Ann Thorac Surg 2002;74:624-628
© 2002 The Society of Thoracic Surgeons
Review
Manual control and tracking—a human factor analysis relevant for beating heart surgery
Volkmar Falk, MD, PhD*a
a Department of Cardiac Surgery, Heartcenter, University of Leipzig, Leipzig, Germany
* Address reprint requests to Dr Falk, Klinik für Herzchirurgie, Universität Leipzig, Herzzentrum, Strümpellstr. 39, 04289 Leipzig, Germany
e-mail: falv{at}medizin.uni-leipzig.de
 |
Abstract
|
|---|
Manual control and tracking are fundamental to human factors and define a metric framework which determines the limits of surgical precision. This review provides a brief analysis of factors that are relevant for targeted motions. Knowing and accepting the limitations of human performance may help to optimize performance in off-pump surgery.
|
Introduction
|
|---|
Despite more widespread use and improved techniques, beating heart surgery is still described as "more demanding" or "technically more challenging." Substantial learning curves have been published demonstrating higher patency rates, less anastomotic stenosis, more complete revascularizations, as well as less need for conversions in more recent series [1–8]. While these learning curves represent some skill acquisition over time (or development of an internal model of the task), they also reflect better patient selection based on criteria that facilitate beating heart surgery. The important role of human factor research in cardiac surgery has recently been outlined [9, 10]. This includes the study of social interactions, coping strategies, stress management and behavior in the operating room, and the analysis of learning curves, as well as a basic understanding of ergonomics and human-machine interface technology. The purpose of this review is to provide an analysis of manual control and tracking, and to outline some of the cognitive elements that are important for targeted motions. While these factors present only a small part in the field of human factor research, they may have some implications for the practice of beating heart surgery.
|
Physiologic parameters of hand motion
|
|---|
Movements involving the fingers and the hand are typically performed at low frequencies and, even in skilled activities, intended movements occur in the 4 to 7 Hz frequency range [11]. The average acceleration-deceleration impulse bandwidth for human hand movements (bandwidth is defined as the range of frequencies that contribute to the total acceleration-deceleration impulse) during target acquisition is 5.6 Hz for the first and 9.2 Hz for subsequent impulses, with an average period of 265 ms for the first and 218 ms for succeeding impulses, respectively [12]. However, studies of tracking performance in which the operator must follow a continuously varying input signal yield lower bandwidth estimates of about 1 Hz [13]. Accordingly, at a heart rate of 60/min (equaling 1 Hz), most humans (including surgeons) will be already at their natural limit for tracking a moving target. During grasping and pointing movements, the velocity profiles of the motion are usually bell-shaped with an initial phase of acceleration followed by a deceleration phase. Peak and mean velocities of targeted motions decrease with smaller size objects and smaller distances towards an object [14]. The typical peak and mean velocity of the wrist during reaching motions is 0.6 to 1.2 m/s and 0.3 to 0.5 m/s, respectively [15, 16].
Human hand movement has a number of inherent involuntary components, including tremor, jerk, and a low frequency drift. In one study, the ability of microsurgeons attempting to hold still a microsurgical instrument yielded a window of width of 202 ± 110 µm. Actuating the instrument (by pressing a button at the shaft) increased the window to some 538 ± 302 µm. For holding still and actuating, 81.8% and 95.9% of the total power spectral densities occurred at the lower end of the spectrum below 1 Hz [17]. This is an important finding because this low frequency component of error overlaps in frequency with voluntary components of motion during tracking.
Typically microsurgeons obtain relative positioning in the range of 100 to 200 µm, and occasionally 50 µm have been achieved (with rested elbows and microscopic view) [18]. In other words, the geometric accuracy to aim a needle at the desired target (arterial wall) under ideal conditions (nonmoving target = arrested heart) will be in the range of 0.1 to 0.2 mm. In small coronaries with a diameter of 1 mm, this position error already equals 20% of the vessel diameter (that is, at a rested object without additional inaccuracy caused by motion artifcats). This may partially explain the fact that the rate of early bypass occlusions dramatically increases with smaller vessel diameter [19].
