Open Access
Add to BibSonomy
Abstract
Feasibility and clinical utility of a multi-resolution foveated laparoscope (MRFL) was previously tested in a porcine surgical study. The study revealed several clinical limitations of the system including moisture proofing, working distance, image quality, low light performance, color accuracy, size, and weight. In this paper, we discuss the root causes of these limitations and strategies to correct them, present the design and prototyping of a new high throughput multi resolution foveated laparoscope (HT-MRFL), and demonstrate the HT-MRFL prototype performance in comparison to the MRFL and simulated performance metrics.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Since its rise to popularity in the 1990’s, minimally invasive surgery (MIS) has gradually replaced open surgery as the gold standard for numerous procedures including appendectomies and cholecystectomies [1]. Compared to open surgery, laparoscopic MIS has demonstrated reduced post-operative pain, medical costs, recovery times, and hospital stays [2,3]. Despite its many advantages, contemporary MIS is fundamentally limited in safety and efficiency by traditional laparoscopic imaging systems. Standard surgical laparoscopes suffer from two main drawbacks.
First, traditional laparoscopes, like most imaging systems, exhibit a trade-off between instantaneous field of view (FOV) and spatial resolution. Most surgical laparoscopes have full FOVs around 70°. To produce highly detailed images, standard laparoscopes are typically used at working distances (WDs) less than 50 mm. The resulting “keyhole image” only shows a maximum region of interest (ROI) of ∼56 × 42 mm2. As a result, accidents such as instrument-tissue collisions and unintentional charge coupling from energized to non-energized instruments and tissue occur and go unnoticed outside the FOV [4,5] The likelihood of such occurrences involving the laparoscope itself is increased by the nonstationary nature and small working distance of a traditional laparoscopes.
Secondly, traditional laparoscopes are handheld devices. Typically, a surgical assistant continuously holds and repositions the laparoscope throughout an intervention, which causes fatigue. Furthermore, it is common that the surgeon and assistant must “entangle” themselves to position the camera at a suitable viewing angle. Such situations can be awkward, ergonomically taxing, and can limit the ranges of motion of the surgeon and assistant.
In response to the safety and efficiency issues detailed above, Qin et al. designed a multi-resolution foveated laparoscope (MRFL) that simultaneously captures a high resolution, closeup view of the surgical ROI and a wide, low resolution “stadium” view of the entire surgical site [6–11]. The optical design of the MRFL includes tunable lenses and a scanning device that allow the user to keep the laparoscope stationary but change the zoom and position, respectively, of the closeup view relative to the wide view. Clinical testing of the MRFL prototype revealed several limitations including moisture proofing, image quality and brightness, low-light performance, bulk, working distance and color accuracy.
In this paper, we propose and demonstrate a high throughput multi-resolution foveate laparoscope (HT-MRFL) that addresses the clinical limitations of its predecessor. The paper is structured as follows. Section 2 presents the redesigned system and explains how each clinical limitation is resolved. Section 3 briefly discusses system design and prototyping considerations. Section 4 comparatively evaluates the HT-MRFL prototype performance.
2. Optical approach
2.1 MRFL limitations and root issues
The optical layout of the MRFL prototype developed by Qin et al. is shown in Fig. 1. The endoscopic portion of the MRFL prototype features an objective and two relay lenses responsible for collecting the initial image and relaying it to an eyepiece. The eyepiece collimates the light and images an intermediate pupil onto a scanning 90R:10 T plate beam splitter (BS). The wide-view probe (WVP) refocuses the light that passes through the BS to form a wide-angle image of the full surgical field on a detector. The light that is reflected by the BS is directed through the zoom-view probe (ZVP). The ZVP is a Keplerian zooming system that magnifies a portion of the surgical field, called the “foveated field”, and images it onto a second detector in high resolution. The MRFL is designed as a stationary system that is mounted in place at the start of a surgery. A pair of Optotune 10-30-TC electrically tunable lenses (ETLs) facilitate optical zoom of the ZVP between 2X-3X (with respect to the WVP image) and achieve auto-focus to maintain a selected focal depth independent of the variable zoom factor. The scanning BS enables dynamic steering of the foveated field to any region visible in the WVP image without physically moving the system.
