//Logo Image
Author: Yeh-Liang Hsu, Shih-Tseng Lee, Hao-Wei Lin (2001-06-11); recommended: Yeh-Liang Hsu (2001-10-28).
Note: This paper is published in Biomedical Engineering, Applications, Basis and Communications, Vol. 13, No. 4, August, 2001.

A modular mechatronic system for automatic bone drilling

Abstract

Drilling tools currently used in surgery depends only on the surgeon’s manual skills to stop the penetration when completing a hole. This paper presents a modular mechatronic system for automatic bone drilling in surgery. The development of a “modular system” that is compatible with motor-driven drills that are commercially available, rather than developing a new surgical drill, is emphasized. A fuzzy controller analyzes the electric current consumed by the DC motor of the drill. When break-through is detected, the power will be cut and stops drilling in order to prevent excessive protrusion of the drill bit. In extensive drilling tests on real human skulls, there were no unexpected failure, and the overshoots of all tests were well less than 2mm.

Keywords: bone drilling, fuzzy control, mechatronics.

1. Introduction

In surgery, hand-held motor driven tools are used to manually perform bone-machining procedures such as drilling, reaming, and sawing. Currently, bone-drilling tools used in surgery do not include any means for the control of penetration. To stop the penetration of the drill when a hole is completed depends only on the surgeon’s manual skills. Allotta et al. pointed out that the performance of existing motor-driven drilling tools is limited by the lack of any sensing means suitable for recognizing the crossing of interfaces between hard and soft tissues and, more generally, to discriminate among layers of different tissues [1].

For this purpose, Allotta et al. developed a hand-held drilling tool for orthopedic surgery [1]. The main tool feature is the capability of early detection of interfaces between layers of different bone tissues and automatic feed stop according to the specification of the surgeon. A force sensor is embodied in the drilling tool to detect the sharp drop in thrust when the drill crosses the interfaces between hard and soft tissues of the bone. After comparing several methods, a fuzzy logic controller is used for controlling penetration [2].

Baker et al. also presented a mechatronic drilling tool for precise drilling of flexible bone tissues during ear surgery [3, 4]. By characterizing the tissue from real time drilling data, it is able to control the drilling to complete the break-through with minimum drill bit protrusion. A feed carriage houses the motor, which produces the drilling rotation, and all sensing components that measure axial force and torque on the drill and its displacement. A support is used to help steady the hand held tool. Kaburlasos et al. further developed a two-level fuzzy-lattice learning scheme for on-line estimation of the thickness of a stapes bone using a force/torque pair of drilling profiles [5, 6].

Glauser et al. described a robot dedicated to stereotactic neurosurgery, which consists of the introduction of a small probe with a diameter of 2~3 mm through a hole drilled in the skull, in order to reach a point inside the brain [7]. This point has previously been located on scanner sections and marked by means of a reference system on the patient’s head. A motor-driven drilling tool is used to perforate the bone. In this system, the electric current consumed by drill motor is analyzed. Comparing the electric current to several thresholds, the beginning of drilling, crossing of different layers, and final break-through can be spotted. In this device, the drill must not penetrate beyond 2 mm inside the skull to prevent injuring the dura.

Bouazza-Marouf et al. presented a purpose built manipulator for invasive orthopedic surgery. This manipulator allows a drill-bit guide to be automatically aligned with the planned drilling trajectory. The surgeon can then perform manual completion of the drilling stage. A strain gauge, which monitors the axial drilling force, is incorporated into the design of the drill-feed carriage to provide force feed back for safety enhancement. Following this work, Ong and Bouazza-Marouf described a reliable and repeatable method of break-through detection based on a modified Kalman filter when drilling into long bones [9, 10]. The effects of system compliance and inherent drilling force fluctuation on the profiles of drilling force, drilling force difference between successive samples and drill bit rotational speed, are also taken into consideration.

