Abstract
Background.
To overcome field of view and ergonomic limitations of standard laparoscopes, we are developing a multi-resolution foveated laparoscope (MRFL), which can simultaneously obtain both wide- and zoomed-in-view images. To facilitate the effectiveness of our MRFL, we have been investigating various ways of organizing and visualizing dual-view multi-resolution images acquired by the MRFL. In our prior study, we implemented and compared 6 display modes for the MRFL, assuming a typical clinical environment where a standard (but limited) resolution monitor is available. To take full advantage of our MRFL, displays having sufficient screen resolutions might be advantageous. The present study aims to further understand the effects of view configurations through displays with a standard high-definition (HD) resolution and a 4K resolution. In this study, we compare 3 display modes for limited-resolution displays against a new mode for sufficient-resolution displays.
Methods.
Twenty subjects performed 3 evaluation trials of a touching task with each display mode in an emulated MRFL environment. Various objective measurements including task completion time and the number of collisions, and subjective preference were recorded.
Results.
The new mode showed a better task completion time than the other modes, while it maintained a low number of collisions similar to the others. Moreover, the majority of participants selected the new mode as their most preferred one.
Conclusions.
With a sufficient display resolution, the coregistration between the unblocked and unwarped wide context view and the high-resolution zoomed-in view offered by the new mode was highly effective on both task performance and user preference.
Keywords: laparoscope, display interface, dual views, multi-resolution visualization, focus plus context, overview plus detail
Introduction
Laparoscopy has been successfully employed in many surgical fields because it provides various advantages to patients over open surgery.1,2 However, the current laparoscopic technologies have several limitations.2,3 A major limitation is a trade-off between limited field of view (FOV) for achieving high spatial resolution vs wide FOV for providing better situational awareness.2–4 Standard laparoscopes (SLs) do not simultaneously provide both high-resolution and wide-angle images through a single scope. This limitation can cause serious situations in surgical operations. With an SL, in order to observe fine details of an area of interest in a surgical field, maneuvers must be usually performed at a close-up or highly zoomed-in view. A highly zoomed-in view brings about the loss of peripheral vision and awareness of potentially dangerous situations occurring outside the immediate focus area of SL.
In the current clinical practice, the FOV limitation is handled by manually moving the entire laparoscope back and forth. In this practice, a trained assistant is required for holding and maneuvering the camera. Where the camera is frequently manipulated by the assistant, it can lead to ergonomic conflicts between the surgeon and assistant (e.g., hand crossover).5
To overcome the FOV and ergonomic limitations of standard laparoscope, robotically assisted techniques have been developed. In these techniques, the camera motion is activated by voice, foot pedal, or head motion without the need of a human camera holder. However, they have drawbacks such as delays in task performance due to faulty recognition of command and/or mechanical control of camera movement, and also require significant practice for efficient setup and use.6 There has also been previous work to develop cameras that have a low-profile zoom lens and support high-definition (HD) image quality with automatic focusing.7 However, this approach alone cannot effectively resolve the problem of loss of situational awareness in zoomed-in views.
To address the limitations described above, we are developing a multi-resolution foveated laparoscope (MRFL).8 A schematic layout of our MRFL in clinical use is shown in Figure 1A. A major advantage of our MRFL is that it can simultaneously acquire both wide-angle image and high-magnification (or optically zoomed-in) image of a surgical field in real time through a single scope. In our MRFL, a two-axis optical scanner placed between the wide-angle and high-magnification imaging probes enables to freely steer the high-magnification field toward any selected region of interest within the wide-angle field without moving the MRFL. In our MRFL system, 2 methods are provided to determine the region of interest for the high-magnification view: automatic and manual. In the automatic method, the system detects the position of instrument tooltip in the wide view, and the detected tooltip position is considered as the center of the region of interest. In this method, the assistant for maneuvering the high-magnification view is not required. In the manual method, the center of region of interest is controlled by 4 direction keys (up, down, left, and down) on a keypad. Whenever an operator presses a direction key, the region of interest moves to the corresponding direction. In this method, the operator uses the keypad only. In both methods, there is no ergonomic conflict between the surgeon and camera operator. Note that in our experiment, we used the automatic method only. Figures 1B and 1C show wide-angle and high-magnification views of a live porcine model captured by our prototype MRFL.