An MRFL prototype based on the layout in Fig. 1 was recently tested in an in vivo porcine surgical study to validate its feasibility and clinical utility [12]. The MRFL prototype was utilized successfully to carry out the in vivo porcine surgical tasks while the instrument was mounted in a fixed position on an articulated arm as shown in Fig. 2. Although this in-vivo study validated the key imaging features of the MRFL design, it revealed five major optical and mechanical limitations: (1) the ZVP image contrast is inadequate; (2) the ZVP image is too dim at higher zoom factors and overall lacks low light performance and color accuracy; (3) the nominal working distance of 120 mm is too shallow; (4) the system is bulky and heavy; and (5) condensation forms on the inner surfaces of the objective lens during porcine surgery despite efforts to seal the system. The underlying causes of each problem are discussed below, and correction strategies are presented.
Fig. 2. Photos depicting the setup (left) and use (right) of an MRFL prototype for porcine surgery during a clinical study. The MRFL is mounted to an articulated arm affixed to the operating table railing. It is fixed in position at the beginning of the surgery and left in place until the operation is complete.
The deficiencies of ZVP image contrast, brightness, and color accuracy are resultant of the objective and relay designs, assembly, and budget. Designing dual imaging probes with differing FOVs inherently introduces a constant throughput disparity between the wide-view and zoomed view probes. This is somewhat alleviated by the adoption of asymmetric splitting ratios [11], but the ZVP throughput dependence on zoom factor is unavoidable and significant. Thus, the zoomed-view image is adequately bright at lower magnifications (Fig. 3(a)) but too dark for clinical use at higher magnifications (Fig. 3(b)). The low overall image contrast of the zoomed view, however, is problematic independent of zoom factor and is due to optical design limitations and budget. The MRFL prototype was built as a proof-of-concept and budgeted as such. The numerical apertures (NAs) of the objective and relay lenses were limited by the maximum outer lens diameters that can be accommodated by commercially available 10-mm endoscope packaging sleeves, which contain fiber bundles for illumination and tubes to house the optics. The objective NA inevitably limits the overall light throughput, spatial resolution, and image contrast. Additionally, the MRFL optical design necessitates placement of intermediate image planes near lens surfaces. As a result, surface imperfections and dusts on these lenses significantly decrease image quality. Finally, the MRFL prototype uses two non-medical grade CCD sensors (Teledyne Flir CM3-UC-13S2C). The CM3-UC-13S2C detector is adequate for the wide-view probe, but its mosaic color filtering, 3.75 µm pixel size, and 37.68 dB maximum signal to noise ratio (SNR) exacerbate the zoomed-view throughput concerns and limit the color accuracy for both wide-view and zoom-view probes.
Fig. 3. A demonstration of some of the MRFL prototype drawbacks. The zoom-view image brightness and quality decrease significantly when zoom is changed from (a) 2X to (b) 3X. The extracorporeal portion of the MRFL prototype (c) is bulky and encroaches on instrument working spaces. Condensation (indicated by green arrows) forms inside the objective lens and can be seen in the (d) zoomed view and (e) wide view images.
The MRFL features a low-profile endoscope probe to decrease potential interferences and collision between the laparoscope and surgical instruments. Its 190 mm insertable probe length is about half that of traditional laparoscopes for which probes lengths typically exceed 350 mm. The MRFL was designed with a 120 mm nominal WD to accommodate the shortened insertable length. Though it is substantially larger than a typical 50 mm WD found in standard scopes, clinical testing revealed that an even larger WD is favorable. The ETL focal length ratios can be calibrated to achieve focused 2X-3X zoom at greater (or shorter) working distances, but image quality degrades as the targeted calibration distance strays further from the nominal value. The image quality of the MRFL prototype is not high enough for this trade-off to be practical. Clinical testing revealed that the MRFL prototype is too bulky to work around and too heavy to be mounted efficiently for clinical purposes (Fig. 3(c)). The MRFL prototype features a streamlined optical design aimed at reducing its form factor and bulk. The bulk and weight of the prototype, however, are greatly attributed to the scanning device (Zaber T-OMG), which was, to our knowledge, the only commercially available, electrically controlled scanning device at that time that met the 2D scanning speed, angle range, accuracy, and repeatability requirements.