This paper presents the development of a modular mechatronic system for automatic bone drilling in surgery. The development of a “modular system”, rather than a new drilling tool, is emphasized. One of the major objectives of this research is to develop “add-on” devices that are compatible with current DC motor-driven drills that are commercially available.

There are three major modules: the control unit, the feed carriage, and the supporting arm. The control unit consists of a control box and a PC. The control box supplies power to the drill, and in the mean time, a fuzzy controller analyzes the electric current consumed by the DC motor of the drill in order to prevent excessive protrusion of the drill bit. Detection of drill bit break-through can be achieved with most DC motor-driven drills by simply plugging the drill into the control box. When break-through is detected, the power will be cut and stops drilling.

The surgeon can also clamp the drill to the feed carriage for automatic feed. The fuzzy controller also controls the feed rate of the drill according to the drilling situation. Parameters of current cutting condition such as feed rate and penetration depth of the drill bit are provided to the surgeon on the PC screen in real time. Finally, the feed carriage can be attached to a supporting arm. Integrating with an optical positioning device, completely automatic bone drilling can be achieved.

This system has undergone extensive drilling test on real human skulls under various cutting conditions, using both industrial drills and surgical drills. There were no unexpected failure, and the overshoots of all drilling tests were well less than 2mm.

2. The modular architecture of the system

As mentioned earlier, one major objective of this research that was set up in the early development stage is to develop “add-on” devices that are compatible with motor-driven drills that are commercially available, rather than developing a new surgical drill. Therefore, it is important that the system we developed must have a modular architecture.

Figure 1 shows the three modules of the system: the control unit, the feed carriage, and the supporting arm. The control unit consists of a control box and a personal computer. Under the modular design consideration, electric current consumed by the DC motor of the drill while drilling the bone is used as the sensing signal, instead of building force sensors into the drill to measure axial force. A hand held motor-driven drill can be plugged into the control box, and the surgeon can perform drilling task as usual. The control box supplies power to the drill, and in the mean time, the electric current consumed by the DC motor of the drill is analyzed. This electric current has a direct relation with the cutting torque on the drilled.

Figure 1. The three modules of the system

The control box converts this electric current into voltage signals. Figure 2 shows a typical plot of time vs. voltage when drilling a piece of human skull. Since a human bone consists of an outer shell of cortical bone around a central mass of cancellous bone, there are two distinct peaks in Figure 2. This pattern is very similar to the figures of time vs. axial force obtained by Bouazza-Marouf et al. [8], and Allotta et al. [1] when drilling bones. Fuzzy control schemes similar to those used by the authors above are implemented in the personal computer to discriminate among layers of different tissues using the variation of the electric current. When break-through (represented by the second peak and a sharp drop) is detected, the power to the drill will be cut and stops drilling.

Using this add-on control unit alone augments the manual skills of surgeons without changing current surgical practice, which should be helpful in gaining initial clinical acceptance.

Figure 2. Time vs. voltage when drilling human skulls

The surgeon can also choose to clamp the drill on the sliding block of the feed carriage to feed the drill automatically. As shown in Figure 3, the form of this feed carriage is designed to be a hand tool for the surgeon to hold with both hands to perform drilling operation. The weight of our prototype feed carriage is less than 1 kg. The surgeon can hold the handle on the left side, and push the strut in front against the skull to steadily support the drill.

Figure 3. A drill clamped to a feed carriage

A step motor (400 pulse/rev) drives the sliding block (together with the drill) forward alone a power screw. The fuzzy controller in the control unit also controls the feed rate. The normal feed rate is set at 0.5 mm/s. When the second layer of cortical bone is detected, the feed rate is reduced. When break-through is detected, the step motor either stops feeding or auto reverses, according to the specification of the surgeon. As shown in Figure 3, there is a rotation handle bar at the back of the step motor, which may be used for manual feed if necessary.

