A Perspective on Medical Robotics

INVITED PAPER

1 A Perspective on

3 Robots are reducing surgeon hand-tremor, assisting in spine and joint-replacement,

4 positioning surgical needle guides, and coordinating medical imaging

5 with surgical procedures.

6 By Russell H. Taylor, Fellow IEEE

7 ABSTRACT | This paper provides an overview of medical as industrial production, inspection and quality control, 34 8 robotics, from the perspective of a researcher who has been laboratory automation, exploration, field service, rescue, 35 9 actively involved in the field for 17 years. Like all robot systems, surveillance, and (as discussed below) medicine and health 36 10 medical robots fundamentally couple information to physical care. Historically, robots have often been first introduced 37 11 action to significantly enhance humans’ ability to perform to automate or improve discrete processes, such as 38 12 important tasksVin this case surgical interventions, rehabili- painting a car or placing test probes on electronic circuits, 39 13 tation, or simply helping handicapped people in daily living but their greatest economic influence has often come 40 14 tasks. Research areas include modeling and analysis of indirectly as essential enablers of computer-integration of 41 15 anatomy and task environments, interface technology between entire production or service processes. 42 16 the Bdata world[ and the physical world, and study of how As this paper will argue, medical robots have a similar 43 17 complex systems are put together. This paper will discuss these potential to fundamentally change interventional medicine 44 18 research areas and illustrate their interrelationship with as enabling components in much broader computer- 45 19 application examples. Although the main focus will be on integrated systems that include diagnosis, preoperative 46 20 robotic systems for surgery, it will also discuss the relationship planning, perioperative and postoperative care, hospital 47 21 of these research areas to rehabilitation and assistance robots. logistics and scheduling, and long-term follow-up and 48 22 Finally, it will include some thoughts on the factors driving the quality control. Within this context, surgical robots and 49 23 acceptance of medical robotics and of how research can be robotic systems may be thought of as Bsmart[ surgical tools 50 24 most effectively organized. that enable human surgeons to treat individual patients 51 with improved efficacy, greater safety, and less morbidity 52 25 26 KEYWORDS | Computer-integrated surgery; human–machine than would otherwise be possible. Further, the consistency 53 27 cooperative systems; medical robotics; rehabilitation robotics; and information infrastructure associated with medical 54 28 robotic assistive systems; surgical assistants; telerobotics; robotic and computer-assisted surgery systems has the 55 29 telesurgery IEEEpotential to make Bcomputer-integrated surgery[ as 56 important to health care as computer-integrated manufac- 57 turing is to industrial production. 58 30 I. INTRODUCTION 59 This paper is not intended to be a survey, in the 31 The ability of robotic systems to couple information to traditional sense. Other papers in this special issue provide 60 32 physical action in complex ways has had a profound a comprehensive overview of major technology themes in 61 33 influence on our society. Applications include such fields medicalProof robotics, as well as related work on robotic sys- 62 tems for rehabilitation and human assistance. Other sur- 63 veys may be found in a recent IEEE TRANSACTIONS ON 64 65 Manuscript received September 26, 2005; revised February 7, 2006. This work ROBOTICS special issue on medical robotics [1], [2] and was supported in part by the National Science Foundation under ERC Grant elsewhere (e.g., [1], [3]–[5]). 66 EEC9731478, in part by the National Institutes of Health, in part by the National 67 Institute of Science and Technology, in part by the Whitaker Foundation, in part by IBM, Instead, the goal is to provide a perspective on how in part by Siemens Corporation, in part by General Electric, in part by Northern Digital, surgical, rehabilitation, and assistive robots relate to 68 in part by Intuitive Surgical Systems, and in part by Integrated Surgical Systems. AQ1 69 The author is with the Department of Computer Science, Johns Hopkins University, broader themes of computation, interface technology, Baltimore, MD 21218 USA (e-mail: [email protected]). and systems. This perspective is informed, first, by the dis- 70 Digital Object Identifier: 10.1109/JPROC.2006.880669 cussion and experience reported in many workshops over 71

0018-9219/$20.00 2006 IEEE Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 1 Taylor: A Perspective on Medical Robotics

Table 1 Complementary Strengths and Limitations of Robots and such as Bno fly zones[ preventing robots from moving tools 100 Humans [4] into dangerous proximity to delicate anatomical structures. 101 A third advantage is the inherent ability of medical 102 robots and CIS systems to promote consistency while 103 capturing detailed online information for every procedure. 104 This Bflight data recorder[ information can be invaluable in 105 mortality and morbidity assessments of serious incidents, 106 but the true potential is much more far-reaching. Poten- 107 tially, statistical analysis comparing outcome measures to 108 procedure variables may produce both better understand- 109 ing of what is most important to control and, ultimately, to 110 safer and more effective interventions. This data can also be 111 a valuable tool for training, skill assessment, and certifi- 112 cation for surgeons. 113 Similarly, robotic systems for rehabilitation or for 114 assistance in daily living must offer real advantages if they 115 are to be adopted. Once again, acceptance will come from 116 exploiting the complementary abilities of humans (who may 117 have disabilities) and machines to accomplish tasks that 118 might not otherwise be feasible or practical for unassisted 119 humans. Typical benefits may include more efficient or 120 consistent performance of exercise following injury or 121 surgery; partial restoration of function through Bintelligent[ 122 prostheses, either for long-term use or during recovery; and 123 cooperative aids for our aging population. Acceptance in 124 72 the past fifteen years (e.g., [6]–[10]); and, second, by my these areas will also be crucially dependent on economic 125 73 own experiences at IBM Research and at Johns Hopkins and social factors such as cost, ruggedness, ease of use, and 126 74 University (JHU). My primary focus will be on medical human–machine communication capabilities. 127 75 robotics and computer-integrated surgery (CIS) systems, 76 which have been the major focus of my own research over B. Surgical CAD/CAM 128 77 the past 17 years. However, there are important synergies The basic information flow of CIS systems is illustrated 129 78 betweenroboticsforCISandforsuchfieldsasrehabili- in Fig. 1. Preoperative planning typically starts with two- 130 79 tation and assistance for elderly or handicapped people, dimensional (2-D) or three-dimensional (3-D) medical 131 80 and I will touch on these related areas as well. images, together with information about the patient. These 132 images can be combined with general information about 133 human anatomy and variability to produce a computer 134 81 II. BASIC SYSTEM CONCEPTS: MEDICAL model of the individual patient, which is then used in 135 82 ROBOTICS IN COMPUTER-INTEGRATED surgical planning. In the operating room, this information 136 83 SURGERY AND REHABILITATION

84 A. Factors Driving Acceptance of MedicalIEEE Robotics 85 Just as with manufacturing robots, medical robots and 86 CIS systems must provide real advantages if they are to be 87 accepted and widely deployed. 88 First, and perhaps most obvious, is the ability of 89 computer-integrated systems to significantly improve sur- 90 geons’ technical capability, either by making existing pro- Proof 91 cedures more accurate, faster, or less invasive or by making 92 it possible to perform otherwise infeasible interventions. In 93 these cases, the advantages often come from exploiting the 94 complementary strengths of humans and robotic devices, as 95 summarized in Table 1. A second, closely related, advantage 96 is the potential of computer-integrated systems to promote 97 surgical safety by: 1) improved technical performance of 98 difficult procedures; 2) on-line monitoring and informa- 99 tion supports for surgical procedures; and 3) active assists Fig. 1. The information flow of CIS systems.

