A System for Laparoscopic Surgery Ergonomics and Skills Evaluation

Abstract

This article presents a system for tracking, recording, and analysis of instrument and surgeon's arm motion in minimally invasive surgeries. The captured trajectories can be objectively analyzed for both ergonomic assessment and skills evaluation. The system consists of two special infrared (IR) markers that are used for 6 degrees of freedom (DOF) laparoscopic instrument tracking and a set of 3DOF IR markers attached to elbows and shoulders. A compact IR camera tracks and records the markers during a standardized training task (eg, suturing). The instrument markers were purposely designed to provide good tracking while minimizing their volume. The accuracy of the instrument markers was evaluated showing a root mean square error of 0.61 mm, 1.0 mm, and 2.4 mm at distances from the camera of 0.5 m, 0.68 m, and 1 m respectively. Furthermore, some sample trajectories were recorded during an in-trainer suturing task. The Results section presents the values of basic skills metrics computed from the acquired data.

Introduction

L aparoscopic surgeries have become more common primarily because of their benefit to the patient. Although they provide shorter patient recovery time, they can be more challenging to perform. Urologists can be forced into uncomfortable and potentially harmful postures by the instruments they use, which may lead to long-term strain injuries (eg, laparoscopic surgeon's thumb).¹ To identify and rectify such problems, ergonomists evaluate working postures using either direct observation or motion analysis systems.² Person and associates³ assessed the ergonomics of laparoscopic surgery, showing that the ergonomic scores were high throughout the procedure, especially for the wrist.

Because laparoscopy is a relatively new procedure with a steep learning curve,⁴ many surgeons are undergoing training while new instruments are designed or improved. Currently, expert surgeons assess residents through direct observation. Alternate evaluation approaches based on instrument tracking were proposed in the last decade.^5
–7 These use three-dimensional positional data provided by tracking devices that are attached to instruments and/or the surgeon's hands. Dosis and colleagues⁸ presented an evaluation of a motion analysis system—Imperial College Surgical Assessment Device. Resident surgical skills were ranked based on time to task completion, number of left-hand motions, number of right-hand motions, left path length, and right path length. Research is still ongoing to find the most relevant metric for skills assessment.^9,10

All the previous studies used off-the-shelf tracking systems that are designed for long-range tracking. While these systems provide the needed data, they are expensive and not used at their maximal capacity. In this article, we present a system that is purposely designed and built for laparoscopic instrument tracking—LapTrack. The system comprises a stereoscopic camera that tracks custom instrument and body markers. LapTrack data can be used both in ergonomic analysis and skills evaluation in a simulated operative environment for training.

Materials and Methods

Overall system design and cost

The system consists of several infrared (IR) markers that are attached to the laparoscopic instruments, and elbows and shoulders, respectively. A BumbleBee^® 2 stereoscopic camera by Point Grey Research ($1895) that is fitted with IR filters (FSR-RG780 from Newport; $54) tracks the markers during a standardized training task (eg, suturing). Two custom-designed markers provide laparoscopic instruments' positions while several other individual light-emitting diode (LED) markers provide arm and posture information. The cost of each instrument marker was $50.

Laparoscopic instrument markers

The instrument marker is composed of two individual rings and a helix. The marker presented in Figure 1A comprises an acrylic cylinder covered with aluminum film with the helical pattern cut out. Infrared LEDs are inserted into small holes at the ends of the cylinder to illuminate the marker's helical pattern through the cut-out areas of the film. The overall size of the marker is 25.4 mm diameter and 50.8 mm in length. The camera filters ensure that only the IR marker is visible in the image, as in Figure 1B. Custom software processes the real-time stereoscopic images and provides the 6 degrees of freedom position of the marker with respect to the camera.

FIG. 1.

(a) The camera and instrument marker; (b) software image of marker.

The accuracy tests were performed using a Thermo CRS robotic manipulator rated with a repeatability of±0.05 mm. The robotic gripper with the marker attached to it was moved in 30 different points on a plane parallel with the camera. Three sets of tests were performed with the camera placed at distances of 500, 680, and 1000 mm away from the marker. The positions provided by the robot's encoders were compared against the positions provided by the tracking system. First, both tracker and robot points were represented in the same coordinate system. Then, the errors between the robot positions and the marker positions were computed.

