gms | German Medical Science

GMS Current Topics in Computer and Robot Assisted Surgery

Deutsche Gesellschaft für Computer- und Roboterassistierte Chirurgie (CURAC)

ISSN 1863-3153

Evaluation of dynamic performance of the Interactive Milling System MicroAssistant

Research Article

Search Medline for

  • corresponding author Carsten Lenze - Division of Automation and Measurement Technology (AMT), Department of Computer Science, University of Oldenburg, Oldenburg, Germany
  • author Ralf Eckert - Division of Automation and Measurement Technology (AMT), Department of Computer Science, University of Oldenburg, Oldenburg, Germany
  • author Andreas Hein - Division of Automation and Measurement Technology (AMT), Department of Computer Science, University of Oldenburg, Oldenburg, Germany

GMS CURAC 2006;1:Doc13

The electronic version of this article is the complete one and can be found online at: http://www.egms.de/en/journals/curac/2006-1/curac000013.shtml

Published: October 9, 2006

© 2006 Lenze et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.en). You are free: to Share – to copy, distribute and transmit the work, provided the original author and source are credited.


Abstract

This paper is about evaluation steps of the kinematic system MicroAssistant. The MicroAssistant is an interactive surgical robotic system which is designed for microsurgical interventions in particular middle ear surgeries. For this field of application an accuracy of approximately 0.5 mm is required. The MicroAssistant is in development and not yet clinical evaluated. Therefore the technical evaluation is addressed mainly in this context and clinical exactness can not be derived through this evaluation. The interactive system MicroAssistant was evaluated in this paper in two stages. At first the experiment was designed to evaluate following influences: instability of the kinematic, errors of the axes and DC-motor control, errors from sensors and errors due to the kinematic model of the robotic system. In the second experiment interaction with milling was evaluated. Following influences are evaluated additional: counterforces during milling, flexibility of the manipulator, mounting inexactnesses. For experiment 1 the mean error is about 0.1 mm which is acceptable. For experiment 2 the errors are about 0.4 mm which has to be optimized further.

Keywords: computer assisted surgery, micro surgery, robot-assisted surgery, accuracy of milling system, evaluation of kinematic structure, interactive milling system


Introduction

Medical Background

The MicroAssistant is in preclinical evaluation but addresses clinical applications in the field of micro surgery. The motivation is derived from middle ear surgery were the surgeon is often confronted with very small and sensitive structures like tiny bones (hammer, anvil and stirrup) or nerves and vessels. For evaluation the first focus is the milling process which is necessary for widening the operation field as well as for cleaning otosclerotic regions out of the middle ear. In this context there are some general problems:

1) The size of these structures which should be manipulated are possible hard at the edge of the human manipulating skills of approximately 0.5 mm. The given value of 0.5 mm is not fixed through publications yet and therefore approximated in this context.

2) While getting access to the operation field in stapedectomy interventions the surgeon has to mill with extensive force, whereas shortly after he has to do precise and tremble free positioning of the surgical instrument [1].

3) In many cases the surgeon decides to use a microscope for the surgery. Thereby the coordination of the optical sensed scene and the handling of the haptic sensed surgical instrument is very difficult to combine to a precise work.

These problems could lead into complications (injury of sensitive structures like nerves or vessels) or cost-intensive following surgeries.

State of the art

The MicroAssistant has to be categorized under the main topic of computer aided systems (CAS). These CAS can be further divided into navigation systems and robotic systems. Navigation systems are displaying information about the anatomical situation in combination with the actual position of an used instrument. A commercial navigation system for the usage in ear, nose and throat surgery (ENT) is for example Vector Vision ENT by BrainLab. The disadvantage of navigation systems is that tremor can not be reduced and a slipping off can not be prevented, too. Robotic systems can be differentiated by their control principles [2]: 1) automatic systems, 2) telemanipulative systems and 3) interactive systems.

1) Automatic systems are systems which control either position, velocity or force. Due to this, automatic systems transform planning data on preoperative images to movements of a surgical instrument along planned trajectories. The specific surgical intervention step is fulfilled automatically but always with the possibility of interrupting it. Examples in this field are the Robodoc (ISS, Sacramento, USA) and the Caspar (Orto Maquet, Germany) system. A system concerning milling operations is the medical robotic system RONAF [3]. Disadvantages of these automatic systems are the limited capability of intervention, especially in unpredictable intraoperative incidents. Another disadvantage of these automatic systems is that they are basically built up out of industrial robots which are adapted for medical applications. These systems are quite large in size which results in a restricted work area for the surgeon.