Another source of performance limitation is the so-called psychological refractory period (PRP). This period lasts for about 300 msec and separates successive outputs from the decision-making stage. This is in addition to the minimum reaction time delay that is approximately 180 msec for a visual input. Interestingly, the PRP is rather fixed and does not change with experience [13].
|
Manual control and tracking
|
|---|
Working on the arrested heart presents a self-paced or internally paced task. With self-pacing, the operator can control the rate of movement of the target being followed (ie, anastomotic suture). During periods of low variability (good vessel, anastomotic side wall) the operator can speed up, while during periods of high variability (calcified area, tip of anastomosis) the operator can slow down the task, keeping the demands of the tracking or control task near the limits of his or her abilities. Beating heart bypass grafting requires tracking of an externally paced target, namely the moving target vessel. In this situation, the operator lacks one important dimension of control [13]. Motion of the heart occurs in three dimensions and can be described as a smoothly varying combination of sine waves. There are only few data on the extent of cardiac surface motion. In a pig model, the calculated elliptical area (two-dimensional area; motion in the Z-axis was not measured) of the unrestrained heart covered 73 ± 43 mm2. Using a stabilizer, this motion was reduced to 1.3 ± 0.5 mm2 [20]. In another study, sonomicrometry crystals were used to detect excursion of the left anterior descending coronary artery (LAD). Motion and velocity were analyzed on a beat-to-beat basis in the X, Y, and Z planes using triangulation theory, before and after placement of a vacuum-assisted stabilizer. Analysis of the LAD motion in planar space demonstrated a bi-phasic pattern in all three planes. The stabilizer dampened the motion of the LAD to a monophasic pattern. Stabilization resulted in a significant reduction of excursion (11.36 ± 1.74 vs 5.99 ± 1.30mm; p < 0.05), maximum Cartesian velocity (141.80 ± 29.73 vs 86.55 ± 29.45 mm/s; p < 0.05), and average Cartesian velocity of the LAD (44.30 ± 7.02 vs 21.46 ± 4.54; p < 0.05). Although a significant reduction in the complexity of motion, amount of motion in the direction of greatest travel, and velocity of the LAD after application of a stabilizer was demonstrated, residual motion was present despite state-of-the-art stabilization [21]. Even with immobilization, the amplitude of target motion can exceed 1 mm or almost the target vessel diameter. In this complex multidimensional (the target moves in three planes) tracking task, the operator tries to make the controlled system (needle) follow the changes of the target system (coronary). According to Fitt’s law (see Appendix), the speed of the control movement depends on the index of difficulty (ID) of the response. With the heart beating, a certain time window is open to do a targeted motion (ie, stitch). If the situation is complicated, for example, by wall irregularities (allowing only a small width of the target area in which the operator wishes the movement to terminate), the ID increases and may cause an increase in the time to perform the movement that may be actually too long to allow the movement to be performed without a phase error.
To simulate beating heart conditions, we have developed an endoscopic trainer consisting of a moving plate and a fixed task pad. The plate is driven by three servomotors that are controlled by a single chip computer. The amplitude and frequency can be changed independently by changing the angle of tilt, elevation, and speed to allow for different motions (mimicking stabilzed and unstabilized epicardial surface motion) (Fig 1).
Using a handheld instrument, 10 subjects were asked to touch targets (target diameter was 1 mm) on the plate in different patterns with increasing ID. The time between hits (target touch signal) as well as misses (ground plate touch) were electronically recorded. With increasing index of difficulty and higher frequencies, the time necessary to perform the tasks, as well as the error rate, increased (Figs 2, 3). With frequencies greater than 90/min, tracking became impossible.
