IROS2019国际学术会议论文集 2479

A Testbed for Haptic and Magnetic Resonance Imaging Guided Percutaneous Needle Biopsy Evelyn Mendoza1 and John P Whitney1 Abstract Magnetic resonance imaging MRI is used for soft tissue pathology diagnosis and assistance with targeted lesion procedures However limited physician access to patients in the MR scanner bore requires iterative positioning and imaging which extends the procedure time and increases patient risk We present a teleoperated system and instrumented testbed for robotic magnetic resonance imaging guided needle biopsy particularly targeting transperineal prostate biopsies The device has a 1 1 kinematic mapping and uses a low friction hydrostatic transmission based on antagonistic pairs of rolling diaphragms It has 3 DOFs and a passive 2 DOF swivel which holds a biopsy needle at the end effector Measured force hysteresis is 0 085N single DOF and 0 44N full 3 DOF manipulator at the needle driving point To quantitatively assess the system s haptic qualities impulse modal testing for 1 DOF 300 Hz bandwidth and 3 DOF 40 Hz bandwidth was conducted Task performance was also assessed through membrane puncture testing with tissue phantoms puncture events as low as 0 1N are reliably detectable from transmission hydraulic pressure sensors I INTRODUCTION The number of MRI guided interventions has increased 10 annually from 1996 to 2010 with 30 2 million in 2010 1 MRI scanners offer the ability to diagnose soft tissue pathology have a guided intervention and asses the results of the intervention in one single session The bore of MRI machines is narrow and it provides limited space between the patient and physician This requires iterative positioning of the patient and biopsy tools between scans not taking advantage of the real time imaging provided by the machine 2 Robotic assistance systems improve patient access and tool positioning but MR environment compatibility require ments constrain their design and limit their transparency 3 Some robotic solutions include a piezoelectically actuated stereotactic neurosurgery robot 4 a pneumatically actuated robot for aortic valve replacement 5 a novel MRI powered robot actuator for tissue biopsy 6 a pneumatically actuated cannula robot 7 and a passive cable driven mechanism for thoracic and abdominal interventions 8 MR Conditional2robotic systems are complicated due to challenges of working with compatible materials including This work was supported in part by the National Science Foundation s Cyber Human Systems under grant 1617122 and 1615891 Corresponding author John Peter Whitney 1Authors are with the Department of Mechanical and Industrial Engi neering Northeastern University Boston MA 02115 USA e mail men doza ev husky neu edu j whitney northeastern edu 2Under the new ASTM F 2503 standard items designated as MR Safe must be completely non metallic Metallic but non ferrous items are considered MR Conditional These two terms replace the previously common designation of MR Compatible 0 1 2 2 DOF swivel joint biopsy needle needle guide plate xe ze ye xn A B 2 cm ceramic ball bearings kevlar cord AT3 timing belts 8 mm output shaft Titanium fasteners Fig 1 color A Kinematic model of haptic 3 DOF manipulator biopsy needle and needle plate Coordinates axes for forwards kinematics B MR conditional rotary actuator fi tted with aluminum parts brass and titanium fasteners ceramic ball bearings and kevlar cord timing belts actuators and sensors that are non magnetic Standard mo tors cannot be used and pneumatic or hydraulic actuation systems require bulky pumps that are diffi cult to place In teleoperational systems the physician will not directly have access to the interventional tools thus tactile sensation is lost To regain haptic feedback and make the procedure more intuitive one solution is to mount force sensors at the distal end to measure insertion forces and render a haptic display to the user Another solution is to develop a manually actu ated passive system with no electronic components Haptic feedback has many benefi ts including improved accuracy faster task times and enhanced learning that have been demonstrated for other teleoperated robot assisted minimally invasive surgeries 9 Several groups have reported the de IEEE Robotics and Automation Letters RAL paper presented at the 2019 IEEE RSJ International Conference on Intelligent Robots and Systems IROS Macau China November 4 8 2019 Copyright 2019 IEEE velopment of MR conditional master slave