Lastly, condensation forms on the objective surfaces (Fig. 3(d)-(e)) due to inadequate sealant and application technique. Construction and marine sealants were tried as options but were too viscous to be applied precisely and effectively. Medical-grade optical sealant was extremely effective, but the water-like viscosity and fast setting time made precision application difficult using the equipment available in our lab. Furthermore, the MRFL objective design does not include a window to facilitate moisture proofing.
2.2 Correction strategies
The issues of image quality and brightness, working distance, and moisture proofing are addressed in the optical design of the HT-MRFL objective and relay lenses. Figure 4 shows the optical layout of the HT-MRFL design. An optical window is integrated as the first element in the objective design to accommodate potting and facilitate moisture proofing. Application and testing of the sealant, however, may be contracted to a professional before clinical testing.
One of the inherent challenges for designing an MRFL system is its inadequate illumination through a standard illumination fiber bundle due to a much larger area to be imaged. Another inherent challenge is that the MRFL optical throughput is much smaller than a standard laparoscope of the same entrance pupil diameter due to its substantially longer WD. Furthermore, the optical throughput of the zoomed- probe is inherently much smaller than the wide-view probe due to the optical zoom factor, about ¼ and 1/9 for 2X and 3X zooms, respectively. Therefore, to correct the issues of inadequate image brightness and the throughput disparities discussed above, the HT-MRFL objective and relay lenses must collect enough light by maximizing their optical throughput while adhering to physical and practical design constraints. Given that the working distance of an endoscope is far greater than the entrance pupil diameter (EPD) of the objective, the small angle approximation is applied to express the object-space throughput as,
The HT-MRFL ZVP uses a medical grade camera system (Toshiba IK-HD1). The camera head (IK-HD1H) features a three-CCD detector setup in which each color channel (red, green, and blue) is imaged to a separate individually addressable full-frame sensor. The pixel size of each sensor is 4.65 µm and the camera head’s maximum SNR is 56 dB. The individually controllable sensors for each color channel provide higher levels of color control and correction than those of the Bayer-filtered CM3-UC-13S2C detector, and the larger pixels and SNR increase low light performance. Furthermore, the camera controller (IK-HD1C) is purpose built for medical imaging and features integrated image processing features specifically designed to improve in vivo imaging.
Lastly, scanning device options were limited when designing the MRFL prototype. A compact scanning mirror device (Optotune MR-15-30) has since become commercially available. The MR-15-30 is significantly smaller and lighter than the T-OMG used in the MRFL and achieves the required 2D scanning speeds, angle ranges, accuracy, and repeatability. As shown in Fig. 4, the path splitting and steering section of the HT-MRFL was redesigned from that of the MRFL to accommodate the MR-15-30. Rather than image an intermediate pupil onto a scanning plate beamsplitter, as was done in the MRFL prototype (Fig. 1), the HT-MRFL eyepiece images an intermediate pupil past a cube beamsplitter and onto the MR-15-30, which acts as a folding and scanning mirror. In this setup, the splitting and steering functions are decoupled without losing utility. The distribution of light to the two arms of the system is controlled by interchanging beamsplitters with different splitting ratios, and the subfield imaged by the ZVP is controlled by adjusting the pitch and yaw of the compact MR-15-30. Accounting only for the optical components and scanning devices, the redesigned path splitting and steering section reduces the overall system weight by 304.4 g, which is just over 20% of the MRFL’s original 1,497 g weight. Actual weight savings will likely be significantly higher since the MR-15-30 does not require a large metal mounting frame like the T-OMG.
