment of visuo motor axes and the need for high dexterity In response to these limitations robotics and computer assistance have been integrated into the clinical workflow to provide augmentation of surgical skills in terms of enhanced dexterity and precision However such systems effectively decouple surgeons actions from their resulting interactions with the surgical site thus depriving surgeons of tactile feedback M B Molinero G Dagnino J Liu W Chi M E M K Abdelaziz and G Z Yang are with The Hamlyn Centre for Robotic Surgery Imperial College London UK e mail dagnino giulio T M Y Kwok and C Riga are with the Department of Surgery Amigo Catheter Robotics Inc NJ USA CorPath GRX Corindus Vascular Robotics MA USA and the R one Robocath France which manipulate standard catheters and guidewires Unfortunately none of the above platforms excluding Sensei X2 provides haptic feedback Therefore contact forces M B Molinero G Dagnino J Liu W Chi M E M K Abdelaziz T M Y Kwok C Riga and G Z Yang Haptic Guidance for Robot Assisted Endovascular Procedures Implementation and Evaluation on Surgical Simulator Figure 1 The CathBot system Motion command on the master manipulator A are sent to the slave robot B to manipulate a catheter or a guidewire while the 2D navigation system not used in the work reported in this paper provides visual guidance C Adapted from 17 2019 IEEE RSJ International Conference on Intelligent Robots and Systems IROS Macau China November 4 8 2019 978 1 7281 4003 2 19 31 00 2019 IEEE5398 between the manipulated instruments and the anatomy are not perceived by the surgeon thus making the already challenging tasks even more complex and dangerous for the patient Solutions to this problem have been proposed by several research groups 6 14 by embedding force sensors at the proximal or distal portion of the catheter to provide haptic feedback However incorporating force sensors into small instruments like catheters and guidewires is practically difficult expensive and time consuming also it requires modifying commercial medical instruments thus introducing clinical usability issues such as safety biocompatibility and sterilization Vision based force sensing through shape and position analysis of the catheter and the vasculature is a promising solution as demonstrated in preliminary work reported in 15 17 Here catheter and vasculature are segmented on intra operative fluoroscopic images and used to provide real time haptic feedback to the surgeon based on catheter vessel relative proximity However the 2D nature of fluoroscopic imaging used in these works limits the haptic feedback to the imaging plane i e the catheter vessel contacts out of the imaging plane or normal to the imaging plane are not detected This clearly represents a major limitation in terms of clinical usability In this paper we hypothesize that the introduction haptic feedback based on 3D vision hereafter called 3D vision based haptic guidance during robot assisted endovascular procedures can further improve surgical performance and safety while overcoming the issues highlighted above We describe the design development and assessment of 3D dynamic haptic guidance for our robotic platform the CathBot The main objective of this research is the design and evaluation of the 3D haptic guidance with the resulting contributions 1 presenting the first framework for the creation and testing of haptic feedback and active constraints in endovascular procedures based on 3D vision 2 pilot end user study with vascular surgeons to understand its applicability usefulness and acceptance to robot assisted endovascular procedures Section II introduces the Hamlyn s CathBot system and describes the 3D vision based haptic guidance and the methodology to create active constraints Section III presents the surgical simulator developed to implement and test the 3D vision based haptic guidance under controlled and repeatable conditions and its integration with the CathBot system Section IV reports the experimental evaluation including setup results and discussion Section V concludes the paper offering future research directions II HAPTIC GUIDANCE BASED ON 3D VISION Research in our group at the Hamlyn Centre focuses on overcoming the current principal technological limitations of endovascular procedures namely limited catheter guidewire maneuverability lack of adequate 3D navigation and lack of contact force sensing 5 by creating the CathBot robotic platform 13 17 18 This is a teleoperated system in a master slave configuration that provides improved catheter guidewire maneuverability and control by offering an ergonomic manipulator that is designed to transmit haptic feedback to the operator Fig 1 Force feedback is generated through vision by tracking the tip of the catheter and the vasculature on 2D images 17 please refer to the online video contribution in 17 However as anticipated before with this technique 3D anatomy and depth information are lost and the haptic feedback is calculated only on the imaging plane representing a major limitation of the system in terms of clinical usability and safety This paper continues the improvement in haptic feedback by introducing for the first time haptic guidance based on 3D vision to robot assisted