View larger version (76K):
[in this window]
[in a new window]
|
Fig 1. Control algorithm of experimental setup: the beating heart simulator is driven by three servomotors (Servo) that are controlled by a microprocessor. Frequency and amplitude of motion can be freely selected. Hits and misses at the taskpad are automatically recorded and put into a database.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig 2. Time (T) to touch two targets in a horizontal line at three different distances (distance increases from A1 to A3) at rest and varying frequencies (35 and 50/min) and amplitudes (st = stabilized, nst = nonstabilized). With increasing distance (A1 < A2 < A3, and thus increased index of difficulty), the time to touch the second target increases. This is in accordance with Fitt’s law (see Appendix). With higher frequencies and larger amplitudes (nonstabilized), the time required to perform the task increases.
|
|
View larger version (24K):
[in this window]
[in a new window]
|
Fig 3. Error (E) rate for a more complex pattern-touch task at rest and different frequencies (35 and 50/min) and amplitudes (st = stabilized, nst = nonstabilized). With higher frequencies and larger amplitudes (nonstabilized), the error rate increases.
|
|
|
Information processing
|
|---|
In an information-processing model of the human operator, three major information processing stages are essential: perception, decision making, and response control. In the perceptual stage, information about the states of the target and controlled system is acquired. This includes visual and tactile inputs. In the decision-making stage, this information is evaluated and used to plan or select a response. In the final response-control stage, the response is executed while the immediate effect is fed back to the operator via the perceptual stage. Response and effect are also stored in the operator’s memory and used to generate "experience." There are three levels of control processes involved in this model of information processing. At the lowest level, the operator’s performance can be modeled as a mechanistic feedback device where the operator makes ad hoc corrections to unanticipated error conditions. On the next level, the operator makes use of inherent predictability in the time varying changes of the target system by using short- and long-term memory. This system is most effective only after experience and practice have allowed development of an internal model of the target system’s behavior. Interestingly, although the decision-making process at this level is more complex to anticipate target system behavior, the overall control task becomes easier because the operator is no longer occupied by a large number of low-level control decisions of the unpredicted error-response type. At the third level, the operator uses general-purpose strategies to overcome his own (response time) as well as hardware limitations. This additional level of response integration allows more time in between control responses [13]. An example for applying this strategy is not to follow the target (coronary) while it is displaced during the cardiac cycle, but to wait for it to return and to execute the directed motion at its starting position which can be anticipated based on a similar motion during previous beats ("wait and see" strategy). This behavior can be observed in most surgeons performing beating heart surgery and confirms experimental results that show that experienced operators take advantage of the redundancy in the target’s motion. Less skilled operators sample target position information of a constantly moving target more frequently. This can lead to an information overflow with impairment of performance. To take advantage of redundancy in the track of the target, an internal model must be developed by the operator. This model allows the operator to make predictions about the future state (or position) of the target and minimizes the need to sample excessively and process information from external sources. Another useful strategy in tracking sine-wave motions is the locking strategy. The operator attempts to generate a sine-wave response that matches the apparent frequency and amplitude of the signal being followed. With a locking strategy, operators can track single sine-wave signals up to 2 Hz before going out of phase with higher frequencies. However, these results apply only for tracking of single sine-waves. The beating heart however moves as a combination of various sine-waves with different amplitude and frequency. In this scenario, locking strategies are less effective. Correct tracking (correct amplitude and in phase) is therefore unlikely to occur above the normal human bandwidth of 1 Hz or a heart rate greater than 60 beats/min (= 1 Hz).
|
Implications for beating heart coronary artery bypass grafting
|
|---|
The described factors define the metric framework for manual control as well as some of cognitive elements that set natural limits for surgical precision. Most of the suggestions that can be derived from this analysis lead to trivial recommendations that reflect the daily life experience in the OR. Obviously surgeons should strive for the best possible immobilization. This includes selection of the appropriate stabilizer and choosing the right retractor to minimize epicardial surface motion. Given the position error of 0.1 mm which will apply for most surgeons, large target vessels should be selected for beating heart surgery as the diameter to position error ratio increases with small vessel size. In large vessels, target width can be larger, decreasing the index of movement difficulty and thus decreasing the time necessary to perform the movement. As a consequence, successful tracking with less phase error is more likely to occur.