3. HT-MRFL optical design and prototype
The final HT-MRFL lens layout is shown in Fig. 4. To meet endoscope packaging and performance requirements, the objective and relay designs feature all custom optics. To meet budget constraints, the remainder of the design, including eyepiece, wide-view probe, and zoom-view probe, prioritizes the use of stock optical components (as shown in Fig. 5).
The design specifications of the MRFL and HT-MRFL are compared in Table 1. The HT-MRFL objective is optimized for a nominal WD of 150 mm with an object space NAobj of 0.005, which is nearly twice of the MRFL NA. It is designed to image at a spatial resolution in object space (SRO) of about 5 lps/mm and 10-15 lps/mm (depending on zoom factor) for the wide and zoomed views, respectively. The overall throughputs of the WVP and ZVP in the HT-MRFL are twice as high as those of the MRFL prototype. To ensure the system is compatible with standard surgical trocars, the outer diameters of the objective and relay lens components are constrained such that a custom 12 mm endoscope tubing with integrated fiber optic illumination jacket can be used to house the optics. The first order parameters for most of the subsystems were determined via standard geometrical optics, throughput conservation, and detector considerations. The objective lens and the zoom-view probe, however, are the most highly constrained and complex subsystems in the HT-MRFL. Therefore, algorithms were written and executed via MATLAB that locate first-order solution regions and enable informed design parameter designation for these subsystems.
Table 1. Design specification comparison between the MRFL and the HT-MRFL.
In addition to setting the maximum achievable image quality for the full system, the objective lens places limitations through conservation of throughput on a multitude of system parameters that each pose physical or application-based concerns or limitations. Each concern or limitation was assigned a governing parameter for which an acceptable value range was defined. Functional forms dependent on NAobj were derived for each governing parameter, including objective focal length, using geometrical optics and conservation of throughput. The first order objective design algorithm populates a densely sampled set of NAobj values and runs it through the functional form of each governing parameter. The resulting parameter values are tested against their respective acceptable ranges, and the NAobj set is culled of values resulting in unacceptable outputs. The final set of NAobj values determines the valid first order solution region for the objective lens, which is plotted in Fig. 6(a). The resulting solution region exists over a very small range of NAobj values that likely would not have been found without the creation and implementation of an algorithm like the one described above.
Fig. 6. Plots of the first order parameter solution regions for the (a) objective lens and (b) zoom-view probe designs.
The ZVP is set up as a Keplerian zoom system followed by an imaging lens. The two lens groups in the Keplerian zoom system each feature an ETL and offset lenses that shift and scale the ETL focal range. Each one of these elements has specific acceptable parameter ranges limited by their physical capabilities or the design goals of the system. First order design of the ZVP amounts to finding a balanced solution that achieves the targeted 3 mm image height (determined by the detector) and 2X-3X zoom range while ensuring that the parameters for each component do not wander outside of their respective acceptable ranges. The first order ZVP design algorithm populates sets of possible values for the spacing between the Keplerian lens groups, tz, and the imaging lens focal length, fI. Minimum ETL focal length values for the ZVP design are chosen to be near the lower limit of the Optotune 10-30-TC focal length range with a small buffer to ensure focal stability. All remaining first order design parameters are strategically considered and evaluated at either 2X or 3X zoom for all possible (tz, fI) pairings. The resulting component parameter values for each (tz, fI) pairing are tested against all associated constraints. Input pairings for which all component parameters fall within the constrained limits are considered valid and are plotted to form the solution region in Fig. 6(b).
It should be noted that the FFOV values listed in Table 1 are noticeably different between the two systems. In designing the MRFL and HT-MRFL for clinical use and acceptance, the ZVP FOV is constrained to image a similarly sized area to that imaged by a traditional laparoscope. The zoom factor at which this constraint is met is, however, is assignable. In balancing the design parameters of the HT-MRFL it was advantageous to meet the FOV constraint at a different zoom factor than the MRFL, resulting in different FOV parameters. EPD was adjusted accordingly to yield the desired throughput increase.