Figure 1.
Demonstration of a MRFL for minimally invasive surgery. (A) Schematic layout of MRFL in clinical use. (B) A wide-angle view captured by a camera through the wide-angle imaging probe. (C) A zoomed-in view captured by a camera through the high-magnification imaging probe. Note that the box in B marks the corresponding region of interest captured through the high-magnification probe. MRFL = multi-resolution foveated laparoscope.
Since our MRFL can simultaneously obtain 2 images having different field resolutions and coverages, it is necessary to study how to effectively present the dual-view multi-resolution images using limited display resources in a clinical environment, where a single monitor is preferable due to space constraints. In our previous work,9 we implemented 6 display modes for display interface of our MRFL, namely, overview plus detail (o + d), picture-in-picture overview plus detail (o + d (pip)), focus plus occluded context (f + oc), fixed focus plus occluded context (ff + oc), focus plus warped context (f + wc), and fixed focus plus warped context (ff + wc) modes, and compared them through a human-subject experiment. All of these display modes assumed the use of a 19-inch-wide liquid crystal display (LCD) monitor having a screen resolution of 1440 × 900 pixels, but they differed from each other in terms of window management and positioning the zoom-view window relative to the viewer. The choice of such a display configuration was based on the assumption that a single monitor with a standard screen resolution is commonly available in a typically clinical environment. In the o + d mode, the wide and zoom views were presented on 2 separate windows side by side (Figure 2A). In the o + d (pip) mode, the wide view was overlaid on the upper-left corner of the zoom view. In the f + oc and ff + oc modes, the zoom view was simply overlaid on the wide view. In the f + wc and ff + wc modes, the zoom-view window was overlaid on the wide-view window, where the wide-view image was warped such that the information hidden behind the zoom-view window became visible in a highly distorted region surrounding the zoom-view window (Figures 2B and 2C, respectively). In the f + oc and f + wc modes, the zoom-view window followed the tooltip in the wide-view image, whereas in the ff + oc and ff + wc modes, it was fixed at the center of wide-view window or display screen. Note that in all the display modes (including the new display mode discussed later), the zoomed-in area in surgical field is automatically determined by the position of instrument (or tooltip) tracked in the wide-view image using computer vision techniques10 and considered as the center of region of interest.
Figure 2.
Display modes compared to the focus plus unwarped context view (f + uwc) mode in this experiment. (A) Overview plus detail view (o + d) mode. (B) Focus plus warped context view (f + wc) mode. (C) Fixed focus plus warped context view (ff + wc) mode. Note that in the f + wc and ff + wc modes, the wide view is warped. (The wooden stick and posts are bent in the wide view.)
According to the results of our previous experiment, the f + wc and ff + wc modes showed the best performance in various aspects, including task completion time and the number of collisions. However, the participants selected these modes as their least preferred one and reported that it was mainly because they were unfamiliar with the warped view. The majority of participants selected the o + d mode as their most preferred mode. Warping the wide view in the f + wc and ff + wc modes was necessary to provide unblocked wide context view in a display having a limited screen resolution. Where a display having a sufficient resolution (e.g., 4K and 8K resolutions) is used, however, the wide view does not have to be warped. Since such high-resolution displays were not popular and widespread when we started developing the MRFL, it was assumed that a limited-resolution display (e.g., 1080p or less) would be used. This assumption of display resolution limit, however, is no longer valid, and the high-resolution displays have been proliferated in operation rooms in the past few years. Thus, it is necessary to evaluate the effects of display modes with displays of adequate resolution and develop new guidelines for MRFL display interface.
