3D Vision System Research Articles

Research and Development of a Real-time UAV Flight Visualization Simulation System

On the premise of practical application, this paper describes the principle, technology and implementation methods of a real-time UAV flight visualization simulation system. The overall framework, hardware system composition, UAV flight model establishment, determination of flight control mode and control rate, design of 3D visual simulation system, management and processing of UAV flight state data are discussed, and finally, the actual application effect of the system is evaluated.

Combining robot-assisted surgical system and 3D visualization system for teaching minimally invasive vitreoretinal surgery.

To investigate the feasibility of teaching minimally invasive vitreoretinal surgery with a robot-assisted surgical system and a three-dimensional (3D) visualization system. Enucleated porcine eyes were established as an animal model for removing foreign bodies. Forty medical students were recruited to remove foreign bodies to compare the traditional microscope and the 3D system. One junior resident performed the surgical task with manual and robot-assisted operations on 20 porcine eyes for each group. One senior surgeon evaluated the retinal invasion by a graded injury degree. The learning curve for minimally invasive vitreoretinal surgery was described. Compared with the robot-assisted group, the injury degree was higher in the manual group. For the first ten surgeries, the manual and robot-assisted groups had injuries of 2.60±1.35 (4 to 0) and 1.80±1.62 (4 to 0), respectively. For the last ten surgeries, the injury degrees were 1.90±1.20 (3 to 0) and 0.80±0.42 (1 to 0). Considering the manual and robot-assisted groups together, 95%, 75% and 60% of the students considered surgical manipulation with the 3D visualization system to be more comfortable, easier and clearer, respectively. The robot-assisted surgical system and 3D visualization system may have value in teaching minimally invasive vitreoretinal surgery.

Automatic assessment of dairy cows' rumen function over time and links to feed changes and milk production

A 3-dimensional (3D) vision-based system was previously designed to automatically estimate the rumen motility of individual cows. This longitudinal study aimed to explore the associations between 3D vision-based rumen function assessment and dairy cow feed changes and milk production on a commercial farm. The 3D vision system was attached to an automatic milking robot to estimate the ruminal contraction frequency and rumen fill in 42 lactating cows during each milking event for 66 d. Additionally, we collected data on milk production, milk composition, general health, and changes in feeding practices. The 3D vision system showed that half the cows displayed a drastic decrease in the estimated rumen fill when all cows began grazing. The grazing and decreased rumen fill were also associated with herd-level milk fat depression. Over the 66 d, one cow was detected with reduced milk production and suspected rumen dysfunction by the farmer. The 3D vision system, however, identified this cow as having sudden decreases in estimated ruminal contraction frequency and rumen fill 4 d before detection by the farmer. In this longitudinal study, the 3D vision-based rumen function assessment system showed potential as a useful management-supporting tool for dairy farmers. The system, however, requires further validation with more cows of various breeds and ages. We suggest validating the 3D vision system with rumen boluses, quantified adjustments in feeding practices, more cases with ruminal dysfunction, and systematic health assessments for future studies.

Bin-Picking Solution for Randomly Placed Automotive Connectors Based on Machine Learning Techniques

This paper presents the development of a bin-picking solution based on low-cost vision systems for the manipulation of automotive electrical connectors using machine learning techniques. The automotive sector has always been in a state of constant growth and change, which also implies constant challenges in the wire harnesses sector, and the emerging growth of electric cars is proof of this and represents a challenge for the industry. Traditionally, this sector is based on strong human work manufacturing and the need arises to make the digital transition, supported in the context of Industry 4.0, allowing the automation of processes and freeing operators for other activities with more added value. Depending on the car model and its feature packs, a connector can interface with a different number of wires, but the connector holes are the same. Holes not connected with wires need to be sealed, mainly to guarantee the tightness of the cable. Seals are inserted manually or, more recently, through robotic stations. Due to the huge variety of references and connector configurations, layout errors sometimes occur during seal insertion due to changed references or problems with the seal insertion machine. Consequently, faulty connectors are dumped into boxes, piling up different types of references. These connectors are not trash and need to be reused. This article proposes a bin-picking solution for classification, selection and separation, using a two-finger gripper, of these connectors for reuse in a new operation of removal and insertion of seals. Connectors are identified through a 3D vision system, consisting of an Intel RealSense camera for object depth information and the YOLOv5 algorithm for object classification. The advantage of this approach over other solutions is the ability to accurately detect and grasp small objects through a low-cost 3D camera even when the image resolution is low, benefiting from the power of machine learning algorithms.