Numerous observations imply that beating heart surgery preferably should be performed at heart rates less than 60/min. At this rate, heart movement falls within the maximum human bandwidth for tracking complex motions (1 Hz). Interestingly, most surgeons intuitively accepted this limit in the early days of beating heart surgery and recommendations were made to use beta-blockers or adenosine to slow the heart rate below 60 bpm during off-pump surgery [22, 23]. With the latest generation of stabilizers, residual motion of the beating heart is minimized and most surgeons feel comfortable with higher frequencies. Nevertheless, response time lags increase with higher frequencies (higher frequencies render the signal less predictive and increase the information-processing load for the operator). Thus, in order to minimize the information-processing load, lower frequencies of the target are desirable [24].
In order to develop a successful "wait and see" strategy (third level of control in tracking motions), it may be useful to temporarily overpace patients who are in atrial fibrillation. Even at the cost of a higher frequency, this will restore a predictable motion pattern and thus reduce the bandwidth requirements for the surgeon.
Removal of distractors that compete for the operator’s attention is mandatory in order to prevent information-processing overload. This includes any visual or auditory input that is not relevant to the tracking task (this does not necessarily mean that music should be banned in the OR). It is equally important to minimize background noise. Background noise in this context refers to additional motion by the operator that inflects on the intended tracking motion, ie, muscular tremor. Tremor reduction is achieved by allowing oneself enough rest, avoiding stress in the OR, and by working in the best ergonomic position.
As outlined above, it is important to build an internal model that allows the operator to utilize inherent redundancy in system behavior and to predict future system states or, in other words, to develop an effective "wait and see" strategy. To build an internal model of the beating heart without putting the patient at an additional risk, the use of beating heart training models [25, 26] or simulators may be useful.
|
Appendix
|
|---|
According to Fitt, the Index of Movement difficulty (ID) is defined as
 | ((1)) |
where A is the amplitude of the intended movement (distance) and W is the width of the target area in which the operator wishes to terminate the movement. It is obvious that with an numerical decrease in W (increasing the required accuracy), the ID increases.
The time needed to complete a movement TM can be predicted from its ID (Fitt’s law):
 | ((2)) |
where km is a delay constant with a typical value of 0.177 s for hand movements and cm (s/bit) is a measure of information handling capacity, typically 0.1 s/bit. From Equation 2, it can be concluded that, with an increase in ID, the time to perform the movement increases. This law is made for single-step tracking tasks and becomes inaccurate for constantly moving targets. A modification that also takes the velocity of the moving target into account has been proposed by Jagacinski and colleagues [27]:
 | ((3)) |
where c, d, and e are regression coefficients dependent upon the order of control. V is the target velocity.