Without the assumption of limited display resolution, we implemented a new display mode called focus plus unwarped context view (f + uwc) mode (Figure 3). In this mode, the final displayed image is generated by superimposing the zoom-view image on an upsampled wide-view image. First, the wide-view image obtained by the wide-view camera is enlarged so that the zoom-view and upsampled wide-view images have the same optical magnification. For example, where the magnification ratio of the zoom view to wide view is 3:1, and both the wide-view and zoom-view images have a pixel resolution of 900 × 720, the wide view image is upsampled to a resolution of 2700 × 2160 (= [3 × 900] × [3 × 720]). Second, the zoom-view image is overlaid at the tooltip position on the upsampled wide-view image. Since the zoom-view and upsampled wide-view images have the same magnification, wide and zoom views are smoothly connected. To present the entire final image, a display having the same resolution as (or larger than) the upsampled wide-view image is required. A drawback of this new mode is that compared with the zoom-view area, the wide-view area may be noticeably blurred due to upsampling. The difference in blurriness between the 2 views may affect user performance and preference.
Figure 3.
A new display mode: focus plus unwarped context view (f + uwc) mode. Note that in the f + uwc mode, the wide view is not warped. (The wooden stick and posts are not bent in the wide view.)
In this article, the new display mode is compared with the previously implemented display modes, through a human-subject experiment conducted with an emulated MRFL using a 4K camera. From the previously implemented 6 display modes, we chose the o + d, f + wc, and ff + wc modes (Figure 2) for comparison to the new display mode, based on the results of our previous experiment regarding task performance and participant preference for display mode.
Materials and Methods
Participants
This study was approved by the Institutional Review Board at the University of Arizona. Twenty participants (13 males and 7 females) were recruited from the population of the university. Their ages ranged from 19 to 34 years (M = 23.6 and SD = 3.7). Ten participants also took part in our previous experiment.9 Each participant voluntarily signed the informed consent form. All the participants except for 2 were right handed. No participant had an experience in surgical operation. Note that the number of participants in our experiment was greater than that suggested by Nielsen and Landauer11 for usability study (10–12).
Experimental Setup and User Task
The experimental setup for the previous experiment was reused. The only difference between the previous and this experiment was that in this experiment, a 4K-resolution (3840 × 2160) 27-inch LCD monitor was used, whereas in the previous experiment, a 19-inchwide LCD monitor was used. Figure 4 shows the experimental setup. The black wooden stick had a length of 60 cm and a thickness of .6 cm, and a 6-degree-of-freedom HiBall 3000 tracker (3rdTech Inc, Chapel Hill, North Carolina) was attached to the wooden stick to track the tip position. The participants could reach the posts only through the access hole (diameter = 1.2 cm) located at the right side of and 8 cm away from the 4K camera, while they were seated about 40 cm away from the monitor.
Figure 4.
Experimental setup. (A) An overall view. (B) A close-up view. (C) A right-side view.
A Flea3 FL3-U3-88S2C camera with a 6-mm lens (PointGrey Research Inc, Richmond, Canada) was utilized to emulate the MRFL. The zoom-view and wide-view images having a resolution of 900 × 720 pixels were produced by clipping out and downsampling, respectively, a single image taken by the 4K camera and having a resolution of 2700 × 2160 pixels. As a result, the lateral magnification ratio of the zoom view to wide view was 3:1, and the wide view covered an area 9 times as large as that by the zoom view, which was consistent with the MRFL being developed at our lab. All the display modes were presented in the same 27-inch monitor. In the o + d, f + wc, and ff + wc modes, after arranging the zoom-view and wide-view images, however, the arranged images were enlarged with a magnification factor of 1.83 (= pixel pitch of 19-inch monitor/pixel pitch of 27-inch monitor = .2842 mm/.1557 mm) to maintain the same apparent window size as in the previous experiment. The EyeX eye tracker (Tobii Inc, Stockholm, Sweden) was attached to the 4K monitor to track participants’ gaze positions on the display screen.