Towards real-time 3D visualization with multiview RGB camera array.

A real-time 3D visualization (RT3DV) system using a multiview RGB camera array is presented. RT3DV can process multiple synchronized video streams to produce a stereo video of a dynamic scene from a chosen view angle. Its design objective is to facilitate 3D visualization at the video frame rate with good viewing quality. To facilitate 3D vision, RT3DV estimates and updates a surface mesh model formed directly from a set of sparse key points. The 3D coordinates of these key points are estimated from matching 2D key points across multiview video streams with the aid of epipolar geometry and trifocal tensor. To capture the scene dynamics, 2D key points in individual video streams are tracked between successive frames. We implemented a proof of concept RT3DV system tasked to process five synchronous video streams acquired by an RGB camera array. It achieves a processing speed of 44 milliseconds per frame and a peak signal to noise ratio (PSNR) of 15.9 dB from a viewpoint coinciding with a reference view. As a comparison, an image-based MVS algorithm utilizing a dense point cloud model and frame by frame feature detection and matching will require 7 seconds to render a frame and yield a reference view PSNR of 16.3 dB.

Evaluating robot bin-picking performance based on Box and Blocks Test

The rise of e-commerce, which demands solutions for small order sizes, large product assortment, short delivery times, and variable order quantity has created a need for more advanced warehousing and order-picking systems. The advancements in collaborative robotics have made it possible to automate intralogistics processes, especially the order-picking systems. If the items to be order-picked are stored in small containers, called bins, the process is known as bin-picking. The processes of bin-picking are still mainly manual due to the adaptability, dexterity, and pace of human hands. However, with the emerging technologies, the gap between human and robot dexterity is getting thinner. The integrators of robotic technologies are therefore faced with the challenge of choosing an appropriate robotic system for bin-picking. There is a lack of standardized procedures, which are designed to process such decisions easier. In our research paper, we propose an improved robotic version of the Box and Blocks Test (BBT), which provides a quick evaluation of a complete robotic bin-picking system. Using our improved robot-adapted BBT protocol, we evaluate a sample robotic bin-picking system, consisting of a collaborative robot, a 3D vision system and three types of robotic grippers, comparing the bin-picking performance of each configuration.

Commentary: Digitally assisted vitreoretinal surgery - Redefining vitreo-retina fellowship training.