2) Telemanipulative systems do not provide position control. Preoperative planning is also not possible in such kind of systems. Due to the missing planning step the surgeon has to control the position of the robotic system by himself. For this a direct or indirect sight on the field of intervention is necessary. Telemanipulative systems are for example the daVinci (Intuitive Surgical, Sunnyvale, USA) and ZEUS (formerly Computer Motion Inc. now Intuitive Surgical, Sunnyvale, USA) system. A system that has already been developed for middle ear interventions is the Steady Hand Surgical Assistant System [4] which is also a telemanipulative system.

3) Interactive systems combine the position control as well as the preoperative planning. Uncontrolled values are given interactively through the user. This user interaction awaits continuous inputs. Are these control signals missing such systems interpret the missing signals as a stop signal. With this mechanism an uncontrolled movement can be prevented. Interactive systems are for example the Acrobot [5], neuroArm [6] and the first clinical applied interactive robotic system SurgiScope 2000 [7]. For solving the medical requirements of the focused microsurgery inventions such interactive systems are most capable.

As a result from the state of the art an interactive system (MicroAssistant) [8] is buildup to interactively support the surgeon during microsurgical interventions (Figure 1 [Fig. 1]). The concept of the system is to plan the intervention on preoperative images and to manually position the MicroAssistant above the operation field. For guidance of the positioning an optical position measuring system MicronTracker2 (Clarontech, Toronto, Canada) is embedded. The control of the system is fulfilled through a kind of joystick directly placed at the MicroAssistant.

Kinematic structure

For achieving higher stiffness a parallel kinematic is designed. The kinematic structure is new and detailed described as follows: The kinematic of the MicroAssistant consists of two linear axes driven by electrical DC-motors. The system only allows a movement along a plane (x-y axes) which supports the lateral milling [9]. The plane should ideally positioned vertical to the insertion direction of the instruments. The insertion direction (depth) has to be adjusted by hand. The system only consists of two degrees of freedom to achieve an easy and compact system. The x-y plane is chosen due to an optimal support for force reduction of the surgeon because sideward forces are more exhausting to apply. The control of preparation in the z-axis remains completely by the surgeon. This prevents the system to be of an automatic (surgeon independent) character while controlling the milling process in two directions in order to keep the focus for the surgeon on the delicate one.

To evaluate the positioning accuracy of the system two experiments are done (Experiment 1 and Experiment 2). At first experiment the kinematic is evaluated considering following influences:

  • Errors of guides
  • Errors due to sensor resolution and DC-motor control
  • Errors due to the kinematic model of the robotic system

In the second experiment following additional influences are regarded:

  • Instability of kinematic structure due to counterforces during milling
  • Flexibility of the instrument
  • Mounting inexactnesses

The three influences in experiment 1 are discussed in more detail as follows: The instability of the kinematic comes up through the limited stiffness of the construction. This error source should be negligible. The MicroAssistant consist of two guides KR15 (THK, Tokyo, Japan) attached with two brushless DC-Servomotors Series 2036 (Faulhaber, Schönaich, Germany) and a guide RSR (THK, Tokyo, Japan) which connects the motorized KR15 guides (Figure 2 [Fig. 2]).

The guides KR15 have a positioning accuracy of 0.04 mm (data sheet, THK) and the tolerance of the guide RSR is given with 0.04 mm as well. Sensors used to control the positioning are the integrated hall sensors of the DC-motors. The resolution for these hall sensors is 1024 lines per turn. The guide, which the sensors are connected with, has a lead of 1 mm per turn which results in a sensor accuracy of 0.001 mm. The MicroAssistant is buildup as a parallel kinematic which kinematic model is given as follows:

The motorized guides of the kinematic are taken as the x and y axes of the kinematic coordinate system (see Figure 3 [Fig. 3]). The motor positions are marked as t x and t y . To get the total position of the platforms along the axis the offset hx and hy have to be added. These positions can be readout from the hall sensors. Constructively fixed design parameters are given in Figure 3 [Fig. 3] as k 1 and k 2 . To get the position of the tool center point within the kinematic coordinate system ( kin p tcp ) only by knowing the motor positions, the forward calculation can be calculated as follows:

Equation 1 (1)
with Equation 2 (2)

For controlling the kinematic movement much more important is the backward calculation. A target position within the kinematic coordinate system is given and the motor positions have to be calculated. Equation (3) and (4) show the according calculations:

Equation 3 (3)
Equation 4 (4)
with Equation 5 (5)

The errors which are expected resulting from the shown kinematic model should not arise from mathematical inexactnesses but they result from the constructive design parameters k 1 , k 2 , and h 1 , h 2 . These values vary because of manufacturing tolerances.

Influences coming up in experiment 2 are not as easy definable as the influences already discussed for experiment 1. Therefore the comparison of experiment 1 and experiment 2 is important to identify the effect of these influences on the complete system.