A touching task was considered as a user task. In a task trial, the participants were asked to touch all the red posts in numerical order without colliding with posts other than a designated post. A label where a unique number (between 1 and 8; changed across task trials) was printed was attached to the top of each red post. The numbers could be clearly shown in the zoom view only. To maintain the same optimal travel distance of tooltip across task trials, the participants were asked to locate the tooltip at a lower-right corner in the wide view before starting each task trial and return back to the initial location after touching each designated post.
Independent and Dependent Variables
Display mode and repetition were considered as primary and secondary independent variables, respectively. During each evaluation trial, tooltip position and participant’s gaze position on the display screen were recorded by a computer, and task completion time and the number of collisions were collected by the experimenter. Based on these data, the following 6 objective dependent variables were considered for analysis:
Task completion time.
The number of collisions.
Length of gaze point trajectory on display screen.
Moving speed of gaze point on display screen.
Length of tooltip trajectory.
Moving speed of tooltip.
In addition, participant preference for the display mode was obtained using a posttest questionnaire, which asked to select the least and the most preferred display modes.
Experimental Design and Procedure
A 4 × 3 (display mode × repetition) within-subject design was utilized. Each participant took part in an experimental session consisting of introduction, eye tracker calibration, task trials, and posttest questionnaire. In the introduction, the participant read and signed the consent form, filled out a demographic questionnaire, and listened to an explanation of the 4 display modes and user task. In the eye tracker calibration, the participant had to gaze at 7 positions on display screen one by one. In the task trials, the participant performed 1 training trial followed by 3 evaluation trials for each display mode (16 trials in total). The presentation order of 4 display modes was arranged using a digram-balanced Latin square. After finishing all the task trials, the participant filled out the posttest questionnaire and was paid $20. It took 1–1.5 hours to complete a session.
Statistical Analysis
Statistical analysis was performed using the SPSS version 21 (IBM Corp, Armonk, New York). To investigate statistical significance of differences among the display modes and interactions between the display mode and repetition on the objective dependent variables, two-way within-subject analysis of variance (ANOVA) was used. To measure sphericity on the data, Mauchly’s test was performed. Where a significant violation of sphericity was found, the degrees of freedoms were adjusted using the Greenhouse-Geisser procedure. Following the ANOVA, post hoc multiple comparison tests with Šidák correction were performed. For all the statistical tests, α = .05 was used as the significance level.
Results
According to the ANOVA, there were no significant interactions between the display mode and repetition for all the objective dependent variables [F (2.457, 46.687) = .831, P = .463 for task completion time; F (6, 114) = 2.111, P = .057 for the number of collisions; F (2.307, 43.841) = .617, P = .567 for length of gaze point trajectory; F (3.649, 69.324) = .808, P = .514 for moving speed of gaze point; F (6, 114)= 1.925, P = .083 for length of tooltip trajectory; and F (6, 114) = 1.769, P = .112 for moving speed of tooltip]. Since we were interested in main effects of display mode and interactions between the display mode and repetition, main effects of repetition are not presented and considered in this article.
Task Completion Time
The participants carried out the task most quickly in the f + uwc mode, then in the f + wc mode, then in the ff + wc mode, and most slowly in the o + d mode (Figure 5A). The ANOVA showed that there were significant differences among the display modes [F (1.758, 33.406) = 5.433, P = .011]. The multiple comparison tests revealed that the participants completed the task significantly more quickly in the f + uwc mode than in the o + d and ff + wc modes.
Figure 5.
Means and standard deviations of task completion time (A) and the number of collisions (B).
The Number of Collisions
The f + wc and f + uwc modes showed the smallest number of collisions, followed by the o + d mode, then the ff + wc mode (Figure 5B). However, the ANOVA revealed that there was no significant difference among the display modes [F (2.096, 39.824) = .123, P = .893].