2 Proceedings of the IEEE |Vol.94,No.9,September2006 Taylor: A Perspective on Medical Robotics

137 is registered to the actual patient using intraoperative The second variety, auxiliary surgical supports, generally 192 138 sensing, which typically involves the use of a 3-D work side-by-side with the surgeon and perform such func- 193 139 localization, x-ray or ultrasound images, or the use of the tionsasendoscopeholdingorretraction.Thesesystems 194 140 robot itself. If necessary, the surgical plan can be updated, typically provide one or more direct control interfaces such 195 141 and then one or more key steps in the procedure are as joysticks, head trackers, voice control, or the like. How- 196 142 carried out with the help of the robot. Additional images or ever, there have been some efforts to make these systems 197 143 sensing can be used to verify that the surgical plan is Bsmarter[ so as to require less of the surgeon’s attention 198 144 successfully executed and to assist in postsurgical follow- during use, for example by using computer vision to keep 199 145 up. The coupling of imaging, patient-specific models, and the endoscope aimed at an anatomic target or to track a 200 146 computer-controlled delivery devices can significantly surgical instrument. Their value is assessed using the same 201 147 improve both the consistency of therapy delivery and the measures as for surgeon extenders, though often with 202 148 data available for patient follow-up and statistical studies greater emphasis on surgical efficiency. 203 149 required to develop and validate new therapies. 150 We refer to the process of building a model of the D. Rehabilitation and Assistive Systems 204 151 patient, planning, registration, execution, and follow-up as As our population ages, robotic systems for rehabilita- 205 152 surgical CAD/CAM, stressing the analogy with computer- tion and for helping deal with physical and cognitive 206 153 integrated manufacturing. Typical examples of robotic disabilities will become more and more important [7]. 207 154 surgical CAD/CAM are discussed in Section IV. The ad- Broadly, we can identify four areas of great promise: 208 155 vantages provided by robotic execution in surgical CAD/ 1) systems assisting with physical therapy following in- 209 156 CAM depend somewhat on the individual application, juries or surgery; 2) Bsmart[ prosthetic devices; 3) systems 210 157 but include: 1) accurate registration to medical images; designed to help disabled people in daily living activities; 211 158 2) consistency; 3) the ability to work in imaging envi- and 4) systems designed to help prevent or ameliorate 212 159 ronments that are not friendly to human surgeons; and cognitive and emotional decline. 213 160 4) the ability to quickly and accurately reposition instru- 161 ments through complex trajectories or onto multiple targets. 214 162 In addition to the technical issues inherent in cons- III. THE STRUCTURE AND TECHNOLOGY 215 163 tructing systems that can provide these advantages, one of OF MEDICAL ROBOTIC SYSTEMS 164 biggest challenges is finding ways to reduce the setup Fig. 2 shows the block diagram of a typical CIS system. 216 165 overhead associated with robotic interventions. A second These systems work cooperatively with humans (surgeons 217 166 challenge is to provide a modular family of low-cost robots and other medical personnel) to couple information with 218 167 and therapy delivery devices that can be quickly configured action in the physical world to perform tasks. Broadly, 219 168 into fully integrated and optimized interventional systems research supporting these systems comprises three areas: 220 169 for use with appropriate interventional imaging devices for computer-based modeling and analysis of images, patient 221 170 a broad spectrum of clinical conditions with convenience anatomy, and surgical plans; interface technology relating 222 171 comparable to current outpatient diagnostic procedures. the Bvirtual reality[ of computer models to the Bactual 223 reality[ of the patient, operating room, and surgical staff; 224 172 C. Surgical Assistants 173 Surgery is a highly interactive process and many sur- 174 gical decisions are made in the operating room. The goal of 175 surgical robotics is not to replace the surgeon with a robot, 176 but to provide the surgeon with a new set ofIEEE very versatile 177 tools that extend his or her ability to treat patients. We thus 178 often speak of medical robot systems as surgical assistants 179 that work cooperatively with surgeons. A special subclass of 180 thesesystemsareoftenusedforremotesurgery. 181 Currently, there are two main varieties of surgical 182 assistant robot. The first variety, surgeon extenders,are Proof 183 operated directly by the surgeon and augment or supple- 184 ment the surgeon’s ability to manipulate surgical instru- 185 ments in surgery. The promise of these systems, broadly, is 186 that they can give even average surgeons superhuman 187 capabilities such as elimination of hand tremor or ability to 188 perform dexterous operations inside the patient’s body. The 189 value is measured in: 1) ability to treat otherwise 190 untreatable conditions; 2) reduced morbidity or error 191 rates; and 3) shortened operative times. Fig. 2. Block diagram of typical CIS system.

Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 3 Taylor: A Perspective on Medical Robotics

• real-time data fusion for such purposes as updating 258 models from intraoperative images; 259 • methods for human–machine communication, 260 including real-time visualization of data models 261 and natural language understanding, gesture rec- 262 ognition, etc.; 263 • methods for characterizing uncertainties in data, 264 models, and systems and for using this informa- 265 tion in developing robust planning and control 266 methods. 267 Of course, these themes are highly interrelated and 268 mutually supportive. For example, modern medical image 269 segmentation methods are intimately associated with 270 registration methods. 271 Statistical methods have long been important in 272 Fig. 3. The daVinci surgical robot uses mechanically constrained medical robotics, and their importance is increasing. 273 remote-center-of-motion arms manipulate modular tools under surgeon teleoperator control. The tools use cable drives Examples include the following. 274 to provide high-dexterity manipulation inside the body. • Construction of statistical Batlases[ characterizing 275 (Photo: Intuitive Surgical Systems). anatomic variation over large populations of 276 AQ2 individuals. Such atlases provide a natural frame- 277 work for consolidating a wide variety of informa- 278 tion about disease states, biomechanical modeling 279 225 and systems science and analysis permitting these compo- results, surgical plans, outcomes, etc. 280 226 nents to be combined in a modular and robust way with • Methods for Bdeformably[ registering atlases to 281 227 safe and predictable performance. individual patient images to produce Bmost pro- 282 bable[ patient models, based on available informa- 283 228 A. Modeling and Analysis tion. Such models also naturally incorporate prior 284 229 As medical robotic systems evolve, computational information about possible treatment plans, bio- 285 230 modeling and analysis will become more and more im- mechanical simulations, expected outcomes, etc. 286 231 portant. There is a robust and diverse research community • Methods for correlating information about treat- 287 232 spanning an equally broad range of research topics and ment plans and actual procedure execution with 288 233 techniques. outcome variation, in order to identify key factors 289 234 The core challenge is to develop computationally ef- affecting outcomes and safety. 290 235 ficient methods for constructing models of individual Fig. 4 shows one typical example of the use of these tech- 291 236 patients and populations of patients from a variety of data niques for brain tumor treatment planning. Other exam- 292 237 sources and for using these models to help perform useful ples include statistical modeling of the location of prostate 293 238 tasks. A related challenge is modeling the tasks themselves cancer based on histology specimens, followed by deform- 294 239 and the environment in which the tasks are performedV able registration of this atlas to ultrasound or MRI images to 295 240 whether the operating room, intensive care facility, clinic, 241 or home. Some common themes include: 242 • medical image segmentation and imageIEEE fusion to 243 construct and update patient-specific anatomic 244 models; 245 • biomechanical modeling for analyzing and pre- 246 dicting tissue deformations and functional fac- 247 tors affecting surgical planning, control, and 248 rehabilitation; Proof 249 • optimization methods for treatment planning and 250 interactive control of systems; 251 • methods for registering the Bvirtual reality[ of 252 images and computational models to the Bphysical 253 reality[ of an actual patient; 254 • methods for characterizing treatment plans and indi- 255 vidual task steps such as suturing, needle insertion, Fig. 4. Patient-specific model of a brain tumor patient based on 256 or limb manipulation for purposes of planning, a deformable registration to a statistical atlas incorporating 257 monitoring, control, and intelligent assistance; finite-element simulations and functional data [99], [100].

4 Proceedings of the IEEE |Vol.94,No.9,September2006 Taylor: A Perspective on Medical Robotics

296 create optimized patient-specific biopsy plans [11], [12], the 297 use of statistical atlases to create 3-D bone models from 2-D 298 x-ray images [13]–[15], and statistical analysis of procedure 299 variability in hip arthroplasty [16], [17]. 300 Although the use of patient-specific models for reha- 301 bilitation planning has so far been relatively limited, the 302 potential for applying similar atlas-based techniques 303 incorporating biomechanical simulations has great future 304 potential. Fig. 5. Autonomous motion inside the body. Left: CMU HeartLander 305 B. Interface Technology [44]. Right: S. Supiore Sant’Anna endoluminal robot [101]. 306 Robotic systems inherently involve interfaces between 307 the data world of computers and the physical world. One 308 consequenceofthisisthatroboticshasalwaysbeena 2) Teleoperation and Hands-On Control: Many surgical 350 309 highly interdisciplinary field involving many branches of robots (e.g., [24], [25], [31], [35]) are teleoperated. Two 351 310 engineering research. For medical robotics, the core potential drawbacks of this approach are that more 352 311 challenge is to fundamentally extend the sensory, motor, equipment (i.e., a Bmaster[ control station) is needed, 353 312 and human-adaptation abilities of robotic systems in an and the surgeon is often somewhat removed from the 354 313 unusually demanding and constrained environment. operating table, thus necessitating significant changes in 355 surgical work flow. Early experiences with ROBODOC 356 314 1) Specialized Mechanism Design: Early medical robots [21] and other surgical robots (e.g., [50], [51]) showed that 357 315 (e.g., [18]–[23]) frequently employed conventional indus- surgeons found a form of Bhands-on[ admittance control, 358 316 trial manipulators, usually with modifications for safety in which the robot moved in response to forces exerted by 359 317 and sterility. Although this approach had many advantages the surgeon directly on the surgical end-effector, to be very 360 318 and is still frequently taken for laboratory use or rapid convenient and natural for surgical tasks. Subsequently, a 361 319 prototyping, surgery and rehabilitation applications im- number of groups have exploited this idea for precise 362 320 pose special requirements for workspace, dexterity, surgical tasks, notably the JHU BSteady Hand[ microsur- 363 321 compactness, and work environment. Consequently, the gical robot [52] and the Imperial College Acrobot 364 322 trend has been more and more toward specialized designs. orthopedic system [53]. These systems (Fig. 8) provide 365 323 For example, laparoscopic surgery and percutaneous very high stiffness and precision and eliminate physiolog- 366 324 needle placement procedures typically involve passage or ical tremor while still permitting the surgeon to exploit his 367 325 manipulation of instruments about a common entry point or her natural kinesthetic sense and eye-hand coordina- 368 326 into the patient’s body. In response, two basic designs have tion. Other groups have developed completely freehand 369 327 been widely used. The first, adopted by the Aesop and Zeus instruments that sense and actively cancel physiological 370 328 robots [24], [25], uses a passive wrist to allow the 329 instrument to pivot about the insertion point. The second, 330 adopted by a variety of groups, using a variety of design 331 approaches (e.g., [26]–[33]), mechanically constrains the 332 motion of the surgical tool to rotate about a Bremote center 333 of motion (RCM)[ distal to the robot’s structure. 334 A second problem is the need to provideIEEE high degrees 335 of dexterity in very constrained spaces inside the patient’s 336 body, and at smaller and smaller scales. Typically, the 337 response has been to develop cable actuated wrists (e.g., 338 [34], [35]. However, the difficulties of scaling these 339 designs to very small dimensions have led some groups to 340 investigate bending structural elements (e.g., [36]–[39]), Proof 341 shapememoryalloys[40],microhydraulics[41]orother 342 approaches (see Fig. 6). The problem of providing access to 343 surgical sites inside the body has led several groups to 344 develop semiautonomously moving robots for epicardial or 345 endoluminal applications (e.g., [2], [42]–[44], Fig. 5). 346 The necessity of providing robots that conform to Fig. 6. Dexterity inside the body. (A) daVinci wrist. (B) Waseda 347 human biomechanics and physical constraints has similarly dual-arm end effector for MIS flexible endoscopy [102]. 348 led researchers to develop specialized designs for rehabil- (C) JHU bendable Bsnake[ robot [36]–[39]. (D) Five degrees of 349 itation or assistive applications (e.g., [45]–[49], Fig. 7). freedom, 3-mm-diameter microcatheter robot [103], [104].

Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 5 Taylor: A Perspective on Medical Robotics

Both teleoperation and hands-on control are likewise 395 used in human–machine cooperative systems for rehabilita- 396 tion and disabilities assistance systems. Constrained hands- 397 on systems offer special importance for rehabilitation 398 applications and for helping people with movement dis- 399 orders. Similarly, teleoperation and Bintelligent[ task follow- 400 ing and control is likely to be vital for further advances in 401 assistive systems for people with severe physical disabilities. 402

4) Augmented Reality Interfaces: Once a surgical pro- 403 cedure has begun, a surgeon’s attention is necessarily fo- 404 cused on the patient’s anatomy. Traditionally, surgeons 405 have relied on their natural hand–eye coordination, either 406 Fig. 7. Typical rehabilitation robots. Left: Rehab. Institute of Chicago Manipulandum [49]. Right: Tsukuba Gaitmaster 2 [49]. with direct visualization or (more recently) using endo- 407 scopic video images. Additional information, such as medi- 408 cal image data, has traditionally been posted on a light box 409 somewhere in the operating room. The ability of the 410 371 tremor (e.g., [54], [55]). Still other groups have developed computer to coregister and visualize images, models, and 411 372 passive or semiactive mechanisms for assisting surgeons other task-specific data provides an opportunity to signif- 412 373 manipulate tools or body parts (e.g.,[56]–[59]). icantly improve the surgeon’s ability to assimilate and use 413 all this information. Accordingly, there has been significant 414 374 3) Human–Machine Cooperative Systems: Although one interest in creating Baugmented reality[ information dis- 415 375 goal of both teleoperation and hands-on control is often plays and in using interactive means such as laser pointers 416 376 Btransparency,[ i.e., the ability to move an instrument in surgical assistant systems (Fig. 9). Similar interfaces 417 377 freely and dexterously, the fact that a computer is actually have also been exploited in rehabilitation systems, for 418 378 controlling the robot’s motion creates many more possi- example, in directing a patient’s motions for exercise. 419 379 bilities. The simplest is a safety barrier or Bno-fly zone,[ in 380 which the robot’s tool is constrained from entering certain C. Systems Science 420 381 portions of its workspace. More sophisticated versions Medical robots are complex systems that necessarily 421 382 include virtual springs, dampers, or complex kinematic involve many interacting subsystems, including computa- 422 383 constraints that help a surgeon align a tool, maintain a tional processes, sensors, mechanisms, and human–machine 423 384 desired force, or perform similar tasks. The Acrobot system interfaces. As such, they share the same underlying needs for 424 385 shown in Fig. 8 represents a successful clinical application good system design and engineering practice: modularity, 425 386 of the concept, which has many names, of which Bvirtual well-defined interfaces, etc. However, the fact that they are 426 387 fixtures[ seems to be the most popular (e.g., [60]–[64]). A to be used in clinical applications or otherwise directly 427 388 number of groups (e.g., [65]–[67]) are exploring exten- interact with people imposes some unusual requirements. 428 389 sions of the concept to active cooperative control, in which The most obvious of these is safety. Although there may be 429 390 the surgeon and robot share or trade off control of the multiple valid approaches to robot safety in specific 430 391 robot during a surgical task or subtask. As the ability of circumstances, a few principles are common [68], [69]. 431 392 computers to model and Bfollow along[ surgical tasks The most important of these is redundancy: no single point of 432 393 improves, these modes will become moreIEEE and more failure should cause a medical robot to go out of control or 433 394 important in surgical assistant applications. Proof

Fig. 9. Typical augmented reality interfaces for CIS. Left: CMU image Fig. 8. Cooperatively controlled surgical robots. Left: The JHU BSteady overlay system [106]. Right: Osaka/Tokyo laser guidance Hand[ robot [105]. Right: the Acrobot orthopedic robot [53]. system [107]. See also Figs. 12 (upper left) and 13.

6 Proceedings of the IEEE |Vol.94,No.9,September2006 Taylor: A Perspective on Medical Robotics

434 endanger a patient. A second, and often equally important, easy to image in CT and x-ray fluoroscopy. These factors 487 435 principle is that the computer’s model of the task environ- have made orthopedics an important application domain in 488 436 ment must correspond accurately to the actual environment. the development of Surgical CAD/CAM. 489 437 This is especially important for robotic systems that execute One of the first successful surgical CAD/CAM robots 490 438 plans based on preoperative images. With careful design and was the ROBODOC system [21], [70], [71] for joint 491 439 implementation, it is possible to practically eliminate the replacement surgery, which was developed clinically by 492 440 possibility that the robot will somehow Brun away[ or make Integrated Surgical Systems from a prototype developed at 493 441 an inappropriate motion. But this does little good if the IBM Research in the late 1980s. Since this system has a 494 442 image, robot, and physical patient coordinate systems are not number of features found in other surgical CAD/CAM 495 443 correctly registered to each other. Similarly, it is vital to robots, we will discuss it in some detail. 496 444 ensure that the procedure is planned correctly and appro- In ROBODOC joint replacement surgery, the surgeon 497 445 priately. Surgical robots are not surgeons. They are surgical selects an implant model and size based on an analysis of 498 446 tools that must be used correctly by surgeons. Consequently, preoperative CT images and interactively specifies the 499 447 it is vital that the surgeon have a clear understanding of the desired position of each component relative to CT 500 448 capabilities and limitations of the robotic system. coordinates. In the operating room, surgery proceeds 501 449 It is important to realize that surgical robots can often normally up to the point where the patient’s bones are to 502 450 enhance patient safety. First, the robot can provide better be prepared to receive the implant. The robot is moved up 503 451 control over process parameters (force, precision, etc.) to the operating table, the patient’s bones are attached 504 452 that can affect outcomes. Second, the robot typically does rigidly to the robot’s base through a specially designed 505 453 not suffer from momentary lapses of attention, although fixation device, and the transformation between robot and 506 454 (of course) the human operator may. Third, a robotic CTcoordinatesisdeterminedeitherbytouchingmultiple 507 455 system can be programmed to include Bvirtual barriers[ points on the surface of the patient’s bones or by touching 508 456 preventing the surgical tool from entering a forbidden preimplanted fiducial markers whose CT coordinates have 509 457 region unless the surgeon explicitly overrides the barrier. been determined by image processing. 510 458 Achieving these advantages requires careful design and The surgeon hand guides the robot to an approximate 511 459 validation, as well as rigorous testing and validation. initial position using a force sensor mounted between the 512 460 Another safety-related issue of special concern to regu- robot’s tool holder and the surgical cutter held by the tool 513 461 latory bodies is careful documentation and rigorous proce- holder. The robot then cuts the desired shape while moni- 514 462 dures in development, testing, and maintenance of toring cutting forces, bone motion, and other safety 515 463 medical robots. Sterility and biocompatibility are of sensors. The surgeon also monitors progress and can inter- 516 464 specific concern for surgical robots; these considerations rupt the robot at any time. If the procedure is paused for any 517 465 can impose unusual design constraints, especially in choice reason, there are a number of error recovery procedures 518 466 of materials. availabletopermittheproceduretoberesumedorre- 519 467 Other systems considerations are of practical interest started at one of several defined checkpoints. Once the 520 468 mainly for researchers and developers. The most impor- desired shape has been cut, surgery proceeds manually in 521 469 tant of these is the vital importance of integration testbeds. the normal manner. 522 470 Medical systems, especially those involving image guid- 471 ance, are extremely difficult to develop without access to 472 all the pieces needed to do complete experiments.IEEE 473 IV. EXAMPLES OF MEDICAL 474 ROBOTICS SYSTEMS 475 A. Robotic Orthopedic Surgery 476 Geometric precision is often an important consideration 477 in orthopedic surgery. For example, orthopedic implants Proof 478 used in joint replacement surgery must fit properly and 479 must be accurately positioned relative to each other and to 480 the patient’s bones. Osteotomies (procedures involving cut- 481 ting and reassembly of bones) require that the cuts be made 482 accurately and that bone fragments be repositioned ac- 483 curately before they are refastened together. Spine surgery 484 often requires screws and other hardware to be places into Fig. 10. The ROBODOC system for hip and knee surgery was one 485 vertebrae in close proximity to the spinal cord, nerves, and of the first successful applications of robotics to surgical 486 important blood vessels. Further, bone is rigid and relatively CAD/CAM [21], [70], [71].

Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 7 Taylor: A Perspective on Medical Robotics

delivering the therapy through a series of percutaneous 558 access steps, assessing what was done, and using this 559 feedback to control therapy at several time scales. The 560 ultimate goal of current research is to develop systems that 561 execute this process with robotic assistance under a variety 562 of widely available and deployable image modalities, 563 including ultrasound, x-ray fluoroscopy, and conventional 564 MRI and CT scanners. 565 Current work at JHU and related work at Georgetown 566 University is typical of this activity. This approach has 567 emphasized the use of Bremote center-of-motion[ (RCM) 568 manipulators to position needle guides under real-time 569 image feedback. This work has led to development of a 570 modular family of very compact robotic subsystems 571

Fig. 11. Parallel link robots that attach directly to the patient’s [79]–[83] optimized for use in a variety of imaging en- 572 bone. The left system [108] is used for hip surgery and the vironments, as well as a simple image overlay device for 573 right system [109] is intended for spine surgery. use in CT environments (Fig. 12). These devices have been 574 used clinically at JHU and have been evaluated for spine 575 applications at Georgetown [84], [119]. Many other groups 576 have also investigated the use of robotic devices with real- 577 523 After preclinical testing demonstrated an order-of- time x-ray and CT guidance, including [85]–[87]. 578 524 magnitude improvement in precision over manual surgery, Anumberofgroupshavedevelopedroboticdevicesfor 579 525 the system was applied clinically in 1992 for the femoral use with ultrasound-guided (e.g., [88]–[90]) and MRI- 580 526 implant component in primary total hip replacement guided (e.g., [91]–[94] needle placement. In-MRI systems 581 527 (THR) surgery. Subsequently, it has been applied success- represent both an unusual challenge and an unusual 582 528 fully to both primary and revision THR surgery, as well as opportunity for medical robotics. On the one hand, the 583 529 knee surgery [72]–[74]. strong magnetic fields and very stringent electrical noise 584 530 A number of other robotic systems for use in joint requirements of MRI significantly limits design flexibility. 585 531 replacement surgery were subsequently proposed, includ- On the other hand, MRI imaging offers unprecedented 586 532 ing the CASPAR system [75], which was very similar to tissue discrimination and imaging flexibility that can be 587 533 ROBODOC, and the cooperatively guided Acrobot [53] exploited by compact robots operating inside the scanner 588 534 (Fig. 8, right). More recently, several groups have environment. 589 535 proposed small parallel-link robots attaching directly to 536 the patient’s bones (Fig. 11). Similarly, there has been 537 extensive progress in so-called surgical navigation for 538 orthopedics (e.g., [8], [76], [77]), in which the surgeon 539 manipulatestoolsfreehandwhileacomputergenerates 540 corresponding displays based on 2-D or 3-D images. 541 One significant consequence of the ability of medical 542 robots and navigation systems to help surgeons carry out 543 plans accurately is that the planning itself becomesIEEE more 544 valuable. In turn, this has increased the potential 545 importance of 3-D modeling of bone from 2-D and 3-D 546 images, finite-element biomechanical analysis, and meth- 547 ods for image-based real-time registration of bone models 548 to x-ray and ultrasound images.

549 Proof B. Robotically Assisted Percutaneous Therapy 550 One of the first uses of robots in surgery was posi- 551 tioning of needle guides in stereotactic neurosurgery 552 [18], [20], [78]. This is a natural application, since the 553 skull provides rigid frame-of-reference. However, the Fig. 12. Needle placement under image guidance. Top left: in-CT 554 potential application of localized therapy is much broader. freehand placement with image overlay device [110]. Top right: x-ray 555 Percutaneous therapy fits naturally within the broader guided nephrostomy needle placement [81]–[83]. Bottom: in-CT 556 paradigm of Surgical CAD/CAM systems. The basic pro- kidney biopsy with fiducial structure on needle driver to assist 557 cess involves planning a patient-specific therapy pattern, robot-to-scanner registration [111], [112].

8 Proceedings of the IEEE |Vol.94,No.9,September2006 Taylor: A Perspective on Medical Robotics

have great potential in rehabilitation and in providing 620 more general assistance in daily living activities for the 621 infirm or disables. Although this area is not strongly tied to 622 my own expertise and experience, I will discuss these 623 applications briefly here. An excellent survey and detailed 624 discussion of the relevant research problems may be found 625 in a recent of the International Advanced Robotics 626 Program workshop reporton medical robots [7]. 627 Interactive systems for physical therapy and rehabilita- 628 tion share many of the characteristics of surgical assistance. 629 Exercise robots such as those in Fig. 7 usually must come in 630 contact with the patient, and often must constrain the 631 patient as well. This raises obvious safety and ergonomic 632 challenges, which may also be viewed as research oppor- 633 tunities. The potential of these systems to customize indi- 634 vidual treatment plans based on patient-specific 635 biomechanical simulations and real-time monitoring of 636 Fig. 13. Some surgical assistant systems. Clockwise from upper left: patient performance represents a significant opportunity. 637 visual servoing of laparoscopic instrument relative to organ [113]; Another, longer term, opportunity is the possibility of 638 stereo video overlay of visually tracked laparoscopic ultrasound image 639 in daVinci robot [114]; image feedback controlled laparoscopic combining the capabilities of surgical systems with reha- ultrasound robot [115]; and laboratory setting for Bskill acquisition[ bilitation robots. In this scenario, a patient might present 640 for suturing and similar tasks [116]–[118]. with symptoms such as pain or mobility difficulties and 641 would receive suitable diagnostic tests and imaging to 642 permit a patient-specific model to be developed. An ap- 643 propriate therapy plan would be developed from this 644 590 C. Minimally Invasive Robotic Surgery model. If the plan includes surgery, the procedure would be 645 591 Teleoperated robots have been used for close to 15 years carried out with the assistance of appropriate technology, 646 592 to assist surgeons in minimally invasive procedures, such as robots or navigation aids. A customized and updated 647 593 first, to hold endoscopes or retractors (e.g., [24], [50], rehabilitation plan would then be executed, taking account 648 594 [95], [96]) and, later, to manipulate surgical instruments of any unexpected events in surgery. Patient progress 649 595 (e.g., [24], [25], [31], [35]). Although there have been would be monitored both to promote optimal recovery and 650 596 some spectacular long-distance demonstrations (e.g., [31], to provide statistical data correlating procedural variables 651 597 [32], [97], [98]) most uses still occur within a local (plan, patient anatomy, execution variability) with out- 652 598 operating room environment. Currently, practical clinical comes for the purposes of improving overall care. 653 599 useismoreorlesslimitedtosurgeonextenderuses,in One of the most rewarding things for any engineer is to 654 600 which the robot mimic the surgeon’s hand motions, and develop technology and systems that directly help people. 655 601 the surgeon relies on visual information from endoscopic Surgical and rehabilitation robots clearly fulfill this crite- 656 602 cameras for feedback. rion. But systems helping individuals with severe disabi- 657 603 However, sufficient progress in modeling and analysis lities, such as the assistive robot in Fig. 14 (left) are in a 658 604 has now been made so that medical robots with special category. These systems pose many of the same 659 605 characteristics of true surgical assistants. SomeIEEE examples human–machine research challenges as surgical assistants, 660 606 are shown in Fig. 13. Research challenges associated with 607 the development of more capable assistants include: 608 1) interactive fusion of preoperative models with real-time 609 images and manipulator feedback, especially for deforming 610 organs; 2) development of ways to describe surgical tasks 611 and task steps that can be related in real time to these 612 Proof models; 3) development of effective menus of assistive 613 capabilities that can be adapted to these models in real 614 time; and 4) development of ways for the computer to 615 Bfollow-along[ the progress of the procedure so as to offer 616 exactly the most appropriate assistance at any given time.

617 D. Rehabilitation and Assistance in Daily Living 618 Although this paper has focused on robotic systems for Fig. 14. Robots for assistance in daily living. Left: exact dynamics 619 surgery and other direct medical interventions, robots also assistive robot manipulator [49]. Right: CMU Nursebot [49].

Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 9 Taylor: A Perspective on Medical Robotics

661 such as modeling tasks and work environments, under- world[ to the physical world of patients and clinicians, and 701 662 standing human intentions, providing meaningful assis- to the system science that makes it possible to put every- 702 663 tance and feedback without being unduly intrusive, etc. thing together safely, robustly, and efficiently. Progress in 703 664 There are some obvious differences, as well. In particular, these areas will most fruitfully be made within the context 704 665 better means must be found to develop patient-specific of well-defined applications or families of application. 705 666 human–machine interface, while at the same time finding Careful attention must also be paid to the advantages that 706 667 common elements that can be standardized on. Over time, the robotic subsystem will provide, at least potentially, 707 668 research on methods of direct coupling of systems to brain within the larger context of the application and hospital, 708 669 and nerve signals to robots and sensors will enable a new, clinic, or home environment in which it will be deployed. 709 670 more capable generation of prostheses and assistive de- Academic researchers, such as the author of this paper, 710 671 vices. Interestingly, fitting of such devices to individual can contribute to progress in these areas, but we cannot do 711 672 patients may be enabled by more precise and delicate it alone. To an even greater extent than in other subspe- 712 673 surgical robots. cialties of robotics, industry has unique expertise that is 713 674 As our population ages, we will all become more absolutely essential for successful development and deploy- 714 675 susceptible to the physical and mental frailties that come ment of medical robot systems. Also, the surgeons who will 715 676 with growing older. This will inevitably pose enormous use these systems have unique insights into the problems to 716 677 challenges for our working population. Nursing and daily be solved and into what will and will not be accepted in the 717 678 living care personnel will be stretched thinner and thinner, operating room. All groups must work together for progress 718 679 and old people may become increasingly isolated. Robotic to be made, and they must work together practically from 719 680 systemssuchastheBNursebot[ in Fig. 14 (right) have the very beginning. Our experience has been that building a 720 681 significant potential to help improve both the ability of strong researcher/surgeon/industry team is one of the most 721 682 human care givers to help other people and of people challenging, but also one of the most rewarding aspects of 722 683 needing help to sustain independent lives. Progress is likely medical robotics research. The only greater satisfaction is 723 684 to be incremental, as general robotic capabilities improve. the knowledge that the results of such teamwork can have a 724 685 Conversely, as these systems become more important eco- very direct impact on patients’ health. Medical robotics 725 686 nomically, they are likely to serve as testbeds for developing research is very hard work, but it is worth it. h 726 687 a broad range of multiuse robotic capabilities.