Methods

The experimental setup for experiment 1 is as follows: The MicroAssistant is situated on a table and not on the mounting arms (Figure 4 [Fig. 4]). The influences of mounting inexactnesses can be eliminated by this setup. Furthermore a pencil is used as an instrument and replaces the milling instrument, which is used later. For getting information of the positions the pencil as the tool center point follows the driven path of the kinematic and records the path on a sheet of paper.

As geometry a spiral of different diameters (10, 20 and 30 mm) were chosen. Beginning from the middle point of the spiral the kinematic moves to the outer position of the spiral diameter. The spiral diameter is reduced with 1, 2 or 3 mm difference to the middle point per turn respectively to the chosen diameter of the spiral. This spiral path is automatically controlled by our control system. For starting position of the spiral within the work area of the MicroAssistant the middle of both axes is chosen.

The path can be analyzed through following steps (Figure 4 [Fig. 4]): 1) The recorded path is scanned with a resolution of 300 dpi. With this resolution the dimensions of a pixel is 0.08 mm. 2) Within a special analyze software 36 measurement rays, starting from the middle of the spiral, are drawn circular with 10 degree differences. 3) The crossings of the measurement rays with the spiral path of the kinematic are marked. 4) The marked points are statistically evaluated for different values. The positioning accuracy can be calculated by the differences of the marked points along one measurement ray. In addition the thickness of the drawn spiral path can be concluded as the accuracy of the dynamic motion of the MicroAssistant. It can be assumed that a non constant movement of the kinematic will result in an uneven thickness of the spiral path. For this evaluation not only the crossing of the measurement rays with the spiral path has to be marked, but also the line thickness at this marked point.

In experiment 2 the MicroAssistant is evaluated with interaction of milling [10]. The system buildup is close to the expected usage of the system in surgical application (see Figure 5 [Fig. 5]). The system is mounted on two mounting arms above a plastic skull. In the skull a hole is inserted for an easy exchange of milling models. These milling models are produced out of plaster which has similar qualities as bone. To show the accuracy of the system a circular geometry is milled similar to the diameter of an ear channel of about 10 mm. To measure the radii the starting position was chosen as the middle point of the circle. In the image analysis the milled areas can be marked and pixels can be counted to get together with the resolution of the image a value in millimeters. To demonstrate the general milling capability two different diameters of the milling instrument tips were used to mill the same circle. For milling a standard surgical milling instrument were used (Aesculap, Tuttlingen, Germany).


Results

In experiment 1 without milling interaction the results have to be divided into positioning accuracy and dynamic motion accuracy.

For the positioning accuracy the values are related to the spiral path distance along the measurement rays (see Figure 6 [Fig. 6]). For the 10 mm spiral the results are given through a mean value of 0.96 mm with a standard deviation of 0.04 mm. The maximum error is 0.84 mm. The expected value for the mean value is 1 mm which results in a mean error of -0.04 mm. For the 20 mm spiral the mean value is 2.13 mm with a standard deviation of 0.05 mm. The maximum error is 0.53 mm. The expected value of the mean value is 2.0 mm which results in a mean error of 0.13 mm. The result for the 30 mm spiral is a mean value of 2.96 mm with a standard deviation of 0.06 mm. The maximum error is 0.38 mm. At the 30 mm diameter spiral a value of 3 mm is expected for the mean value which results in a mean error of -0.04 mm.

The values concerning the dynamic motion accuracy are given for the 10 mm spiral with a mean value of 0.52 mm and a standard deviation of 0.05 mm. The result for the 20 mm spiral is for the mean value 0.49 mm with a standard deviation of 0.05 mm. The 30 mm spiral has a mean value of 0.53 mm with a standard deviation of 0.06 mm. For this experiment values no expected reference value can be derived because the line thickness was dependent of the pencil. In the context of the dynamic motion accuracy only the standard deviation is of interest.

Experiment 2 with interaction of milling [10] was divided into two further experiments regarding the milling instrument tip diameter. The first experiment with a tip diameter of 2.7 mm resulted (Figure 7 [Fig. 7]) in a mean error of 0.4 with a standard deviation of 0.2 mm. The maximum error is 1.3 mm. In the second experiment with a tip diameter of 1.7 mm the mean error was 0.4 mm with a standard deviation of 0.4 mm and a maximum error of 1.6 mm.


Discussion

The mean error in experiment 1 for the position accuracy of the 10 and 30 mm spiral with 0.04 is very low. Even the mean error of 0.13 for the 20 mm diameter spiral is half than the mean error through interaction of milling in experiment 2. This shows that the influences of counterforces during milling, flexibility of the manipulator and mounting inexactnesses is significantly high. Furthermore it should be pointed out that for the mounting and restraint devices the necessity of improvement is given. The high maximum error values appear in only few measurements. For these measurement points it was difficult to find the exact line because the circle was near the middle point. For the dynamic motion accuracy the values for the standard deviation are constantly very low which leads to the conclusion, that the kinematic moves smooth along the given path with a regular continually movement.