|
References
|
|---|
- Calafiore A.M., DiGiammarco G., Teodori G., et al. Midterm results after minimally invasive coronary surgery (LAST-operation). J Thorac Cardiovasc Surg 1998;115:763-771.[Abstract/Free Full Text]
- Cremer J., Mügge A., Wittwer T., et al. Early angiographic result after revascularization by minimally invasive direct coronary artery bypass. Eur J Cardiothorac Surg 1999;15:383-388.[Abstract/Free Full Text]
- Diegeler A., Falk V., Matin M., et al. Minimally invasive coronary artery bypass grafting without cardiopulmonary bypass: early experience and follow-up. Ann Thorac Surg 1998;66:1022-1025.[Abstract/Free Full Text]
- Arom K.V., Flavin T.F., Every R.W., et al. Safety and efficacy of off-pump coronary artery bypass grafting. Ann Thorac Surg 2000;69:704-710.[Abstract/Free Full Text]
- Hart J.C., Spooner T., Pym J., et al. A review of 1582 consecutive Octopus off-pump coronary bypass patients. Ann Thorac Surg 2000;70:1017-1020.[Abstract/Free Full Text]
- Diegeler A., Matin M., Falk V., et al. Indication and patient selection in minimally invasive and off-pump coronary artery bypass grafting. Eur J Cardiothorac Surg 1999;16(Suppl 1):S79.[Abstract/Free Full Text]
- Jegaden O., Mikaeloff P. Off-pump coronary artery bypass surgery. The beginning of the end?. Eur J Cardiothorac Surg 2001;19:237-238.[Free Full Text]
- Diegeler A., Matin M., Kayser S., et al. Angiographic results after minimally invasive coronary bypass grafting using the minimally invasive direct coronary bypass grafting (MIDCAB) approach. Eur J Cardiothorac Surg 1999;15:680-684.[Abstract/Free Full Text]
- Carthey J., de Leval M.R., Reason J.T. The human factor in cardiac surgery: errors and misses in a high technology domain. Ann Thorac Surg 2001;72:300-305.[Abstract/Free Full Text]
- De Leval M.R., Carthey J., Wright D.J., Farewell V.T., Reason J.T. Human factors and surgical outcomes: a multi-center study. J Thorac Cardiovasc Surg 2000;119:661-672.[Abstract/Free Full Text]
- Kunesch E., Binkofski F., Freund H.J. Invariant temporal characteristics of manipulation hand movements. Exp Brain Res 1989;78:539-546.[Medline]
- Draper J.V. Human factors in telemanipulation: perspectives from the Oak Ridge National Laboratory experience. In: Kom W.S., ed. . Telemanipulator technology and space robotics. Boston: Proc SPIE, 1994:162-174.
- Knight J.L. Manual control and tracking. In: Salvendy G., ed. Handbook of human factors. New York: John Wiley & Sons, 1987:182-218.
- Darling W.G., Cole K.J., Abbs J.H. Kinematic varability of grasp movements as a function of practice and movement speed. Exp Brain Res 1988;73:225-235.[Medline]
- Gentilucci M., Castiello U., Corradini M.L., Scarpa M., Umilata C., Rizzolatti G. Influence of different types of grasping on the transport component of prehension. Neuropsychologia 1991;29:361-378.[Medline]
- Jones L. Dextrous hands: human, prosthetic, and robotic. Presence 1997;6:29-56.
- Reviere C.N., Rader R.S., Khosla P.K. Characteristics of hand motion of eye surgeons. IEEE Eng Med Biol Soc 1997;4:1690-1693.
- Taylor R.H., Jensen M., Whitcomb L., et al. A steady-hand robotic system for microsurgical augmentation. Rob Res 1999;12:1201-1210.
- Paz M.A., Lupon J., Bosch X., Pomar J.L., Sanz G. Predictors of early saphenous vein aortocoronary bypass graft occlusion. The GESIC Study Group. Ann Thorac Surg 1993;56:1101-1106.[Abstract]
- Borst C., Jansen E.W., Tulleken C.A.F., et al. Coronary artery bypass grafting without cardiopulmonary bypass and without interruption of native coronary flow using a novel anastomosis restraining device ("Octopus"). J Am Coll Cardiol 1996;27:1356-1364.[Abstract]
- Koransky ML, Tavana ML, Yamaguchi A, Robbins RC. Quantification of mechanical stabilization for the performance of off-pump coronary artery surgery. http://www.ctsnet.org/abstract/6956.
- Bel A., Perrault L.P., Faris B., Mouas C., Vilaine J.P., Menasche P. Inhibition of the pacemaker current: a bradycardic therapy for off-pump coronary operations. Ann Thorac Surg 1998;66:148-152.[Abstract/Free Full Text]
- Robinson M.C., Thielmeyer K.A., Hill B.B. Transient ventricular asystole using adenosine during minimally invasive and open sternotomy coronary artery bypass grafting. Ann Thorac Surg 1997;63(Suppl):S30.