The HT-MRFL optics were designed and optimized in Zemax. Figure 7 and Fig. 8 plot the nominal MTF performance comparison between the HT-MRFL (plotted in green lines) and the MRFL (plotted in orange lines) for the WVP and ZVP, respectively. Three field positions, corresponding to normalized field heights of H = 0, H = 0.5, and H = 0.75, are evaluated for the WVP. Similarly, two field positions corresponding to normalized field heights of H = 0 and H = 0.5 are evaluated for the ZVP at two ROIs, one centered in the WVP FOV and one located near the edge. Given the applications of the system, the full-field (H = 1) curves do not deviate significantly from those plotted and were therefore omitted to ensure plot legibility. The plots clearly show that the nominal MTF values for both the WVP and ZVP of the HT-MRFL are on average 30% higher than those of the MRFL corresponding to the same spatial frequencies and same field positions.
Fig. 7. Plot comparing the designed nominal wide-view MTFs of the MRFL (orange) and HT-MRFL (green).
Fig. 8. Plots comparing the designed nominal zoomed-view MTFs of the MRFL (orange) and HT-MRFL (green) zoom-view probes for central (left) and edge (right) ROI scanning positions. MTFs are plotted for normalized field heights of H=0 and H=0.5.
The HT-MRFL optics are toleranced in CodeV to meet practical performance requirements and assume precision machined opto-mechanical tolerances similar to those of the MRFL prototype. The custom optics were fabricated according to the performance tolerance requirements. Considering the current budget and time constraints, however, we were unable to fabricate custom optical-mechanical housing using precision machinery. Instead, a preliminary benchtop assembly was built for testing using custom 3D-printed parts and stock opto-mechanics (as shown in Fig. 9). Housings, spacers, mounts and alignment aids were designed in SolidWorks and printed in-house on a 3D Systems ProJet MJP 2500 Plus printer. Larger degrees of performance degradation from the nominal performances shown in Figs. 7 and 8 are thus anticipated.
4. Performance evaluation
HT-MRFL performance is evaluated on three fronts: SRO; MTF; and simulated clinical imaging. SRO was measured for the MRFL and HT-MRFL using a negative mirrored 1951 USAF resolution target placed at the nominal WDs of 120 and 150 mm, respectively. Lighting conditions were kept constant between the systems, but detector settings were varied to achieve best performance. Wide-view SROs were measured at normalized field heights of H = 0, 0.5, and 0.75 and are listed in Table 2. Zoom-view SROs were measured at normalized field heights of H = 0 and 0.5 for central and edge ROI scanning positions. The results are compiled in Table 3. To account for rotational asymmetry at H = 0.5, four measurements taken at differing radial angles are averaged.
Table 2. Comparison of measured spatial resolutions (lp/mm) of the MRFL and HT-MRFL wide-view imaging probes at normalized field heights of 0.0, 0.5, and 0.75.
Table 3. Resolution comparison of MRFL and HT-MRFL Zoom probes at 2X and 3X zooms for central and edge ROIs. Resolution measurements were taken at normalized field heights of 0.0 and 0.5.
As seen in Table 2, the HT-MRFL wide-view probe SRO is roughly double that of the MRFL prototype across all fields and falls just shy (on average) of the 5 lp/mm design goal. Table 3 shows consistent and considerable improvement in horizontal SRO from the MRFL to the HT-MRFL across all fields, zooms, and scanning locations. HT-MRFL vertical SROs are, on average, higher than those of the MRFL but the results are less consistent. Such inconsistency is likely due to a combination of alignment error, ghost images, and gravitational effects on the liquid tunable lenses. Furthermore, the HT-MRFL benchtop prototype falls short of the 10-15 lp/mm zoomed-view design goal. As mentioned previously, the HT-MRFL was originally designed for assembly as a practical prototype with high-precision tolerances but is assembled as a benchtop system with 3D printed tolerances due to budget and time constraints. SRO is likely to improve and meet all resolution goals in future stages, however, when budget permits precision fabrication and assembly of a practical prototype.