Length of Gaze Point Trajectory on Display Screen
The participants’ gaze points on the display screen were most moved in the o + d mode, then in the f + uwc mode, and then in the f + wc and ff + wc modes (Figure 6A). The ANOVA showed that there were significant differences among the display modes [F (1.470, 27.94) = 40.29, P < .0001]. According to the multiple comparison tests, the gaze points were significantly more moved in the o + d mode than in the other modes, and significantly less moved in the ff + wc and f + wc modes than in the f + uwc mode.
Figure 6.
Means and standard deviations of length of gaze point trajectory on display screen (A), moving speed of gaze point on display screen (B), length of tooltip trajectory (C), and moving speed of tooltip (D).
Moving Speed of Gaze Point on Display Screen
The participants’ gaze points were moved fastest in the o + d mode, then in the f + uwc mode, and then in the f + wc and ff + wc modes (Figure 6B). The ANOVA revealed that there were significant differences among the display modes [F (2.124, 40.36) = 91.29, P < .0001]. The multiple comparison tests showed that the gaze points were significantly moved faster in the o + d mode than in the other modes, and significantly moved slower in the ff + wc and f + wc modes than in the f + uwc mode.
Length of Tooltip Trajectory
The tooltip was most moved in the o + d mode, then in the ff + wc mode, then in the f + uwc mode, and then in the f + wc mode (Figure 6C). The ANOVA showed that there were significant differences among the display modes [F (3, 57) = 6.457, P = .001]. According to the multiple comparison tests, the tooltip was significantly less moved in the f + wc and f + uwc modes than in the other modes.
Moving Speed of Tooltip
The tooltip was fastest moved in the ff + uwc mode and most slowly in the f + wc mode (Figure 6D). However, the ANOVA revealed that there was no significant difference among the display modes [F (3, 57) = 1.701, P = .177].
Preference for Display Mode
The majority of participants selected the f + uwc mode as the most preferred one. For the least preferred mode, similar numbers of participants selected the other modes, respectively (Figure 7).
Figure 7.
Subjective preference for display mode.
Discussion
The f + uwc mode showed a better task completion time than the other display modes, while it maintained a low number of collisions similar to the others. Moreover, the majority of participants selected this mode as their most preferred display mode. These results indicate that the unwarped wide context view was highly effective on both task performance and user preference, although it was noticeably blurred due to upsampling. It seems that the drawback of the f + uwc mode (difference in blurriness between wide and zoom views) is not significant. The spatial awareness is obviously easier and faster in the f + uwc mode than in the f + wc and ff + wc modes, since in the latter modes an effort for retrieving correct spatial relationship from the distorted wide view is required. We think that this effort difference led to a difference in task completion time between the f + uwc mode and the f + wc and ff + wc modes.
The participants less moved their eyes at a lower speed in the f + uwc mode than in the o + d mode, but more at a higher speed than in the ff + wc and f + wc modes. This implies that the participants might feel less eye fatigue in the f + uwc mode than in the o + d modes, but more than in the ff + wc and f + wc modes. However, the simple comparison may be insufficient, because the f + uwc mode was presented in a different-size display. Note that for the f + uwc mode, a 27-inch monitor was considered, whereas for the other modes, a 19-inch monitor was assumed. Although all the display modes were presented in the same 27-inch monitor, the apparent image sizes for the o + d, ff + wc, and f + wc modes were maintained the same as those in the previous experiment using a 19-inch monitor. If a larger display size was considered for these modes, the amount and speed of eye movement might increase, which might lead to an opposite result between the f + uwc mode and the ff + wc and f + wc modes.
The participants moved the tooltip significantly less in the f + wc and f + uwc modes than in the other modes, but at a similar speed regardless of the display modes. This implies that they felt less arm fatigue in the f + wc and f + uwc modes than in the other modes. Note that the results of tooltip movement are independent of the display size.