endovascular procedures This represents also a step forward towards introducing dynamic navigation into endovascular procedures one of the CathBot future objectives Figure 2 shows how haptic guidance based on 3D vision is created The haptic device used in this work is the CathBot manipulator introduced in 17 Fig 1A and 2A The main component of the manipulator is a tube like structure that can be moved linearly and rotated axially Any linear movement causes a railing to slide through a Linear Motor LM LM1247 Faulhaber while the torque Figure 2 3D vision based haptic guidance the surgeon manipulates the catheter through the CathBot mechanical manipulator A The 3D navigation system a surgical simulator in this research calculates the 3D distance and orientation of the catheter tip with respect to the vessel wall B C This information is processed by the controller and coupled with surgeon s motion commands to generate force feedback and haptic guidance 5399 generated by rotations is transmitted through a belt to a Brushless Motor BLM BLDC 2057B Faulhaber LM and BLM are used to generate motion commands on the slave side of the system i e to move a catheter or guidewire and to transmit haptic feedback back to the user Haptic feedback is perceived as frictions which increase proportionally to the catheter vessel distance i e the closer the catheter is to the vessel the higher is the force feedback generated into the master manipulator to inform the surgeon of the proximity of the wall Contacts between catheter and vessel wall should not be completely avoided as they are used by the operator to navigate the instrument usually not steerable through the vasculature However the magnitude of the catheter vessel hits should be minimized while allowing operator to freely navigate the instrument Therefore friction like forces were chosen for haptic rendering instead of other options e g repulsive forces Here we propose to solve the problem by measuring the distance from the instrument tip e g a catheter and a 3D model of the vasculature generated pre operatively by CT data This is done by acquiring patient specific CT images and segmenting them using dedicated software in our case 3D Slicer The 3D pose position and orientation of the catheter tip Pt xt yt zt xt yt zt is provided by the simulator described in the next section In a real case Pt can be provided by the NDI Aurora system as in 18 by placing an EM 6DOF sensor to the tip of the catheter To calculate the 3D distance between the instrument tip and the vessel wall a bespoke tracking method was developed including ray casting and collision detection algorithms Firstly Fig 2B a ray casting algorithm is applied 100 equally spaced rays are cast in different directions starting at the actual position of the catheter tip these rays form a sphere i e the red sphere in Fig 2B Secondly the collision detection algorithm detects the collisions between the rays and the colliders placed on the vessel wall Finally Fig 2C the algorithm selects the point on the vessel wall Pw xw yw zw xw yw zw with minimum distance d from the catheter tip Pt This information is then used to model the damping factor f as described in 1 1 2 1 1 1 2 where d is the Cartesian distance between the catheter tip Pt and the vessel wall Pw D is the local vessel diameter fmax is the maximum friction achievable z is the angle between the longitudinal axes of the catheter tip Pt and the normal to the vessel wall in Pw z is computed using the quaternion product between the QPw quaternion orientation of the vessel wall and the inverse of the QPt quaternion pose of the catheter tip 1 0 0 0 0 0 0 3 acos 1 1 4 where wi ax by cz are components of the quaternions It is wort noting that if the catheter tip is in contact with d 0 or perpendicular to z 0 the vessel wall then f is equal to fmax The damping modelled in 1 is then used in 5 to provide haptic feedback perceived as a friction to the surgeon through the haptic manipulator Fig 2 The LM and BLM generates respectively the linear friction perceived when pushing and pulling the catheter and the rotational friction perceived when twisting the catheter The haptic feedback is generated for both LB and BLM as 19 5 where Vm is the motors velocity control output I is the motor current control input and f is the damping derived in 1 When the surgeon applies a force on the tube handle to manipulate the catheter a motor current I proportional to the force applied is generated The corresponding motor velocity LM and BLM is directly proportional to the force applied described by I and inversely to the damping factor f Considering 1 and 5 this means that when the surgeon pushes the catheter towards the vessel wall then the friction generated by the motors equation 5 increases accordingly We modelled this friction to act as repulsive active constraint that guide the surgeon in manipulating the catheter through the vasculature It is also dynamic as it adapts to the motion of the vasculature due to heartbeat and respiration thus further minimizing undesired and dangerous contacts between the manipulated instrument and the vessel Here we hypothesize that such haptic guidance can provide enhanced assistance to the surgeon and improve the performance