systems In 10 a 6 DOF telesurgery system was developed using a Fabry Perot interferometry sensor and a pneumatic haptic device with a bandwidth of 5 Hz A sensor at the distal end provided force feedback to the physician Its position tracking error was 0 70 0 35 mm However they need a shielded controller box to operate the robot In 11 a novel passive parallel master slave mechanism with 3 DOFs for translation and a 2 DOF gimbal was evaluated with an average stiffness of 1 7 N mm and it is mechanically coupled In 12 a teleoperated MR compatible hydrostatic transmission system with force feedback was developed The system has 1 DOF with a bandwidth of 20 Hz However the master actuator is a conventional motor that needs to be placed outside the MR room Additionally 13 presents an extensive review of robots developed for needle based interventions since 2014 Many of the issues encounter by the MRI environment are effectively avoided by using mechanical transmissions either as 1 an end to end passive mechanical transmission or 2 a passive transmission system used to shift non MR conditional electric actuators outside the high gauss zone A completely passive system 1 is preferred in some cases due to intrinsic safety and simplicity but achieving high transparency low friction is very challenging remotely actuated systems 2 are more complex but enable features such as virtual fi xtures and tremor fi ltering We present a novel haptic 3 DOF bilateral teleoperation system for remote needle biopsies under MR guided interventions confi gured here for fully passive benchtop testing Our device relies on rotary hydrostatic actuators that uses rolling diaphragm seals 14 This eliminates the high static friction found in hydraulic and pneumatic sliding o x ring or cup type seals Its haptic capabilities and high transparency arise from its high series stiffness light weight and low friction We are initially targeting transperineal prostate interventions because of accessibility to the affected area Device requirements take into account the average prostate size of 5 cm x 4 cm x 3 cm 15 In vivo prostate biopsies using 18 gauge needles results in forces ranging from 9 2 N right after piercing skin to 5 N while puncturing the prostate capsule and reaches a maximum force of 18 N 16 The maximum depth needed for prostate biopsies is 150 mm from the perineal surface Force resolution as low as 0 2 N is de sirable to detect minimum force capsule membrane puncture events 17 II DESIGN DESCRIPTION In the following discussions we refer to experiments on two hardware setups A 3 DOF biopsy device the complete device shown in Fig 2 A It has three hydrostatic rotary actuators for both master and slave Secondly a 1 DOF transmission testbed we isolate only the base DOF for the biopsy devide to study a single pair of hydrostatic rotary actuators Fig 1 B in isolation This confi guration provides one degree of freedom teleoperation In both cases refer ences to linear motions and needle forces refer to driving point motion and force of the biopsy needle at a moment B A master side slave side ultrasound probe biopsy needle agar based tissue phantom ultrasound image hydraulic swivels master side w counterweights 40 000 CPR encoders Fig 2 color A 3 DOF manipulator instrumented with encoders on all three axes and differential pressure sensors on each hydraulic line Hydraulic connections to the two serially connected actuators pass through low friction ball bearing hydraulic swivels SMC KS near the base 0 axis of rotation B Setup for ultrasound guided benchtop MRI proxy needle biopsy testing in agar tissue phantoms arm 180 mm from the rotational center 0 Our haptic 3 DOF biopsy device uses six of the hybrid air water rotary actuators presented in 14 In this design each degree of freedom employs an individual water line that maintains high series stiffness and shares one common pneumatic line with pressurized air maintaining a positive preload pressure under all conditions This feature ensures the hydrostatic transmission is backdrivable and bidirectional without requiring duplicate agonist antagonist hydraulic con nections for each degree of freedom 18 The rotary actuator is based on pairs of rolling diaphragm sealed cylinders balanced via timing belts as shown in Fig 1 B This current implementation is a passive master slave device with a 1 1 kinematic mapping Fig 2 B shows the complete device The 3 DOF biopsy device is designed with the actuators close to the base to