At experiment 2 the errors are much higher than the errors form experiment 1. This shows at the one hand that the effect of the additional influences are very high and on the other hand the measurement method is be seen as a critical error influence. After analyzing the results the measurement analysis is been pointed out as the major error source. Due to the method chosen for placing the middle point of the millings a shift of the middle point is done which results in higher mean and especially maximum errors. For the milled circle with the maximum error of 1.6 mm a visual placed circle was placed in the image which results in a reduction of the maximum and mean error as follows (only one circle): the mean error is 0.2 mm with a standard deviation of 0.15 mm and the maximum error is 0.45 mm. Some additional influences are the fixation of the milling tip in the milling instrument which is about 0.2 mm and should be the instability of the kinematic structure due to counterforces during milling as well as the mounting inexactnesses (kinematic, plaster model, milling instrument).


Conclusions

The evaluation for experiment 1 is satisfying at this point of the development. Without redesigning the complete kinematic system these errors are acceptable. The twice as high errors in experiment 2 resulting from milling interaction should be further optimized. For further evaluation and development steps the MicroAssistant should be analyzed with 1) user interaction and also in combination with 2) the positioning control for achieving an interactive robotic system. The user interaction will be evaluated in combination of our tactile human machine interface. And for the positioning control planning data hast to be included. The planning step will be carried out on preoperative CT data sets. With these further improvements also different geometries will be regarded for analyzes. Through the combination with user interaction additional errors arise which affect the accuracy of the MicroAssistant. These influences should be:

  • Counterforces applied from the user
  • Inexactnesses due to the cognition of the user
  • Problem of latency time (capacity of reaction) coming up through the ability of the user to apply control sequences

For clinical applications neither spirals nor milling circles are relevant but from theses geometries technical exactnesses of the system could be derived. In clinical application as an interactive robotic system the MicroAssistant will use preoperative images for planning intervention. With the expected resolution of CT or MRI images it could be stated that the MicroAssistant will not apply further inexactness to the intervention.


References

1.
Helms J, Jahrsdoerfer RA. Kopf- und Hals-Chirurgie in 3 Bänden, Band 2: Ohr. Stuttgart: Thieme; 1998.
2.
Lueth, TC, Hein A, Albrecht J, Demirtas M, Zachow S, Heissler E, Klein M, Menneking H, Hommel G, Bier J. A Surgical Robot System for Maxillofacial Surgery. In: IEEE Int. Conf. on Industrial Electronics, Control, and Instrumentation (IECON), Aachen, Germany, Aug. 31-Sep. 4, 1998. p. 2470-5.
3.
Stolka P, Henrich D. Building Local Maps in Surgical Robotics. In: IEEE Int. Conf. on Intelligent Robots and Systems (IROS). Edmonton, Alberta/Canada, Aug. 2 - 6, 2005.
4.
Berkelman PJ, Rothbaum L, Roy J, Lang S, Whitcomb LL, Hager G, Jensen PS, de Juan E, Taylor RH, Niparko JK. Performance Evaluation of a Cooperative Manipulation Microsurgical Assistant Robot Applied to Stapedotomy. In: Lecture Notes in Computer Science: Medical Image Computing and Computer-Assisted Interventions MICCAI, The Fourth International Conference, Utrecht, Netherlands, 2001. p. 1426-8.
5.
Ho SC, Hibberd RD, Davies BL. Robot Assisted Knee Surgery. IEEE Eng Med Biol. 1995;14(3).
6.
McBeth PB, Deon FL, Rizun PR, Sutherland GR. Robotics in neurosurgery. Am J Surg. 2004;188(4A Suppl):68S-75S.
7.
Hein A, Lueth TC. Image-based control of interactive robotics systems. In: Second International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI'99), Cambridge, England. September 19-22, 1999. p. 1125-32.
8.
Lenze C, Chaudhri N, Volling P, Hein A. Concept for a navigated micro surgical assistant system for middle ear surgery. In: CURAC, München, Germany, Oct. 8-9, 2004.
9.
Hein A. Mikrorobotik - Innovative Ansätze in der Medizin. In: PTK 2004 - Internationales Produktionstechnisches Kolloquium, Berlin, Germany, Sept. 27-29, 2004. p. 431-8.
10.
Lenze C, Hein A. Preliminary Evaluation of an Interactive Milling System. In: Proc. of CARS 2005. International Congress Series, ICS5219. Berlin, Germany: Elsevier; 2005. Vol. 1281C. p. 559-64.