Elbow, shoulder, and wrist markers

To evaluate the ergonomics of the instruments, several other parameters have to be recorded, such as hand, wrist, and elbow positions. These are derived using individual LEDs that are attached to elbows and shoulders. Assuming that the hand is in contact with the instrument handle, the hand position can be estimated from the instrument marker position. Figure 2A shows all the markers that are attached to a person.

FIG. 2.

(a) All markers attached to a subject; (b) stick figure of that subject.

The wrist position is measured using resistive flex sensors. The resistance of these sensors is a function of the amount the sensor is bent, the bend radius, and the pressure (if any) that is applied on the sensor.

Ergonomic analysis and skills assessments

The following measures can be computed for the ergonomic analysis of the wrist:

1. t_opt The amount of time for which the wrist angles are within±20% from the neutral position.

2. t_extreme The amount of time for which the wrist angles are within±20% from the extreme values.

3. p_opt The percent of the total procedure time for which the wrist angles are within±2 0% from the neutral position.

4. p_extreme The percent from the total procedure time for which the wrist angles are within±20% from the extreme values.

The system also records raw position data for instrument, hand, elbow, and shoulder. The postprocessing provides the following measures:

1. Hand range of motion for left and right hands: The range of motion is the maximum distance between two points belonging to the analyzed trajectory.

2. Posture angle data: Mean left elbow angle, maximum left elbow angle, minimum left elbow angle, mean right elbow angle, maximum right elbow angle, minimum right elbow angle, mean left forearm-instrument angle, maximum left forearm-instrument angle, minimum left forearm-instrument angle, mean right forearm-instrument angle, maximum right forearm-instrument angle, minimum right forearm-instrument angle are computed automatically from the tracking data. The software can also provide the histograms for angle distributions.

In addition to the numeric values, stick figures can also be displayed for any time instance of a given experiment, as shown in Figure 2B. Recorded trajectories can be used for surgical skills evaluation, as described in the next section.

Results

System accuracy

The system provided a similar trajectory with that of the robot; some slight errors are evident in the direction corresponding to the distance from the camera. Figure 3 illustrates the error at each set-point position of the trajectory for the 680 mm distance test. Table 1 presents the root mean square (RMS) error computed at different distances. In practice, it is necessary to place the camera about 600 mm from the instruments to cover the entire range of motion. The results show that the system provides good tracking accuracy (1 mm RMS) in the range relevant for the intended application.

FIG. 3.

Error at each set-point position of the trajectory for the 680 mm distance test.

Table 1.

Root Mean Square Error Computed at Different Distances

Distance	500 mm	680 mm	1000 mm
RMS error	0.61 mm	1 mm	2.4 mm

RMS=root mean square.

Trajectory analysis sample

One expert (E) and two novices (M and N) with different familiarity with the procedure performed a suturing task three times. The trajectories of both instruments were recorded, and two different skills metrics were computed. The first one was the trajectory length. Table 2 presents the values for the three subjects. The mean length of the expert trajectories (0.12–left instrument; 0.12–right instrument) is smaller than those of the novices (0.69–left instrument M, 0.37–right instrument M; 0.21–left instrument N; 0.2741–right instrument N). This is in agreement with previously reported results showing a correlation between instrument trajectory length and skill.⁹

Table 2.

Trajectory Lengths for an Expert and Two Novices

Experiment #	Left E	Right E	Left M	Right M	Left N	Right N
1	0.1666	0.1660	0.2825	0.2473	0.1440	0.1771
2	0.0965	0.0883	0.2708	0.3264	0.1559	0.1941
3	0.1047	0.1171	0.9799	0.5519	0.3559	0.4510
Mean	0.1226	0.1238	0.6955	0.3752	0.2186	0.2741

E = expert; M, N = novice.