While in the literature we could not find any study where the f + uwc mode was compared to the ff + wc or f + wc mode, we could find several studies where it was compared to the o + d mode. Baudisch et al12 conducted 2 experiments. In their first experiment considering a circuit board inspection task and a map navigation task, task completion time was significantly faster in the f + uwc mode than in the o + d mode. In their second experiment considering a car driving task, the number of collisions with obstacles was significantly lower in the f + uwc mode than in the o + d mode. Chen et al13 conducted 2 experiments involving a single moving target tracking task and a multiple moving target identification task, respectively. Their study revealed that the f + uwc mode showed a significantly faster task completion time than the o + d mode, and the participants preferred the f + uwc mode to the o + d mode. More recently, Seo et al14 conducted an experiment considering a navigation task in a simulated disaster environment, and showed that operator performance was better in the f + uwc mode than in the o + d mode. Although the user tasks in these studies were different from ours and were not directly related to laparoscopic operations, the results from these studies support our results for comparison between the f + uwc and o + d modes. Regarding the number of collisions, while Baudisch et al’s second experiment showed a significant difference between the 2 modes, our experiment did not. Our user task or test environment might be too simple to show a significant difference. Note that Baudisch et al’s test environment was dynamically changed during the task trial, whereas ours was not. In the o + d mode, user’s attention on the wide and zoom views must be divided, because the wide and zoom views are presented at separate locations. For the tasks carried out in an environment where the contents are rapidly changed, the degradation of user performance due to the split attention in the o + d mode may be more significant.
Comparing the results of our previous and this experiments among the o + d, ff + wc, and f + wc modes, the trends of rankings (or shapes of graph charts) among the 3 display modes in this experiment were quite consistent with those in the previous experiment.
For a display providing a screen resolution large enough for the f + uwc mode, the f + uwc mode is not always the best. Where display size or pixel pitch is too small, the apparent size of zoom view may be too small to observe fine details. In this case, 2 options can be considered. One is to digitally zoom in the whole image and crop the area out of the display screen, and the other is to enlarge the zoom-view image only and warp the surrounding wide-view image. The first option has a critical drawback that the peripheral area must be lost, and thus surgeon’s situational awareness must be reduced. The second option is obviously the f + wc mode. For the tasks where the situational awareness in peripheral area is not required or entire surgical field can be clearly shown in a FOV reduced by cropping, the first option would be a good choice since the wide view is not warped. For the other tasks, the second option (f + wc mode) would be better since it still provides peripheral information and showed task performance comparable to the f + uwc mode. On the other hand, where display size or pixel pitch is too large, the amount and speed of user’s eye movement may increase, and this may lead to increase surgeon’s eye fatigue, which can affect task performance. It would be worthy to find out an optimal display size for each display mode.
In our experiment, we did not choose surgeons or surgical trainees as participants. This was because the object-touching task considered in the experiment is a basic and primary task in laparoscopic operation but does not require skilled techniques. We think that for this task, the experimental results from the nonexpert group would be similar to those from surgeon group.
Conclusions
We further understood how different display modes to present dual-view multi-resolution images in a MRFL system affected task performance and user preference, by comparing the newly implemented display mode (f + uwc mode) to the previously implemented display modes (o + d, ff + wc, and f + wc modes). It was found that the unblocked and unwarped wide context view was highly effective on both task completion time and user preference for the touching task which is a basic task for laparoscopic surgical operations. Although the experimental results showed that the f + uwc mode was superior to the o + d, ff + wc, and f + wc modes, the f + uwc mode is not always available. Depending on the display size, screen resolution, and clinical scenarios, other display modes can be still useful. Our future work will be to validate the results with laparoscopic surgeons performing simulated surgical procedures. Another thread of future work will include a study to find out suitable display modes depending on various clinical scenarios.
Acknowledgments
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institutes of Health [grant number 1R01EB18921-01].