Acknowledgment 727 688 V. PERSPECTIVES: WHITHER ARE Althoughthispaperisintendedasapersonalperspec- 728 689 WE TENDING AND HOW CAN tive on the field of medical robotics, this perspective has 729 690 WE GET THERE? necessarily been shaped by the author’s experiences in 730 691 In less than two decades, medical robotics has developed working with others. The author is grateful to all of these 731 692 from a subject of late-night comedy routines into a growing many colleagues and collaborators. Similarly, one notable 732 693 field engaging the attention of hundreds of active trend over the past several years has been the explosion of 733 694 researchers around the world. If work on related technical excellent work in the field. It is no longer possible to 734 695 areassuchasmedicalimageanalysisisincluded,thereare produce a truly inclusive survey. The author is acutely 735 696 thousands of researchers involved. conscious that much excellent work has gone uncited. To 736 697 This research is challenging, interdisciplinary, and sy- those who have been passed over, please accept the author’s 737 698 nergistic. Progress is needed across the board in the model- apologies. Finally, the author expresses appreciation to the 738 699 ing and analysis required for medical robotic applications,IEEEmany government and industry partners who have partially 739 700 for the interface technologies required to relate the Bdata funded some of the work reported here. 740 REFERENCES medical robotics,[ in Standard Handbook [7] T. Kanade, presented at the Int. AQ3 B of Biomedical Engineering and Design, Advanced Robotics Program Workshop [1] R. Taylor and D. Stoianovici, Medical M. Kutz, Ed. New York: McGraw-Hill, on Medical Robotics, Hidden Valley, robotic systems in computer-integrated 2003, pp. 29.3–29.45. PA, 2004. surgery,[.Probl. Gen. Surg., vol. 20, pp. 1–9, [5] R. H. Taylor, S. Lavallee, G. C. Burdea,Proof and [8] K. Cleary, Workshop report: Technical 2003. R. Mosges, Computer Integrated Surgery. requirements for image-guided spine [2] P. Dario, B. Hannaford, and A. Menciassi, Cambridge, MA: MIT Press, 1996. procedures (April 17–20, 1999), Georgetown BSmart surgical tools and augmenting [ [6] K. Curley, T. Broderick, R. Marchessault, Univ. Medical Center, Washington, DC, devices, .IEEE Trans. Robot. Automat., 1999, p. 113. vol. 19, no. 5, pp. 782–792, Oct. 2003. G. Moses, R. Taylor, W. Grundfest, B [ E. Hanley, B. Miller, A. Gallagher, and [9] R. H. Taylor, G. B. Bekey, and J. Funda, AQ4 [3] R. H. Taylor, Medical robotics, in M. Marohn, Integrated research team final presented at the NSF Workshop on Computer and Robotic Assisted Knee and Hip report: Surgical roboticsVThe next steps Computer Assisted Surgery, Surgery, A. DiGioia, B. Jaramaz, F. Picard, September 9–10 2004, Telemedicine and Washington, DC, 1993. and L. P. Nolte, Eds. Oxford, U.K.: Oxford Advanced Technology Research Center B Univ. Press, 2004, pp. 54–59. [10] R. H. Taylor and S. D. Stulberg, Medical (TATRC), U.S. Army Medical Research and robotics working group section report,[ [4] R. H. Taylor and L. Joskowicz, Materiel Command, Fort Detrick, MD, presented at the NSF Workshop Medical BComputer-integrated surgery and TATRC Rep. 04-03, Jan. 2005.

10 Proceedings of the IEEE |Vol.94,No.9,September2006 Taylor: A Perspective on Medical Robotics

Robotics and Computer-Assisted Medical to development,[.Surg. Endosc., vol. 8, enhanced miniaturization,[ in Proc. IEEE Interventions (RCAMI), Bristol, U.K., 1996. pp. 63–66, 1994. Int. Conf. Robotics and Automation, 2005, [11] D. Shen, Z. Lao, J. Zeng, W. Zhang, [25] H. Reichenspurner, R. Demaino, M. Mack, pp. 351–357. I. Sesterhenn, L. Sun, J. W. Moul, D. Boehm, H. Gulbins, C. Detter, B. Meiser, [39] N. Simaan, R. Taylor, A. Hillel, and P. Flint, E. H. Herskovits, G. Fichtinger, and R. Ellgass, and B. Reichart, BUse of the voice BMinimally invasive surgery of the upper C. Davatzikos, BOptimization of biopsy controlled and computer-assisted surgical airways: Addressing the challenges of strategy by a statistical atlas of prostate system ZEUS for endoscopic coronary artery dexterity enhancement in confined spaces,[ cancer distribution,[.Med. Image Anal., bypass grafting,[.J. Thorac. Cardiovasc. Surg., in Surgical RoboticsVHistory, Present vol. 8, pp. 139–150, 2004. vol. 118, no. 1, pp. 11–16, Jul. 1999. and Future Applications, R. Faust, Ed. [12] D. Shen, Y. Zhan, and C. Davatzikos, [26] B. Eldridge, K. Gruben, D. LaRose, J. Funda, Commack, NY: Nova, 2006. BSegmentation of prostate boundaries from S. Gomory, J. Karidis, G. McVicker, [40] K. Ikuta, M. Tsukamoto, and S. Hirose, ultrasound images using statistical shape R. Taylor, and J. Anderson, BA remote center BShape memory alloy servo actuator system model,[.IEEE Trans. Med. Imag., vol. 22, of motion robotic arm for computer assisted with electric resistance feedback and no. 4, pp. 539–551, Apr. 2003. surgery,[.Robotica, vol. 14, pp. 103–109, application for active endoscope,[ in [13] M. Fleute and S. Lavallee, BNonrigid 1996. Computer-Integrated Surgery, R. H. Taylor, 3-D/2-D registration of images using [27] J. Funda, K. Gruben, B. Eldridge, S. Gomory, S. Lavallee, G. Burdea, and R. Mosges, Eds. statistical models,[ in Proc. MICCAI ’99, and R. Taylor, BControl and evaluation of Cambridge, MA: MIT Press, 1996. pp. 138–147. a 7 axis surgical robot for laparoscopy,[ pp. 277–282. [14] J. Yao and T. Russell, BDeformable 2D-3D presented at the 1995 IEEE Int. Conf. [41] K. Ikuta, H. Ichikawa, K. Suzuki, and B medical image registration using a statistical Robotics and Automation, Nagoya, Japan. T. Yamamoto, Micro hydrodynamic pelvis model: Experiments and accuracy [28] D. Stoianovici, L. Whitcomb, J. Anderson, actuated multiple segments catheter for [ factors,[ IEEE Trans. Med. Imag., 2003. R. Taylor, and L. Kavoussi, BA modular safety minimally invasive therapy, in Proc. IEEE Int. Conf. Robotics and Automation, [15] O. Sadowsky, K. Ramamurthi, surgical robotic system for image-guided [ 2003, pp. 2640–2645. L. M. Ellingsen, G. Chintalapani, percutaneous procedures, in Proc. Medical J. L. Prince, and R. H. Taylor, Image Computing and Computer-Assisted [42] M. Carrozza, L. Lencioni, B. Magnani, B BAtlas-assisted tomography: Registration Interventions Conf. (MICCAI-98), pp. 404–410. S. D’Attanasio, and P. Dario, The of a