The second metric was the mean distance between trajectories. For any task, one can define an optimal trajectory, or this can be acquired from an expert. The mismatch between the trainee and the ideal trajectories provides means to rank his or her skills. Table 3 shows that the metric performs well for the left instrument; the mean distance between expert trajectories (0.4) is lower than the mean distance between expert and novice trajectories (1.4 between E and M; 1.3 between E and N). The results for right instrument trajectories (Table 4) were inconclusive. The mean distance between expert and novice M (0.4) equals the mean distance between expert trajectories (0.4). The cumulative average, however, shows that the distance between expert trajectories is 0.4, whereas the distances between E and M and E and N are, respectively, 0.9 and 1.15.

Table 3.

Mean Difference (in cm) Between Left Instrument Trajectories

	E	M	N
E	0.4	1.4	1.3
M		0.6	0.2
N			0.3

E=expert; M, N=novice.

Table 4.

Mean Difference (in cm) Between Right Instrument Trajectories

	E	M	N
E	0.4	0.4	1.0
M		0.5	0.3
N			0.4

E=expert; M, N=novice.

The initial validation results indicate that LapTrack data can discriminate between novice and expert surgeons based on instrument trajectory length. Currently, the system is undergoing a comprehensive validation study that involves 30 subjects with different surgical skill levels.

Role in Endourology

Urologists are on the leading edge of laparoscopic surgery. It is now considered standard of care to perform donor nephrectomies laparoscopically; innovations in laparoscopic prostatectomies and cystectomies are emerging. Gofrit and coworkers¹¹ assessed the injuries related to laparoscopic urology surgery and found the laparoscopic operating theater a hostile ergonomic environment. Furthermore, new instrument design/evaluation necessitates data for ergonomics and skills analysis.

This article demonstrates an accurate and precise tracking-based system for ergonomics and skills evaluation in laparoscopy. The goal of this work is to ultimately provide the location and orientation data of the markers in three-dimensional space for later processing. The results show that the system can track instrument motions and surgeon's postures with millimetre accuracies at reasonable distances from the camera. Different metrics used for ergonomics assessment or laparoscopic skills evaluation can be computed from the recorded trajectories.

Footnotes

Acknowledgment

The project was partially supported by St. Joseph Hospital Surgical Associates Clinical Research Grant, Natural Sciences and Engineering Research Council of Canada, and Canada Foundation for Innovation.

Disclosure Statement

No competing financial interests exist.

Abbreviations Used

References

Horgan

, O'Riordan

, Doctor

. Neuropraxia following laparoscopic procedures: An occupational injury. Minim Invasive Ther Allied Technol, 1997; 6:33–35(am)

Genaidy

, Al-Shedi

, Karwowski

. Postural stress analysis in industry. Appl Ergon, 1994; 25:77–87.

Person

, Hodgson

, Nagy

. Automated high-frequency posture sampling for ergonomic assessment of laparoscopic surgery. Surg Endosc, 2001; 15:997–1003.

Raja

. The impact of the learning curve in laparoscopic surgery. World J Laparosc Surg, 2008; 1:56–59.

Shah

, Darzi

. Surgical skills assessment: An ongoing debate. BJU Int, 2001; 88:655–660.

Riener

, Reiter

, Rasmus

et al. Acquisition of arm and instrument movements during laparoscopic interventions. Minim Invasive Ther and Allied Technol, 2003; 12:235–240.

Rasmus

, Riener

, Reiter

et al. In vivo kinematic measurement during laparoscopic cholecystectomy. Surg Endosc, 2004; 18:1649–1656.

Dosis

, Aggarwal

, Bello

et al. Synchronized video and motion analysis for the assessment of procedures in the operating theater. Arch Surg, 2005; 140:293–299.

Egi

, Okajima

, Kawahara

et al. Scientific assessment of endoscopic surgical skills. Minim Invasive Ther Allied Technol, 2010; 19:30–34.

10.

Leong

, Nicolaou

, Atallah

et al. HMM assessment of quality of movement trajectory in laparoscopic surgery. Med Image Comput Comput Assist Interv, 2006; 9:752–759.

11.

Gofrit

, Mikahail

, Zorn

et al. Surgeons' perceptions and injuries during and after urologic laparoscopic surgery. Urology, 2008; 71:404–407.