Footnotes
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: Drs Sangyoon Lee, Mike Nguyen, and Allan Hamilton have no conflicts of interest or financial ties to disclose. Dr Hong Hua has a patent multi-resolution foveated endoscope/laparoscope (pending) and a patent optical article and illumination system for endoscope (US Patent No. 10 481 386), and has no other conflicts of interest or financial ties to disclose.
References
- 1.Cuschieri A Minimal access surgery and the future of interventional laparoscopy. Am J Surg. 1991;161(3):404–407. [DOI] [PubMed] [Google Scholar]
- 2.Heemskerk J, Zandbergen R, Maessen JG, Greve JWM, Bouvy ND. Advantages of advanced laparoscopic systems. Surg Endosc. 2006;20(5):730–733. [DOI] [PubMed] [Google Scholar]
- 3.Parker WH. Understanding errors during laparoscopic surgery. Obstet Gynecol Clin N Am. 2010;37(3):437–449. [DOI] [PubMed] [Google Scholar]
- 4.Pierre SA, Ferrandino MN, Simmons WN, et al. High definition laparoscopy: Objective assessment of performance characteristics and comparison with standard laparoscopy. J Endourol. 2009;23(3):523–528. [DOI] [PubMed] [Google Scholar]
- 5.Kim SP, Shah ND, Weight CJ, et al. Contemporary trends in nephrectomy for renal cell carcinoma in the United States: Results from a population based cohort. J Urol. 2011;186(5):1779–1785. [DOI] [PubMed] [Google Scholar]
- 6.Bernie JE, Venkatesh R, Brown J, Gardner TA, Sundaram CP. Comparison of laparoscopic pyeloplasty with and without robotic assistance. J Soc Laparoendosc Surg. 2005;9(3):258–261. [PMC free article] [PubMed] [Google Scholar]
- 7.Tsai FS, Johnson D, Francis CS, et al. Fluidic lens laparoscopic zoom camera for minimally invasive surgery. J Biomed Opt. 2010;15(3):030504. [DOI] [PubMed] [Google Scholar]
- 8.Qin Y, Hua H. Optical design and system engineering of a multiresolution foveated laparoscope. Appl Optic. 2016; 55(11):3058–3068. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Lee S, Hua H, Nguyen M, Hamilton AJ. Comparison of six display modes for a multi-resolution foveated laparoscope. Surg Endosc. 2019;33(1):341–351. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Jiang X, Zheng B, Atkins MS. Video processing to locate the tooltip position in surgical eye-hand coordination tasks. Surg Innovat. 2015;22(3):285–293. [DOI] [PubMed] [Google Scholar]
- 11.Nielsen J, Landauer TK. A mathematical model of the finding of usability problems. Paper presented at Proceedings of the CHI 1993 Conference on Human Factors in Computing Systems; April 24-29, 1993; Amsterdam, the Netherlands:206–213. [Google Scholar]
- 12.Baudisch P, Good N, Bellotti V, Schraedley P. Keeping things in context: A comparative evaluation of focus plus context screens, overviews, and zooming. Paper presented at: Proceedings of the CHI 2002 Conference on Human Factors in Computing Systems; April 20-25, 2002; Minneapolis, MN:259–266. [Google Scholar]
- 13.Chen K-W, Lin C-W, Chiu T-H, Chen MY-Y, Hung Y-P. Multi-resolution design for large-scale and high-resolution monitoring. IEEE Trans Multimed. 2011;13(6):1256–1268. [Google Scholar]
- 14.Seo SH, Rea DJ, Wiebe J, Young JE. Monocle: Interactive detail-in-context using two pan-and-tilt cameras to improve teleoperation effectiveness. Paper presented at: Proceedings of the 26th IEEE International Symposium on Robot and Human Interactive Communication; August 28-September 1, 2017; Lisbon, Portugal:962–967. [Google Scholar]