deformable Atlas to compensate [29] R. H. Taylor, J. Funda, B. Eldridge, development of a microrobot system [ for limited-angle cone-beam trajectory,[ in K. Gruben, D. LaRose, S. Gomory, for colonoscopy, in Proc. 1st Joint Conf. Proc. 3rd IEEE Int. Symp. Biomedical Imaging: M. Talamini, L. R. Kavoussi, and Computer Vision, Virtual Reality and Macro to Nano (ISBI), 2006, pp. 1244–1247. J. Anderson, BA telerobotic assistant for Robotics in Medicine (CVRMed II) and [ Medical Robotics and Computer Assisted [16] D. Larose, L. Cassenti, B. Jaramaz, J. Moody, laparoscopic surgery, .IEEE Eng. Med. Surgery (MRCAS III), 1997, pp. 779–789. T. Kanade, and A. DiGioia, BPost-operative Biol. Mag., vol. 14, no. 3, pp. 279–287, measurement of acetabular cup position May/Jun. 1995. [43] L. Phee, A. Menciassi, S. Gorini, G. Pernorio, B using X-ray/CT registration,[ in Proc. 3rd [30] R. Taylor, P. Jensen, L. Whitcomb, A. Arena, and P. Dario, An innovative Int. Conf. Medical Image Computing and A. Barnes, R. Kumar, D. Stoianovici, locomotion principle for minirobots moving [ Computer-Assisted Intervention (MICCAI), P. Gupta, Z. Wang, E. deJuan, and in the gastrointestinal tract, in Proc. 2000, pp. 1104–1113. L. Kavoussi, BA steady-hand robotic IEEE Int. Conf. Robotics and Automation [ (ICRA 2002), pp. 1125–1130. [17] A. DiGioia, J. Moody, R. LaBarca, C. Nikou, system for microsurgical augmentation, and B. Jaramaz, BClinical measurements of in Proc. Medical Image Computing and [44] N. Patronik, C. Riviere, S. E. Qarra, and B acetabular component orientation using Computer-Assisted Interventions Conf. M. A. Zenati, The HeartLander: A novel surgical navigation technologies,[ in Proc. 1st (MICCAI), 1999, pp. 1031–1041. epicardial crawling robot for myocardial [ Annu. Meeting CAOS Int., 2001, p. 19. [31] M. Mitsuishi, T. Watanabe, H. Nakanishi, injections, in Proc. 19th Int. Congr. Computer Assisted Radiology and Surgery, [18] Y. S. Kwoh, J. Hou, E. A. Jonckheere et al., T. Hori, H. Watanabe, and B. Kramer, B 2005, pp. 735–739. BA robot with improved absolute positioning A telemicrosurgery system with accuracy for CT guided stereotactic brain colocated view and operation points [45] L. Zollo, S. Roccella, R. Tucci, B. Siciliano, surgery,[.IEEE Trans. Biomed. Eng., vol. 35, and rotational-force-feedback-free master E. Guglielmelli, M. C. Carrozza, and [ B no. 2, pp. 153–161, Feb. 1988. manipulator, in Proc. 2nd Int. Symp. Medical P. Dario, BioMechatronic design and Robotics and Computer Assisted Surgery, 1995, control of an anthropomorphic hand [19] J. M. Drake, M. Joy, A. Goldenberg, and pp. 111–118. for prosthetics and robotic application,[ D. Kreindler, BComputer- and robot-assisted presented at the 1st IEEE/RAS-EMBS resection of thalamic astrocytomas in [32] M. Mitsuishi, S. I. Warisawa, T. Tsuda, Int. Conf. Biomedical Robotics and children,[.Neurosurgery, vol. 29, pp. 27–31, T. Higuchi, N. Koizumi, H. Hashizume, and B Biomechatronics (BioRob 2006), Pisa, Italy. 1991. K. Fujiwara, Remote ultrasound diagnostic system,[ in Proc. IEEE Conf. Robotics and [46] L. Masia, H. I. Krebs, P. Cappa, and [20] P. Cinquin, J. Troccaz, J. Demongeot, Automation, 2001, pp. 1567–1574. N. Hogan, BWhole-arm rehabilitation S. Lavallee, G. Champleboux, L. Brunie, following stroke: Hand module,[ presented F. Leitner, P. Sautot, B. Mazier, A. Perez, [33] S. E. Salcudean, W. H. Zhu, P. Abolmaesumi, B at the 1st IEEE/RAS-EMBS Int. Conf. M. Djaid, T. Fortin, M. Chenic, and S. Bachmann, and P. D. Lawrence, A robot [ Biomedical Robotics and Biomechatronics A. Chapel, BIGOR: Image guided operating system for medical ultrasound, in Proc. 9th (BioRob 2006), Pisa, Italy. robot,[.Innovation et Technonogie en Biologie Int. Symp. Robotics Research (ISRR), 1999, et Medicine, vol. 13, pp. 374–394, 1992. IEEEpp. 195–202. [47] S. Micera, P. N. Sergi, F. Zaccone, B G. Cappiello, M. C. Carrozza, [21] R. H. Taylor, H. A. Paul, P. Kazandzides, [34] K. Ikuta and M. Nokata, Minimum wire [ E. Guglielmelli, R. Colombo, F. Pisano, B. D. Mittelstadt, W. Hanson, J. F. Zuhars, drive of multi micro actuators, .J. Robot. Soc. G. Minuco, and P. Dario, BA low-cost B. Williamson, B. L. Musits, E. Glassman, Jpn., vol. 16, pp. 791–797, 1998. biomechatronic system for the restoration and W. L. Bargar, BAn image-directed robotic [35] G. S. Guthart and J. K. Salisbury, BThe and assessment of upper limb motor function system for precise orthopaedic surgery,[ intuitive telesurgery system: Overview and in hemiparetic subjects,[ presented at the IEEE Trans. Robot. Automat., vol. 10, no. 3, application,[ in Proc. IEEE Int. Conf. Robotics 1st IEEE/RAS-EMBS Int. Conf. Biomedical pp. 261–275, Jun. 1994. and Automation (ICRA 2000), pp. 355–360. Robotics and Biomechatronics [22] J. Adler, A. Schweikard, R. Tombropoulos, [36] N. Simaan, R. Taylor, and P. Flint, (BioRob 2006), Pisa, Italy. and J.-C. Latombe, BImage-guided robotic BA dexterous system for laryngeal [ V Proof[48] S. Kousidou, N. Tsagarakis, C. Smith, and radiosurgery, in Proc. 1st Int. Symp. Medical surgery Multi-backbone bending D. G. Caldwell, BAssistive exoskeleton for Robotics and Computer Assisted Surgery, 1994, snake-like slaves for teleoperated dexterous task based physiotherapy in 3-dimensional vol. 2, pp. 291–297. surgical tool manipulation,[ in Proc. IEEE space,[ presented at the 1st IEEE/RAS-EMBS Int. Conf. Robotics and Automation, 2004, [23] J. Petermann, R. Kober, P. Heinze, Int. Conf. Biomedical Robotics and pp. 351–357. P. F. Heeckt, and L. Gotzen, Biomechatronics (BioRob 2006), Pisa, Italy. BImplementation of the CASPAR system in [37] VV, BHigh dexterity snake-like robotic [49] T. Kanade, Conference Report: International the reconstruction of the ACL,[ in Proc. slaves for minimally invasive telesurgery AQ5 Advanced Robotics Program Workshop on North Amer. Program Computer Assisted of the throat,[ in Proc. Int. Symp. Medical Medical Robotics, Hidden Valley, PA, 2004. Orthopaedic Surgery (CAOS/USA ’99), Image Computing and Computer-Assisted pp. 86–87. Interventions, 2004, pp. 17–24. [50] R. H. Taylor, J. Funda, B. Eldridge, K. Gruben, D. LaRose, S. Gomory, [24] J. M. Sackier and Y. Wang, BRobotically [38] N. Simaan, BSnake-like units using flexible M. Talamini, L. Kavoussi, and J. Anderson, assisted laparoscopic surgery. From concept backbones and actuation redundancy for

Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 11 Taylor: A Perspective on Medical Robotics

BA telerobotic assistant for laparoscopic [64] M. Li and R. H. Taylor, BSpatial motion Computer Integrated Surgery: Technology and surgery,[.IEEE Eng. Med. Biol. Mag. constraints in medical robots using virtual Clinical Applications. Cambridge, MA: MIT (Special Issue on Robotics in Surgery), fixtures generated by anatomy,[ in Proc. Press, 1996, pp. 343–351. vol. 14, no. 3, pp. 279–288, May/Jun. 1995. IEEE Conf. Robotics and Automation, 2004, [79] K. Masamune, G. Fichtinger, A. Patriciu, [51] T. M. Goradia, R. H. Taylor, and L. M. Auer, pp. 1270–1275. R. Susil, R. Taylor, L. Kavoussi, J. Anderson, BRobot-assisted minimally invasive [65] H. Mayer, I. Nagy, and A. Knoll, BSkill I. Sakuma, T. Dohi, and D. Stoianovici, neurosurgical procedures: First experimental transfer and learning by demonstration in a BSystem for robotically assisted percutaneous experience,[ in Proc. 1st Joint Conf. Computer realistic scenario of laparoscopic surgery,[ procedures with computed tomography Vision, Virtual Reality and Robotics in Medicine presented at the IEEE Int. Conf. Humanoids, guidance,[.J. Comput.-Assisted Surg., vol. 6, (CVRMed II) and Medical Robotics and Munich, Germany, 2003. pp. 370–383, 2001. Computer Assisted Surgery (MRCAS III), [66] D. Kragic, P. Marayong, M. Li, [80] R. C. Susil, J. H. Anderson, and R. H. Taylor, 1997, pp. 319–322. A. M. Okamura, and G. D. Hager, BA single image registration method for [52] R. Taylor, P. Jensen, L. Whitcomb, A. Barnes, BHuman–machine collaborative systems CT guided interventions,[ in Proc. 2nd Int. R. Kumar, D. Stoianovici, P. Gupta, for microsurgical applications,[ presented Symp. Medical Image Computing and Z. Wang, E. deJuan, and L. Kavoussi, at the Int. Symp. Robotics Research, Computer-Assisted Interventions (MICCAI ’99), BA steady-hand robotic system for Sienna, Italy, 2003. pp. 798–808. [ microsurgical augmentation, .Int. J. Robot. [67] M. Li, A. Kapoor, and R. Taylor, [81] D. Stoianovici, J. A. Cadeddu, Res., vol. 18, no. 12, pp. 1201–1210, 1999. BA constrained optimization approach to R. D. Demaree, H. A. Basile, R. H. Taylor, [53] M. Jakopec, S. J. Harris, F. R. y. Baena, virtual fixtures,[ in Proc. IEEE/RSJ Int. Conf. L. L. Whitcomb, W. N. Sharpe, and L. R. P. Gomes, J. Cobb, and B. L. Davies, BThe Intelligent Robots and Systems (IROS 2005), Kavoussi, BAn efficient needle injection first clinical application of a hands-on robotic pp. 1408–1413. technique and radiological guidance method [ [ knee surgery system, .Comput. Aided Surg., [68] B. L. Davies, BA discussion of safety issues for percutaneous procedures, in Proc. 1st vol. 6, pp. 329–339, 2001. for medical robots,[ in Computer-Integrated Joint Conf. Computer Vision, Virtual Reality [54] C. V. Riviere, R. S. Rader, and N. V. Thakor, Surgery, R. H. Taylor, S. Lavallee, G. Burdea, and Robotics in Medicine (CVRMed II) and BAdaptive real-time cancelling of and R. Mosges, Eds. Cambridge, MA: MIT Medical Robotics and Computer Assisted physiological tremor for microsurgery,[ Press, 1996, pp. 287–298. Surgery (MRCAS III), 1997, pp. 295–298. presented at the 2nd Int. Symp. on Medical [69] R. H. Taylor, BSafety,[ in Computer-Integrated [82] J. T. Bishoff, D. Stoianovici, B. R. Lee, Robotics and Computer Assisted Surgery Surgery, R. H. Taylor, S. Lavallee, G. Burdea, J. Bauer, R. H. Taylor, L. L. Whitcomb, (MRCAS), Baltimore, MD, 1995. and R. Mosges, Eds. Cambridge, MA: MIT J. A. Cadeddu, D. Chan, and L. R. Kavoussi, B [55] W. T. Ang and C. N. Riviere, BNeural Press, 1996, pp. 283–286. RCM-PAKY: Clinical application of a new robotic system for precise needle network methods for error canceling in [70] B. Mittelstadt, P. Kazanzides, J. Zuhars, [ placement,[.J. Endourol., vol. 12, p. S82, human–machine manipulation, in Proc. B. Williamson, P. Cain, F. Smith, and 1998. 23rd Annu. Int. Conf. IEEE Engineering W. Bargar, BThe evolution of a surgical in Medicine and Biology Soc., 2001, robot from prototype to human clinical use,[ [83] J. Cadeddu, D. Stoianovici, R. N. Chen, pp. 3462–3465. in Computer-Integrated Surgery, R. H. Taylor, R. G. Moore, and L. R. Kavoussi, B [56] M. S. Nathan, B. L. Davies, R. D. Hibberd, S. Lavallee, G. Burdea, and R. Mosges, Eds. Stereotactic mechanical percutaneous renal [ and J. Wickham, BDevices for automated Cambridge, MA: MIT Press, 1996, access, .J. Urol., vol. 159, p. 56, 1998. resection of the prostate,[ in Proc. 1st Int. pp. 397–407. [84] K. Cleary, D. Stoianovici, A. Patriciu, Symp. Medical Robotics and Computer Assisted [71] P. Kazanzides, B. D. Mittelstadt, D. Mazilu, D. Lindisch, and V. Watson, Surgery, 1994, pp. 342–345. B. L. Musits, W. L. Bargar, J. F. Zuhars et al., Acad. Radiol., vol. 9, pp. 821–825, 2002. [57] B. L. Davies, R. D. Hibberd, A. G. Timoney, BAn integrated system for cementless hip [85] M. Loser and N. Navab, BA new robotic and J. E. A. Wickham, BA clinically applied replacement,[.IEEE Eng. Med. Biol. Mag., system for visually controlled percutaneous robot for prostatectomies,[ in Computer vol. 14, pp. 307–313, 1995. interventions under CT fluoroscopy,[ in Integrated Surgery: Technology and Clinical [72] F. Gosse´, K. Wenger, K. Knabe, and Proc. Int. Symp. Medical Image Computing Applications. Cambridge, MA: MIT Press, C. Wirth, BEfficacy of robot-assisted hip and Computer-Assisted Interventions 1996, pp. 593–601. stem implantation: A rediographic (MICCAI 2000), pp. 887–896. [58] C. B. Cutting, F. L. Bookstein, and comparison of matched-pair femurs prepared [86] VV. (2002). [Online]. Available: http:// R. H. Taylor, BApplications of simulation, manually and with the ROBODOC system www.picker.com/www/marconimed.nsf/. [ morphometrics and robotics in craniofacial using an anatomic prosthesis, in Proc. [87] J. Yanof, J. Haaga, P. Klahr, C. Bauer, [ surgery, in Computer-Integrated Surgery, 3rd Int. Conf. Medical Image Computing D. Nakamoto, A. Chatuvedi, and R. Bruce, R. H. Taylor, S. Lavallee, G. Burdea, and and Computer-Assisted Intervention BCT-integrated robot for interventional R. Mosges, Eds. Cambridge, MA: MIT (MICCAI 2000), pp. 1180–1187. procedures: Preliminary experiment and Press, 1996, pp. 641–662. [73] A. Bauer, BPrimary THR using the human–computer interfaces,[ Comput. Aided [59] J. Troccaz, M. Peshkin, and B. Davies, ROBODOC system,[ in Proc. 3rd Annu. Surg., vol. 6, pp. 352–359, 2001. B The use of localizers, robots and synergistic North Amer. Program Computer Assisted [88] K. Surry, W. Smith, G. Mills, D. Downey, [ devices in CAS, .Proc. 1st Joint Conf. Orthopaedic Surgery Conf. (CAOS/USA 1999), and A. Fenster, BA mechanical, three Computer Vision, Virtual Reality and Robotics pp. 107–108. dimensional ultrasound-guided breast biopsy in Medicine (CVRMed II) and Medical Robotics [74] U. Wiesel, A. Lahmer, M. Tenbusch, and apparatus,[ in Proc. Int. Symp. Medical Image and Computer Assisted Surgery (MRCAS III), M.IEEE Borner, BTotal knee replacement using Computing and Computer-Assisted Intervention pp. 727–736. the ROBODOC system,[ in Proc. 1st Annu. (MICCAI 2001), pp. 232–239. B [60] L. B. Rosenberg, Virtual fixtures: Perceptual Meeting CAOS Int. 2001, p. 88. [89] G. Megali, O. Tonet, C. Stefanini, [ tools for telerobotic manipulation, in [75] W. Siebert and S. Mai, BOne year clinical M. Boccadoro, V. Papaspyropoulis, Proc. IEEE Virtual Reality Int. Symp., 1993, experience using the robot system L. Angelini, and P. Dario, BA computer- pp. 76–82. CASPAR for TKR,[ in Proc. CAOS USA assisted robotic ultrasound-guided biopsy [61] S. Park, R. D. Howe, and D. F. Torchiana, 2001, pp. 141–142. system for video-assisted surgery,[ in Proc. B [ Virtual fixtures for robotic cardiac surgery, [76] A. DiGioia, B. Jaramaz, F. Picard, and Int. Symp. Medical Image Computing and in Proc. 4th Int. Conf. Medical Image L. P. Nolte, Computer and Robotic Assisted Computer-Assisted Intervention (MICCAI Computing and Computer-Assisted Knee and Hip Surgery. New York: Oxford, 2001), pp. 343–350. Intervention, 2001, pp. 1419–1420. Proof 2004. [90] G. Fichtinger, E. Burdette, A. Tanacs, AQ6 B [62] M. Li and A. M. Okamura, Recognition of [77] S. Lavallee, P. Sautot, J. Troccaz, P. Cinquin, A. Patriciu, D. Mazilu, L. Whitcomb, and B operator motions for real-time assistance and P. Merloz, BComputer assisted D. Stoianovici, Robotically-assisted prostate [ using virtual fixtures, in Proc. 11th Int. Symp. spine surgery: A technique for accurate brachytherapy with transrectal ultrasound V [ Haptic Interfaces for Virtual Environment and transpedicular screw fixation using CT data guidance preliminary experiments, Int. J. Teleoperator Systems, 2003, pp. 125–131. and a 3-D optical localizer,[ in Proc. 1st Int. Radiat. Oncol. Biol., 2002, (in review). [63] P. Marayong and A. Okamura, Symp. Medical Robotics and Computer-Assisted [91] K. Chinzei, N. Hata, F. Jolesz, and R. Kikinis, BEffect of virtual fixture compliance on Surgery (MRCAS 94), pp. 315–322. BMR compatible surgical assist robot: [ human–machine cooperative manipulation, [78] S. Lavallee, J. Trocaz, L. Gaborit, system integration and preliminary [ in Proc. IEEE/RSJ Int. Conf. Intelligent Robots P. Cinquin, A. L. Benabid, and D. Hoffmann, feasibility study, in Proc. 3rd Int. Conf. and Systems, 2002, vol. 2, pp. 1089–1095. BImage-guided operating robot: A clinical Medical Robotics, Imaging and Computer application in stereotactic neurosurgery,[ in Assisted Surgery, 2000, pp. 921–930.