In our prototype, the maximum feed depth is 60 mm. A potentiometer is built in the feed carriage to provide the measurement of penetration depth of the drill bit in real time. The surgeon can also estimate the thickness of the bone from the three-dimensional image obtained by Computerized Tomography (CT), and specify a maximum penetration depth for safety enhancement. When the maximum penetration depth is reached, the step motor stops feeding or auto reverses, though break-through of the bone has not been detected.

Finally, the feed carriage can be attached to a supporting arm. We did not choose to develop a robotic manipulator with actuators on all joints to support the feed carriage because most industrial robots are too heavy and not suitable for medical use, and the cost for developing such a manipulator will be very high. On the other hand, we realized that drilling operation needs only one translational degree of freedom, which is already provided by the feed carriage. The real need for the supporting arm is that its end effecter (the feed carriage in this case) is able to reach a given point in a given angle deftly and conveniently, but not necessarily automatically. After reaching this point, the supporting arm should have enough stiffness to hold the feed carriage securely and steadily during drilling operation. Positioning precision may not be of ultra most concern.

Under these considerations, we choose a universal arm with magnetic base as our supporting arm. This arm is usually used to hold gauges or indicators in experiments, and is commercially available. It has 3 joints, providing 5 degrees of freedom. The surgeon can manually move the feed carriage to a given point at a given angle, and tighten the joints by simply turning a knob. These joints are then held solid by hydraulic force. The arm has an electric magnetic base, which intends to eliminate vibration and movement. The one used in our prototype has holding force of 140kg, and its arm length is 340mm.

3. The control of penetration

The key issue to the success of this automatic bone drilling system is the ability to discriminate among layers of different tissues, and to control the penetration in order to complete the hole with minimal protrusion. As discussed earlier, many previous research use force sensors embedded in the drill to monitor the axial drilling force. One important goal of our research is to develop “add-on” devices that are compatible to DC motor drills that are commercially available. Therefore we monitor the electric current consumed by the drill instead.

Figure 4 shows a plot of drilling signals vs. time when drilling the human skull. The current consumed by the drill, which has a direct relation with the cutting torque on the drilled, is converted to voltage signals. The sampling rate is chosen to be 10Hz, considering the thickness of the human skull and the feed rate of the drill. There are two peaks as shown in Figure 4, and there is a sharp voltage drop when a cortical wall is drilled through.

Figure 4. Drilling signals vs. time

A threshold has to be set up so that during the drilling process, if a voltage drop larger than the threshold occurs, the controller will record as a break through of the cortical wall. When the second break through of the cortical wall is detected, the power of the drill will be cut off. This threshold should not be too small, or small fluctuations in the drilling signals might be mistakenly recognized as drilling through the cortical wall. On the other hand, if the threshold is too large, the detection of the second break through of the cortical wall will be slow, which will increase the amount of drill bit protrusion (overshoot) when a hole is completed. In particular, the threshold should be less than or equal to the voltage drop between the two neighboring signals right after the second peak, as indicated in Figure 4.

The torque required when drilling a bone depends on the type, diameter, and rotational speed of the drill. Therefore the threshold discussed above may vary from one drill to another. For example, Figure 5 compares the plot of drilling signals vs. time for drill diameter 2.5mm, feed rate 0.41mm/s, and that for drill diameter 1.0mm, feed rate 0.6mm/s. The rotational speed is 11,500 rpm in both cases. As shown in the figure, the drilling signals of both cases have a similar pattern, but the magnitude of the voltage drop for drill diameter 2.5mm is much larger.

Figure 5. Drilling signals vs. time for different drill diameters

For this reason, a calibration mode is provided in our system. When a new drill is used with our system, it should be tested several times in the calibration mode to decide the proper threshold. In the calibration mode, the voltage drop between the two neighboring signals right after the second peak indicated in Figure 4 will be output. Table 1 shows the voltage drops of 8 test runs in the calibration mode for drill diameter 1mm, rotational speed 11,500rpm. The minimum voltage drop in these 8 test runs is 0.13 volt. Therefore the threshold for this drill is chosen to be 0.13 volt. A look-up table of drills and corresponding thresholds can be stored in if many drills are to be used with the system. We can also observe from Table 1 that this voltage drop seems to be independent with the feed rate.