Slanted edge MTF testing was used to characterize the tangential and sagittal MTF of the benchtop prototype’s wide and zoomed views. Measurements were taken at normalized field heights H = 0, 0.5, and 0.75 for the wide view and H = 0 and 0.5 for the zoomed view. As a comparative metric, as-built (97.7% probability) MTF performance of the HT-MRFL was simulated in CodeV for the precision-machined practical prototype and the 3D printed benchtop assembly. Each system is simulated with tolerances and compensators unique to its fabrication and assembly. Figure 10 plots the average (of tangential and sagittal) 2X zoom ZVP MTF curves of the nominal design, simulated practical prototype, simulated benchtop prototype, and measured benchtop prototype at central and edge ROI scanning positions. As seen in Fig. 10, the measured and simulated benchtop prototype MTF curves correlate well in the mid frequency range with minor disparities in the lower and upper frequency ranges. The measured-on axis MTF of the edge ROI takes exception to this trend and far surpasses simulated expectations in the mid frequency range. As anticipated, the measured MTF values for both the center and edge ROIs were noticeably lower than those of the nominal design and the simulated practical prototype with high-precision mechanics. The same comparative analyses were performed for the WVP and ZVP at 3X zoom. In both cases the measured performance was similar to that predicted in the benchtop prototype simulation. The correlation between simulated and measured performance confirms design validity and suggests that a similar trend may be observed if a practical prototype is constructed with precision machined parts in the future.
Fig. 10. HT-MRFL 2X zoomed-view MTF plots comparing the measured MTF of the benchtop prototype with the MTFs of the nominal design, simulated practical prototype with precision machined parts, and simulated benchtop prototype with 3D printed parts. Curves are presented for the central ROI (top) and edge ROI (bottom) scanning positions at normalized field heights H=0 and 0.5.
For comparison, slanted edge MTF testing was performed for the MRFL prototype with the same target and lighting conditions. Relative field heights and scanning positions were kept the same and the target was placed at nominal WD. Results indicate that the HT-MRFL WVP outperforms that of the MRFL at all measured field heights for approximately 95% of unaliased frequencies, with MTF disparities of greater than 0.6 in the mid frequency range. At 2X zoom, the HT-MRFL significantly outperforms the MRFL by MTF margins up to 0.5 at low and mid frequencies and performs on par with the MRFL at higher frequencies. At 3X zoom the two systems exhibit similar MTF curves with the HT-MRFL showing a slight advantage (average MTF disparity <0.2) in the mid-frequency range.
Pork loin was imaged as an analogue to live-tissue to assess relative clinical utility. The same section of pork loin was imaged using the MRFL and HT-MRFL. For each system, the pork loin was placed at nominal working distance and detector gain and white balance were optimized. The HT-MRFL benchtop prototype does not accommodate fiber illumination, thus fluorescent desk lamps were used to illuminate the pork loin. The HT-MRFL zoom-view images were captured using a single fluorescent desk lamp. The MRFL required the addition of a second identical lamp to yield the comparison images shown in Fig. 11.
Fig. 11. Wide-view (top row), 2X zoom (middle row), and 3X zoom (bottom row) images of pork loin captured with the MRFL prototype (left column) and HT-MRFL benchtop prototype (right column). The green and blue boxes in the wide-view images indicate the ROI position and FOV of the corresponding 2X and 3X images, respectively. The HT-MRFL images were captured using a single light source. The MRFL required an additional identical light source.
As seen in Fig. 11, the HT-MRFL images tissue with superior color and detail as compared to the MRFL which required twice the illumination. Thus, the HT-MRFL’s high-throughput design and medical grade ZVP detector deliver the respective intended benefits detailed in section 2.2 and increase overall clinical utility.