minimize device inertia and to give the largest possible working volume while still meeting biopsy needle force insertion requirements We chose a fi ve bar pantograph mechanism for 2 DOF sagittal motion plus one degree of freedom at the base for yaw similar to the Phantom haptic device architecture 19 The rotation of each DOF is referred to as 0 1 and 2 as shown in Fig 1 A The two actuators that constitute 0are grounded and use a hard hydraulic line to maximize stiffness in that joint A previous version of our device used fi ber reinforced rubber hydraulic hoses Unfortunately these hoses expanded with changes in pressure and dragged against the device during operation interfering with haptic transparency The M1m2 M2 rcm2 r2 m1 r1 rcm1 B A M2 M1 0 0 Fig 3 color A Low inertia 2 DOF counterweight principle for the sagittal plane B Tungsten blocks 19 3 g cm3 are used as counterweights for the upper arm M1 and forearm M2 current version of our 3 DOF biopsy device uses aramid reinforced tefl on lined hoses Goodridge 910 and discrete hydraulic swivels at the base for the 1and 2DOFs as shown in Fig 2 A which reduces friction eliminates hose dragging and increases haptic transparency higher hose expansion stiffness The hard line used is aluminum tubing with 8 63 mm inside diameter The fi ve bar linkage arm is made of carbon fi ber tubes and 3D printed parts to reduce weight increase stiffness of the overall system and assure MR compatibility This version of the 3 DOF biopsy device is fully instrumented encoders pressure sensors for testing and is thus not MR safe however a fully MR conditional version of the transmission actuators was designed fabricated and tested The maneuverability provided by our design maximizes the working volume Fig 4 exceeding the 150 mm needle stroke requirement in all directions A MR Conditional Hydrostatic Actuators We modifi ed the hydrostatic rotary actuator presented in 14 to make it fully MR conditional The redesigned actua tor is made with non magnetic materials using aluminum for its body titanium and brass fasteners ceramic ball bearings and kevlar cord timing belts Fig 1 B The actuator has a range of motion of 135 and a maximum continuous torque rating of 4 5 Nm MRI tests performed on this actuator are discussed in section III B Parallel Mechanism Counterweights The pantograph parallel mechanism used for sagittal plane motion permits the top link to be counterweighted at the base Fig 3 A The orientation of the top link is refl ected on the bottom link therefore having a counterweight at the bottom M2counterbalances the forearm Attaching the counterweights at the base of the parallel mechanism is the 0 200 100 200 100 0 0 100200 200 mm 440 mm xe ze ye saggital plane Fig 4 color Range of motion of the manipulator measured via joint encoders mapped to the driving end of the biopsy needle option with the least added inertia In practice there is a small imbalance as the upper arm center of mass does not lie directly between the shoulder and elbow joints due to the mass of the elbow connecting rod The 3 DOF biopsy device counterweights are made of tungsten cubes due to their high density 19 3 g cm3 They are each attached to the shafts of the master side actuators Fig 3 B counterweighting both the master and slave side simultaneously The compact high mass counterweights with a short moment arm minimize the added inertia to the link they balance Operation with and without counterweights is shown in the supplemental video C Needle Operation At the end effector on the slave side a passive 2 DOF swivel mechanism holds a biopsy needle Fig 1 A The operator has the freedom to change both the angle and stroke of the biopsy needle by manipulating the master side A needle plate on the master and slave side assures needle alignment We used phantom tissue on the slave side to test the performance on mock biopsy procedures The needle plate is a grid template use for MRI guided transperineal prostate biopsy made from an acrylic block 100 x120 x25 mm with a grid of 1 3 mm holes spaced 5 mm apart where the biopsy needles are inserted mirroring the guide plates used clinically 20 D Device Kinematics Kinematic analysis is important to understand the range of motion and fi ne manipulation of the end effector Fig 4 as well as to perform proper planning of movement For ward kinematic analysis allows us to fi nd a transformation between the position and orientation of the end effector p xe ye ze T with respect to the joint angles