12 Proceedings of the IEEE |Vol.94,No.9,September2006 Taylor: A Perspective on Medical Robotics

[92] K. Masamune, E. Kobayashi, Y. Masutani, K. Kan, M. G. Fujie, H. Iseki, and [111] G. F. K. Masamune, A. Patriciu, R. Susil, AQ7 M. Suzuki, T. Dohi, H. Iseki, and K. Takakura, BNeuRobot: Telecontrolled R. Taylor, L. Kavoussi, J. Anderson, K. Takakura, BDevelopment of an MRI- micromanipulator system for minimally I. Sakuma, T. Dohi, and D. Stoianovici, compatible needle insertion manipulator for invasive microneurosurgery-preliminary BGuidance system for robotically assisted stereotactic neurosurgery,[ J. Image Guid. results,[ Neurosurgery, vol. 51, percutaneous procedures with computed Surg., vol. 1, pp. 242–248, 1995. pp. 985–988, 2002. tomography,[ Comput. Assisted Surg., vol. 6, [93] G. Fichtinger, A. Krieger, A. Tanacs, [103] K. Ikuta, T. Hasegawa, and S. Daifu, BHyper 2001. L. Whitcomb, and E. Atalar, BTransrectal redundant miniature manipulator Fhyper [112] S. Solomon, A. Patriciu, R. H. Taylor, prostate biopsy inside closed MRI scanner finger_ for remote minimally invasive L. Kavoussi, and D. Stoianovici, BCT guided with remote actuation under real-time image surgery in deep area,[ in Proc. IEEE Conf. robotic needle biopsy: A precise sampling guidance,[ presented at the Medical Image Robotics and Automation, 2003, method minimizing radiation exposure,[ Computing and Computer-Assisted pp. 1098–1102. Radiology, vol. 225, pp. 277–282, 2002. Intervention, Tokyo, 2002, pp. 91–98. [104] K. Ikuta, K. Yamamoto, and K. Sasaki, [113] A. Krupa, J. Gangloff, C. Doignon, [94] S. P. DiMaio, G. S. Fischer, S. J. Haker, BDevelopment of remote microsurgery robot M. F. deMathelin, G. Morel, J. Leroy, N. Hata, I. Iordachita, C. M. Tempany, and new surgical procedure for deep and L. Soler, and J. Marescaux, BAutonomous R. Kikinis, and G. Fichtinger, BA system narrow space,[ in Proc. IEEE Conf. Robotics 3-D positioning of surgical instruments in for MRI-guided prostate interventions,[ and Automation, 2003, pp. 1103–1108. robotized laparoscopic surgery using visual [ in BioRob, Pisa, Feb. 2006. [105] R. H. Taylor, P. Jensen, L. L. Whitcomb, servoing, IEEE Trans. Robotics and [95] J. A. McEwen, C. R. Bussani, A. Barnes, R. Kumar, D. Stoianovici, Automation, vol. 19, pp. 842–853, 2003. G. F. Auchinleck, and M. J. Breault, P. Gupta, Z. X. Wang, E. deJuan, and [114] J. Leven, D. Burschka, R. Kumar, BDevelopment and initial clinical evaluation L. R. Kavoussi, BA steady-hand robotic G. Zhang, S. Blumenkranz, X. Dai, of pre-robotic and robotic retraction systems system for microsurgical augmentation,[ Int. M. Awad, G. Hager, M. Marohn, M. Choti, for surgery,[ in Proc. 2nd Workshop Medical J. Robot. Res., vol. 18, pp. 1201–1210, 1999. C. Hasser, and R. Taylor, BDaVinci canvas: and Health Care Robotics, 1989, pp. 91–101. [106] M. Blackwell, C. Nikou, A. M. DiGioia, and A telerobotic surgical system with integrated, [96] J. A. McEwen, BSolo surgery with automated T. Kanade, BAn image overlay system for robot-assisted, laparoscopic ultrasound [ positioning platforms,[ presented at the medical data visualization,[ Med. Image capability, presented at the Int. Conf. NSF Workshop Computer Assisted Surgery, Anal., vol. 4, pp. 67–72, 2000. Medical Image Computing and Computer-Assisted Intervention, Washington, DC, 1993. [107] T. Sasama, N. Sugano, Y. Sato, Y. Momoi, Palm Springs, CA, 2005. [97] J. Marescaux, J. Leroy, M. Gagner, F. Rubino, T. Koyama, Y. Nakajima, I. Sakuma, D. Mutter, M. Vix, S. E. Butner, and M. G. Fujie, K. Yonenobu, T. Ochi, and [115] P. Abolmaesumi, S. E. Salcudean, W. H. Zhu, M. K. Smith, BTransatlantic robot-assisted S. Tamura, BA novel laser guidance system M. R. Sirouspour, and S. P. DiMaio, B telesurgery,[ Nature, vol. 413, pp. 379–380, for alignment of linear surgical tools: Its Image-guided control of a robot for [ 2001. principles and performance evaluation as medical ultrasound, IEEE Trans. Robot. [ Automat., vol. 18, no. 1, pp. 11–23, Feb. 2002. [98] M. Ghodoussi, S. E. Butner, and Y. Wang, a man-machine system, in Proc. 5th BRobotic surgeryVThe transatlantic case,[ Int. Conf. Medical Image Computing [116] I. Nagy, H. Mayer, A. Knoll, in Proc. IEEE Int. Conf. Robotics and and Computer-Assisted Intervention, 2002, E. U. Schirmbeck, and R. Bauernschmitt, B Automation, 2002, pp. 1882–1888. pp. 125–132. EndoPAR: An open evaluation system for minimally invasive robotic surgery,[ [99] A. Mohamed and C. Davatzikos, BFinite [108] D. S. Kwon, J. J. Lee, Y. S. Yoon, S. Y. Ko, presented at the IEEE Mechatronics and element modeling of brain tumor J. Kim, J. H. Chung, C. H. Won, and B Robotics 2004 (MechRob), Aachen, mass-effect from 3-D medical images,[ J. H. Kim, The mechanism and the Germany. presented at the Int. Symp. Medical registration method of a surgical robot [ B Image Computing and Computer-Assisted for hip arthroplasty, in Proc. IEEE Int. [117] H. Mayer, I. Nagy, and A. Knoll, Kinematics Interventions (MICCAI 2005), Palm Conf. Robotics and Automation, 2002, and modelling of a system for robotic [ Springs, CA. pp. 1889–2949. surgery, in Proc. 9th Int. Symp. Advances in Robot Kinematics, 2004, pp. 181–190. [100] A. Mohamed, D. Shen, and C. Davatzikos, [109] M. Shoham, M. Burman, E. Zehavi, BDeformable registration of brain L. Joskowicz, E. Batkilin, and Y. Kuchiner, [118] I. Nagy, H. Mayer, A. Knoll, E. Schirmbeck, B B tumor images via a statistical model of Bone-mounted miniature robot for surgical and R. Bauernschmitt, The EndoPar system [ [ tumor-induced deformation,[ presented at spinal procedures, in Proc. 2nd Annu. for minimally invasive robotic surgery, the Int. Symp. Medical Image Computing Meeting Int. Soc. Computer Assisted in Proc. IEEE/RSJ Int. Conf. Intelligent Robots and Computer-Assisted Interventions Orthopaedic Surgery (CAOS 2002), and Systems (IROS), 2004, pp. 3637–3642. (MICCAI 2005), Palm Springs, CA. 2002, pp. 59. [119] K. Cleary, D. Stoianovici, A. Patriciu, [101] S. G. A. Menciassi, A. Moglia, G. Pernorio, [110] G. Fichtinger, A. Degeut, G. S. Fischer, D. Mazilu, D. Lindisch, and V. Watson, B C. Stefanini, and P. Da, BClamping tools of a E. Balogh, H. Matthieu, R. H. Taylor, Robotically assisted nerve and facet blocks: B [ capsule for monitoring the gastrointestinal S. J. Zinreich, and L. M. Fayad, Image A cadaveric study, Acad. Radiol., vol. 9, tract,[ in Proc. IEEE Conf. Robotics and overlay guidance for needle insertions in CT no. 7, pp. 821–825, Jul. 2002. [ Automation, 2005, pp. 1321–1326. scanner, IEEE Trans. Biomed. Eng., vol. 52, no. 8, pp. 1415–1424, Aug. 2005. [102] K. Hongo, S. Kobayashi, Y. Kakizawa, J.-I. Koyama, T. Goto, H. Okudera, IEEE ABOUT THE AUTHOR 741 Russell H. Taylor (Fellow, IEEE) received the history of research in computer-integrated surgery and related fields. In 757 742 B.E.S. degree from Johns Hopkins University, 1988–1989, he led the team that developed the first prototype for the 758 743 Baltimore, MD, in 1970 and the Ph.D. degree in ROBODOC system for robotic hip replacement surgery and is currently on 759 744 computer science from Stanford University, Stan- the Scientific Advisory Board of Integrated Surgical Systems. At IBM he 760 745 ford, CA, in 1976. subsequently developed novel systems for computer-assisted craniofa- 761 746 He joined IBM Research in 1976, where he cial surgeryProof and robotically augmented endoscopic surgery. At Johns 762 747 developed the AML robot language. Following a Hopkins, he has worked on all aspects of CIS systems, including 763 748 two-year assignment in Boca Raton, FL, he man- modeling, registration, and robotics in areas including percutaneous 764 749 aged robotics and automation technology re- local therapy, microsurgery, and computer-assisted bone cancer surgery. 765 750 search activities at IBM Research from 1982 until Dr.TaylorisEditorEmeritusoftheIEEETRANSACTIONS ON ROBOTICS AND 766 751 returning to full-time technical work in late 1988. From March 1990 to AUTOMATION, a Fellow of the American Institute for Medical and Biological 767 752 September 1995, he was manager of Computer Assisted Surgery.In Engineering (AIMBE), and a member of various honorary societies, 768 753 September 1995, he moved to Johns Hopkins University as a Professor of panels, editorial boards, and program committees. Dr. Taylor is a 769 754 Computer Science, with joint appointments in Radiology and Mechanical member of the scientific advisory board for Integrated Surgical Systems. 770 755 Engineering. He is also Director of the NSF Engineering Research Center In February 2000, he received the Maurice Mu¨ller award for excellence in 771 756 for Computer-Integrated Surgical Systems and Technology. He has a long computer-assisted orthopedic surgery. 772

Vol. 94, No. 9, September 2006 | Proceedings of the IEEE 13 AUTHOR QUERIES

AUTHOR PLEASE ANSWER ALL QUERIES

AQ1 = Edited as per IEEE style–correct?

AQ2 = Please provide copy of written permission to use any figures copyrighted to another publisher’

AQ3 = Paper title? in Ref. [7].

AQ4 = Paper title? in Ref. [9] AQ5 = Please provide paper title and page span in Ref. [49].

AQ6 = Please update Ref. [90].

AQ7 = Is this the same as Ref. [79]? Pease provide complete information.

Notes: 1) Acknowledgment was captured in affnote field. 2) Reference [84] was split into two references and added Reference [119].

3) Unsequenced figures. Figure 6 was placed after Figure 5.

4) All figures and table are colored.

5) Figures 3 and 10 were not cited. Captions were placed after Figures 2 and 9 respectively. 6) Author’s photo is unsharp.

END OF AUTHOR QUERIES IEEE Proof