Table 1. Voltage drop of 8 test runs in the calibration mode

Test

1

2

3

4

5

6

7

8

Feed rate (mm/s)

1.74

1.31

1.14

0.99

0.76

0.47

0.6

0.55

Voltage drop (V)

0.23

0.13

0.20

0.24

0.21

0.17

0.23

0.18

4. Drilling tests on human skulls

This system has undergone extensive drilling tests on real human skulls under various cutting conditions, using both industrial drills and surgical drills. Most of these tests were performed in a fully automatic feed mode, as shown in Figure 6. The power line of the drill was plugged into the control box connected to a personal computer. The drill clamped to the feed carriage, which was also connected to the personal computer, was manually led to the point to be drilled on the model head. Then the joints of the supporting arm were held solid by turning the knob on the supporting arm. The rest of the drilling operations were all controlled from the personal computer. A piece of real human skull was fixed on the model head for the drilling tests. When a hole was completed, the power of the drill was automatically cut off, and the overshoot of the drill bit was measured.

Figure 6. The tests are performed in a fully automatic feed mode

First an industrial drill was used in the tests. The diameter of the drill was 1mm, effective length 20mm, and the rotational speed was 11,500rpm. The feed rate was fixed at 0.57mm/s and tested for 12 times. Then the feed rate was varied from 0.5mm/s to 2.0mm/s and tested for 13 times. All 25 tests were successful. There were no unexpected failure, and the overshoots of all drilling tests were well less than 2mm. The average overshoot for the 12 times with fixed feed rate was 0.68mm, and the stand deviation was 0.12mm. The average overshoot for the 13 tests when the feed rate varied was 0.66mm, and the standard deviation was 0.20mm.

One factor that affects the amount of overshoot is the shape of the drill head and the diameter of the drill. The drill used in the tests has a 120-degree triangular drill head, as shown in Figure 7. This triangular part will protrude beyond the bone when the bone is drilled through. As shown in Figure 7, the amount of the protrusion of the drill head p is 0.29mm for drill diameter 1.0mm, and p is 0.72mm for drill diameter 2.5mm. We also performed 6 tests using drill diameter 2.5mm, rotational speed 11,500rpm, and feed rate 0.57mm/s. The average overshoot was 1.03mm, and the standard deviation was 0.14mm.

Figure 7. The shape of the drill head

Another factor that affects the amount of the overshoot is the feed rate. Figure 8 shows that the amount of overshoot strongly correlates to the feed rate. For the 13 tests with varied feed rates, Pearson correlation coefficient was 0.87. The maximum feed rate in our tests was 2.0mm/s. At this feed rate we obtained the maximum overshoot 1.1mm. The controller is modified to further reduce the amount of overshoot. As shown in Figure 9, the feed rate is reduced when break through of the first cortical wall is detected. We performed 6 tests using this control scheme, with drill diameter 2.5mm, rotational speed 11,500rpm, and feed rate 0.57mm/s. The average overshoot was reduced to 0.73mm, and the standard deviation was 0.08mm.

Figure 8. Correlation between overshoot and feed rate

Figure 9. The feed rate is reduced when drill-through of the first cortical wall is detected

Finally a surgical drill is used in the drilling test. The diameter of the drill is 1.5mm, and the effective length is 30mm. One major difference between the industrial drills and the surgical drills is that, surgical drills often have very high rotational speeds. Three rotational speeds were tried: 15,000rpm, 45,000rpm, and 75,000rpm. Though the electric currents required at different rotational speeds are different, the voltage signal transferred by the control box of the system exhibits similar drop when breaking through the outer shell of the bone. Initial calibration was done to find the thresholds used in different rotational speeds. Drill bit break through were successfully detected in all 3 speeds, and the overshoots were well less than 2mm.