q 0 1 2 T For the purpose of this analysis the coordinate axes are fi xed as shown in Fig 1 A The device home B daisy chain air preload lines water lines slave sidemaster side 0 1 2 p0 p1 p2 0 p0 A torque sensor handlepressure sensors ppre ppre Fig 5 color A 1 DOF experimental setup B 3 DOF actuator setup with individual water lines for each DOF and a common daisy chain air preload line Each water line is fi tted with a differential pressure sensor position is when all the angles of the parallel mechanism are fi xed at 90 According to 21 the forwards kinematic map is given as xe ye ze s0 L1c1 L2s2 L2 L2c2 L1s1 L1 c0 L1c1 L2s2 1 Where si sin i and ci cos i with L1and L2 as upper arm and forearm lengths respectively Our 3 DOF biopsy device has a workspace of 480 mm in the frontal axis 365 mm in the longitudinal axis and 265 mm in the sagittal axis III EXPERIMENTS AND RESULTS We have fully instrumented the current 3 DOF biopsy de vice with 40 000 count optical encoders US Digital E6 post quadrature and attached 3 differential bidirectional pressure sensors Omega MM series to each degree of freedom Fig 2 Tests documented below were taken with an air preload of 500 kPa We used a NI USB 6341 DAQ unit and developed custom MATLAB programs to record and analyze data A Stiffness and Hysteresis For the characterization of the device we tested the 1 DOF transmission testbed fi rst Fig 5 A A torque sensor Omega TQM202 series was mounted to the master side actuator This enabled us to record torque and pressure simultaneously Fig 6 shows pressure torque and position tracking when the system moves freely unblocked with a stroke of approximately 32 To measure transmission series stiffness for the 1 DOF transmission testbed we blocked the slave side shaft and mounted the same torque sensor fi tted with a 180 mm long acrylic handle to the master side Fig 5 A The handle was Fig 6 color Result from manipulating the open loop 1 DOF system The plots of pressure vs position and torque vs position perfectly match Best fi t line T 0 0048P 0 0122 Fig 7 color Top 1 DOF blocked work loop Bottom Linear mapping between torque and pressure measurements The black plot is the best fi t line of torque vs pressure T 0 0026P 0 0116 pushed back and forth over a distance of approximately 15 mm Fig 7 The stiffness of the transmission was measured to be 28 64 N m rad Using the data obtained from the pressure and torque sensor we found a linear mapping between pressure and torque Fig 7 for the blocked system which corresponds with the diaphragm s specifi ed pressure area of 269 mm2and a timing belt pitch radius of 10 5 mm To compare the stiffness and hysteresis of the 1 DOF transmission tested with the 3 DOF biopsy device we mea sured the work loops in each case over the same stoke Fig 8 For the 1 DOF transmission testbed the 180 mm handle was manually operated with a 0range of 60 For the 3 DOF biopsy device the angles of the four bar linkage arm were fi xed at 90 and the linkage was manually operated with a 0range of 60 It was found that hysteretic loses for 40 2002040 Position deg 0 15 0 1 0 05 0 0 05 0 1 0 15 Torque N m 1 DOF 3 DOF Fig 8 color Free motion workloop The blue plot is the result of rotating the bottom actuator with a 180 mm hadle for the 1 DOF case The red plot is the result of rotating the bottom actuator with the parallel mechanism for the 3 DOF case Ideal systems would exhibit horizontal torque traces with zero torque and zero work loop area the 1 DOF transmission testbed are in the order of 0 085 N and for the 3 DOF biopsy device are 0 44 N 2 of the 20N maximum driving force This is reasonable as it takes more torque to operate the 3 DOF system because of the added friction of multiple actuators hoses and swivels The vertical stiffness of the 3 DOF biopsy device was measured by clamping the master side on its home position of 90 Fig 1 A and stacking 5mm plates with the end ef fector moving upwards The torque of the actuators increased linearly as the plates were inserted After inserting three plates it was found that the vertical stiffness of the system is 380N m The lateral stiffness was also measured to be 380N m using a similar method This stiffness is lower than previous work 11 1700 N m but we could if needed increase endpoint stiffness by reducing workspace range a factor of three is possible while maintaining our stroke requirement and without resorting to a clutching approach We plan to explore the impact of transmission stiffness on