- Crossman E.R. The information capacity of the human motor system in pursuit tracking. Q J Exp Psychol 1960;12:1-16.
- Stanbridge R.L., O’Regan D., Cherian A., Ramanan R. Use of a pulsatile beating heart model for training surgeons in beating heart surgery. Heart Surg Forum 1999;2:300-304.[Medline]
- Izzat M.B., El-Zufari M.H., Yim A.P. Training model for "beating-heart" coronary anastomoses. Ann Thorac Surg 1998;66:580-581.[Abstract/Free Full Text]
- Jagacinski R.J., Repperger D.W., Moran M.S., Ward L.B., Glass B. Fitts’ law and the microstructure of rapid discrete movements. J Exp Psychol 1980;6:309-320.
This article has been cited by other articles:
|
|
|
|
 
V. Falk and F. W. Mohr
Minimally Invasive Myocardial Revascularization
Card. Surg. Adult,
January 1, 2008;
3(2008):
697 - 710.
[Full Text]
|
|
|
|
|
|
|
D. de Canniere, G. Wimmer-Greinecker, R. Cichon, V. Gulielmos, F. Van Praet, U. Seshadri-Kreaden, and V. Falk
Feasibility, safety, and efficacy of totally endoscopic coronary artery bypass grafting: Multicenter European experience
J. Thorac. Cardiovasc. Surg.,
September 1, 2007;
134(3):
710 - 716.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
T. Watanabe, S. Omata, M. Odamura, M. Okada, Y. Nakamura, and H. Yokoyama
Three-dimensional quantification of cardiac surface motion: a newly developed three-dimensional digital motion-capture and reconstruction system for beating heart surgery.
J. Thorac. Cardiovasc. Surg.,
November 1, 2006;
132(5):
1162 - 1171.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
V. Falk, F. Mourgues, L. Adhami, S. Jacobs, H. Thiele, S. Nitzsche, F. W. Mohr, and E. Coste-Maniere
Cardio Navigation: Planning, Simulation, and Augmented Reality in Robotic Assisted Endoscopic Bypass Grafting
Ann. Thorac. Surg.,
June 1, 2005;
79(6):
2040 - 2047.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
V. Falk, T. Walther, S. Jacobs, R. K. Wolf, and F. W. Mohr
Facilitated MIDCAB Using a Magnetic Coupling Device
Ann. Thorac. Surg.,
February 1, 2005;
79(2):
691 - 693.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
P. Cattin, H. Dave, J. Grunenfelder, G. Szekely, M. Turina, and G. Zund
Trajectory of coronary motion and its significance in robotic motion cancellation
Eur. J. Cardiothorac. Surg.,
May 1, 2004;
25(5):
786 - 790.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
S. Jacobs, D. Holzhey, B. B. Kiaii, J. F. Onnasch, T. Walther, F. W. Mohr, and V. Falk
Limitations for manual and telemanipulator-assisted motion tracking--implications for endoscopic beating-heart surgery
Ann. Thorac. Surg.,
December 1, 2003;
76(6):
2029 - 2035.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
V. Falk, T. Walther, H. Stein, S. Jacobs, C. Walther, A. Rastan, G. Wimmer-Greinecker, and F. W. Mohr
Facilitated endoscopic beating heart coronary artery bypass grafting using a magnetic coupling device
J. Thorac. Cardiovasc. Surg.,
November 1, 2003;
126(5):
1575 - 1579.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
T. Athanasiou, P. Kumar, S. Al-Ruzzeh, O. Aziz, A. Charitou, J. Hart, B. Glenville, and R. Casula
Expanded use of suction and stabilization devices in cardiothoracic surgery
Ann. Thorac. Surg.,
October 1, 2003;
76(4):
1126 - 1130.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
W. D. Boyd and K. D. Stahl
The Janus syndrome: a perspective on a new era of computer-enhanced robotic cardiac surgery
J. Thorac. Cardiovasc. Surg.,
September 1, 2003;
126(3):
625 - 630.
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