5. Conclusion
A new generation of multi-resolution foveated laparoscope, the HT-MRFL, is introduced to address the clinical shortcomings of the earlier MRFL prototype. The HT-MRFL design features an optical window to ease moisture proofing, a larger working distance to better facilitate abdominal MIS, a high throughput design to account for dual imaging of mismatched and varying FOVs, a medical grade ZVP detector to increase color accuracy and lowlight performance, and a cutting-edge compact scanning mirror unit to reduce size and weight. Current budget constraints prevent the construction of a practical prototype with precision machined tolerances, but a benchtop prototype was assembled with 3D printed opto-mechanics. Slanted edge MTF testing verifies the benchtop HT-MRFL performs as expected and outperforms the MRFL prototype. Resolution testing further supports these findings, and simulated clinical imaging of a live-tissue analogue highlights obvious improvements in color, image quality, and lowlight performance. In conclusion, benchtop testing validates the HT-MRFL design and warrants construction of practical prototype pending funding.
Funding
National Institute of Biomedical Imaging and Bioengineering (1R01EB18921).
Acknowledgments
We thank Milo Obradovic for supplying the Toshiba camera head and helping source the camera controller unit.
Disclosures
Dr. Hong Hua has issued patents for Multi-Resolution Foveated Endoscope/Laparoscope and for Optical Article and Illumination System for Endoscope and has no other related conflicts of interest or financial ties to disclose. Mr. Jeremy Katz has no conflicts of interest or financial ties to disclose.
Data availability
Data underlying the results presented in this paper not publicly available at this time, but may be obtained from the authors upon reasonable request.
References
1. C. Tsui, R. Klein, and M. Garabrant, “Minimally invasive surgery: National trends in adoption and future directions for hospital strategy,” Surg Endosc 27(7), 2253–2257 (2013). [CrossRef]
2. A. Cuschieri, “Minimal access surgery and the future of interventional laparoscopy,” The American Journal of Surgery 161(3), 404–407 (1991). [CrossRef]
3. J. Heemskerk, R. Zandbergen, J. G. Maessen, J. W. M. Greve, and N. D. Bouvy, “Advantages of advanced laparoscopic systems,” Surg Endosc 20(5), 730–733 (2006). [CrossRef]
4. M. P. Wu, C. S. Ou, S. L. Chen, E. Y. T. Yen, and R. Rowbotham, “Complications and recommended practices for electrosurgery in laparoscopy,” Am. J. Surg. 179(1), 67–73 (2000). [CrossRef]
5. C. R. Voyles, J. Mississippi, R. D. Tucker, and I. City, “Education and Engineering Solutions for Potential Problems With Laparoscopic Monopolar Electrosurgery,” The American Journal of Surgery 164(1), 57–62 (1992). [CrossRef]
6. Y. Qin and H. Hua, “Continuously zoom imaging probe for the multi-resolution foveated laparoscope,” Biomed. Opt. Express 7(4), 1175 (2016). [CrossRef]
7. Y. Qin and H. Hua, “Optical design and system engineering of a multiresolution foveated laparoscope,” Appl. Opt. 55(11), 3058–3068 (2016). [CrossRef]
8. Y. Qin, Z. Zheng, and H. Hua, “Multi-resolution foveated laparoscope,” Frontiers in Optics, FIO 2012, 5–6 (2013).
9. Y. Qin, H. Hua, and M. Nguyen, “Multiresolution foveated laparoscope with high resolvability,” Opt. Lett. 38(13), 2191 (2013). [CrossRef]
10. Y. Qin, H. Hua, and M. Nguyen, “Characterization and in-vivo evaluation of a multi-resolution foveated laparoscope for minimally invasive surgery,” Biomed. Opt. Express 5(8), 2548 (2014). [CrossRef]
11. J. I. Katz, S. Lee, and H. Hua, “Improved multi-resolution foveated laparoscope with real-time digital transverse chromatic correction,” Appl. Opt. 59(22), G79 (2020). [CrossRef]
12. M. Nguyen, J. Nabhani, A. Hamilton, S. Lee, J. Katz, K. Samakar, A. Ghlandian, and H. Hua, “Surgeon Assessment of a Novel Multi-Resolution Foveated Laparoscope,” Surgical Innovation 155335062210811 (2022).