and safety of this delicate procedure However implementing the proposed haptic guidance in a real endovascular procedure is still an open challenge The main issue currently is the lack of dynamic real time intra operative 3D imaging current practice generally relies on 2D fluoroscopy or statically registered pre operatively generated 3D models Therefore in order to assess the proposed 3D vision based haptic guidance we have developed the surgical simulator for robot assisted endovascular procedures which is described in the next section III SURGICAL SIMULATOR FOR ROBOT ASSISTED ENDOVASCULAR PROCEDURES A surgical simulator was developed in Unity to assess the proposed 3D vision based haptic guidance under controllable and repeatable experimental conditions Fig 3 The simulator features 1 generation and visualization of a 3D dynamic model of the vasculature based on pre operative CT images 2 modeling of a vascular catheter based on pre operative CT data 3 implementation of 3D haptic guidance as described 5400 above and 4 assessment of users performance through evaluation metrics In this study a model of an aortic arch Fig 3A B C was generated from pre operative CT images generated by 4D XCAT 20 The model was rigged and animated in Blender to simulate the motions due to heartbeat and breathing The aortic arch model was animated to replicate realistic movements for a more realistic representation of a real intervention This was achieved by incorporating in our model the results obtained by Beller et al 21 and Weber et al 22 on the aortic arch dynamic due to heartbeat and respiratory cycle respectively The final animated model was then imported into Unity where physics was applied to allow interaction with other elements i e the catheter model in the simulated environment A similar procedure was followed to model the catheter Fig 3D E F A commercial 5 Fr angiography catheter Beacon Tip Van Schie 1 by Cook Medical USA was CT scanned and segmented to generate a 3D model The model was rigged in Blender to create a skeleton that determines where the model can bend or deform Once imported into Unity a mesh collider was assigned to each part forming the model and then linked together through configurable joints to model the physics of the catheter This was necessary to ensure that the catheter model behaved realistically and interacted properly with the aortic arch model Fig 3G Once both the aortic arch and catheter models were imported into the simulation environment the colliders assigned and the physics correctly modeled the 3D haptic guidance was implemented as described in section II As a result the user could navigate the model of the catheter within the 3D virtual vasculature while receiving haptic guidance through the master manipulator connected to the simulator The pose of the catheter tip Pt is provided by the simulator in real time Lastly the simulator was designed to assess users performance in terms of catheter navigation precision under two different conditions C 1 no constraints and C 2 3D vision based haptic guidance For each condition two different tasks were available T 1 cannulation of the left ventricle LV and T 2 cannulation of the left subclavian artery LSA both via retrograde infra inguinal access Several evaluation metrics were designed and embedded into the simulator to objectively assess the users performance Namely 1 distance between the catheter tip and the vessel wall 2 number of tip vessel contacts 3 distance between the catheter tip and center of the vessel 4 time to complete the task These metrics recorded during the surgical simulations were then used off line to calculate two more metrics 5 average speed of the catheter and 6 path following RMSE tip path vs vessel center line IV EXPERIMENTAL EVALUATION This section reports the experimental setup and methodology used to evaluate whether the proposed 3D vision based haptic guidance can improve the surgical performance The readers are referred to the attached video for a visual overview of the experimental evaluation The master manipulator and the surgical simulator were used in an experiment designed to compare as introduced earlier two different conditions i e C 1 no constraints and C 2 3D vision based haptic guidance 3 endovascular trained vascular surgeons were asked to complete two different cannulation procedures i e T 1 cannulation of the LV and T 2 cannulation of the LSA The users watched the simulated scene consisting of an animated aortic arch and a catheter model on a laptop screen running on an HP ProBook i5 4200M 2 50GHz and manipulated the catheter model through the master manipulator control algorithms running on a compactRIO 9022 National Instruments at 1kHz In this experiment an anterior posterior AP view of the aortic arch was provided to simulate a 2D fluoroscopic image Fig 3H i e the type of imaging typically used in clinical practice No 3D visual information was provided so that the 3D mensionality of the anatomy could be sensed only through the 3D haptic guidance allowing us to better assess its efficacy The experiment consisted of four phases First the users received information and instructions about the experiment Second they were given 5 minutes to familiarize themselves with