5. Discussions and conclusions

This paper presents a modular mechatronic system for automatic bone drilling in surgery. The development of a “modular system” that is compatible with motor-driven drills that are commercially available, rather than developing a new surgical drill, is emphasized. Integrating with an optical positioning device, completely automatic bone drilling can be achieved by our system. Brain surgery is usually carefully planned using Computerized Tomography (CT) and Magnetic Resonance Imaging (MRI). Current frameless optical positioning systems can integrate with the 3D image obtained from CT or MRI for image guided surgery. In particular, BrainLab VectorVision currently used in Chang Gung Memorial Hospital, where this project is performed, uses the “passive marker technology.” The two marker spheres on the VectorVision “probe” reflect infrared flashes emitted by the camera-system. The cameras capture the marker reflections and the system converts each marker’s spatial position.

To integrate with VectorVision, 3 registration markers are attached on the feed carriage of our system, as shown in Figure 10. To calibrate this additional instrument, the surgeon simply touches the probe to each registration marker in any order to complete full registration, and then this instrument can then be used and visualized simultaneously.

Figure 10. Registration markers

Though the developments and testing of this system were focused on brain surgery, this system should be useful for all surgeries that involve bone drilling.

References

1.      Allotta, B., Giacalone, G., and Rinalidi, L., 1997. “A hand-held drilling tool for orthopedic surgery,” IEEE/ASME Transaction on Mechatronics, vol.2, no.4, p. 218-229.

2.      Allotta, B., Belmonte, F., Dario, P., and Rinalidi, L., 1996. “The control of penetration in a mechatronic drill for orthopedic surgery,” Proc. Multiconf. Computational Engineering in Systems Applications – Symp. Robotics and Cybernetics, Lille, France, pp. 222-227.

3.      Baker, D., Brett, P.N., Griffiths, M.V., and Reyes, L., 1996. “A mechatronic drilling for ear surgerya case study of some design characteristics,” Mechatronics, vol. 6, no.4, p. 461-477.

4.      Baker, D., Brett, P.N., Griffiths, M.V., and Reyes, L., 1996. “Surgical requirements for the stapedotomy tooldata and safety considerations,” IEEE Engineering in Medicine and Biology Society.

5.      Kaburlasos, V.G.; Petridis, V.; Brett, P.; Baker, D., 1997. “On-line estimation of the stapes-bone thickness in stapedotomy by learning a linear association of the force and torque drilling profiles,” Intelligent Information Systems, IIS '97, Proceedings, p. 80 –84.

6.      Kaburlasos, V.G.; Petridis, V.; Brett, P.; Baker, D., 1998. “Learning a linear association of drilling profiles in stapedotomy surgery,” IEEE International Conference on Robotics and Automation, Proceedings, vol. 1, p. 705-710.

7.      Glauser, D., Flury, P., Villotte, N., and Burckhardt, C.W., 1991. “Conception of a robot dedicated to neurosurgical operations,” Fifth International Conference on Advanced Robotics. Robots in Unstructured Environment, vol.1, p. 899-904.

8.      Bouazza-Marouf, K., Browbank, I., and Hewit, J.R., 1996. “Robot-assisted invasive orthopaedic surgery,” Mechatronics, vol. 6, no. 4, p. 381-397.

9.      Ong, F.R., and Bouazza-Marouf, K., 1998. “Drilling of bone: A robust automatic method for the detection of drill bit break-through,” Proceedings of the Institution of Mechanical Engineers, Part H: Journal of Engineering in Medicine, vol. 212, no. 3, p. 209-221.

10.  Ong, F.R., and Bouazza-Marouf, K., 1999. “Detection of drill bit break-through for the enhancement of safety in mechatronic assisted orthopaedic drilling,” Mechatronics, vol. 9, no. 6, p. 565-588.