Gone are the days when vitreoretinal fellows used to stand in a queue to get a glimpse of the complex procedures in the side scope of an analog microscope. This study aims at finding out the relevance of the DAVS platform used in the NGENUITY 3D Visualization System (Alcon, Forth Worth, Texas, USA) along with TrueVision Visualization System (Santa Barbara, California, USA) in surgical steps visualization compared to the conventional analog microscope (CAM). The 3D DAVS visualization system offers a ‘‘heads-up’’ position for the operating surgeon. This system also has high image definition at wider magnifications and can be used at much lower endo-illumination levels compared to the CAM. Moreover, DAVS allows a “surgeon’s view” to the students, scrub nurse/fellow, and anesthesiologist present in the room. Additionally, it allows the scrub nurse/fellow to be in sync with the surgeon, who can anticipate the next steps during the procedure. As a result, this platform can be made an integral component in fellow and resident education. In this study of IJO, the authors have concluded that the DAVS system has an advantage to the observers in terms of comfort, image quality, understanding of surgery, and viewing preference.[1] Few additional factors need to be considered while interpreting the results of this study. Here, each study participant watched a minimum of five surgeries on each of the two platforms comprising three categories of vitreoretinal procedures. It would have been interesting to see the response of fellows or students actively assisting the surgeon during vitreoretinal surgical procedures, such as during scleral or belt buckling, holding contact lenses, or during scleral depression. During vitreoretinal surgery, in most cases, the assistant sits on the right or left side of the patient, facing the patient sidewise. In the DAVS platform, they have to turn to the right or left to see the screen. This would inevitably increase their fatigue and discomfort, jeopardizing active assistance during those surgeries. There are reports where it was depicted that in sufficiently long surgeries, the surgeon must move to relax his/her neck. The television display is always slightly lateralized, causing a negative impact ergonomically, resulting in torticollis. Furthermore, it was noted that people (other than the surgeon) who were not being able to view the display on the same axis had complaints of dizziness or nausea.[2] 3D asthenopia, such as dizziness, after viewing 3D displays is not very uncommon. Wee et al.[1] found that accommodation and binocular vergence may also influence 3D asthenopia significantly.[3] DAVS platform has an initial high expense as currently both the surgical microscope and the 3D technology need to be purchased separately instead of only the operating microscope, which is being used in CAM. There are reports of future developments that aim to have an integrated, stand-alone DAVS system. On the downside, the 3D system has a larger footprint compared to an operating microscope alone; thus, there needs to be significant physical space in the operating room for the anesthesiologists and nurses to maneuver the anesthesia apparatus and vitrectomy machine around the display monitor in a non-obstructing manner.[4] To get a solution, Yoshihiro et al.[5] suggested a microscope-independent camera that can slide over the patient; if developed, it would cause fewer concerns related to positioning and visual obstructions. This would also help in achieving a smaller footprint in the operating room. There is no doubt that the modern vitreoretinal surgical instruments and microscopes have made a paradigm shift in understanding and adopting the nuances of surgical techniques both for the surgeons as well as the trainees. DAVS platform has undeniably changed the perspective of viewing the operating field, and the authors in this article have made their very valid comparisons. More number of participants, a more detailed questionnaire, and varied surgeries with a longer time frame would have made this study’s results more intriguing. This system, though having its own certain limitations, is definitely going to change the way of ophthalmic surgical training in the coming days.

Artificial vision system for the automatic inspection of preforms

The preforming consists of processing the reinforcement to obtain a geometry of the dry fiber, close to that of the final product before being impregnated by the resin during the injection process. During the process, the fibers are manipulated to reach the desired shape are subjected to tensile, shear and torsion stresees that can cause changes in the orientation of the fibers, misalignments,or other defects such as wrinkles or fiber breakage that can affect the performance of the preform . In this work a solution for the automation of the quality control of the preforming process by artificial vision is investigated.. Two approaches are proposed, 3D vision for the geometric and superficial inspection of the fibers and machine learning aproach based on Deep Learning (DL) to detect irregularities in the orientation. The solution based on 3D vision automatically detects defects based on the comparison of a theoretical 3D model and a 3D reconstruction of the real part. A structured light system is used to generate dense and precise point clouds. Once aligned, the dissimilarities between both surfaces are analyzed. The solution is complemented by a 2D vision system based on DL that classifies the irregularities in the orientations of the fabric. The model is trained with images of classified fiber orientations to automatically identify areas of fiber whose orientation does not correspond to the expected one. The combination of both technologies allows to give a complete solution to the automated quality inspection of preforms for manufacturing or defects.

Center of Gravity Coordinates Estimation Based on an Overall Brightness Average Determined from the 3D Vision System

In advanced manufacturing technologies (including complex automated processes) and their branches of industry, perception and evaluation of the object parameters are the most critical factors. Many production machines and workplaces are currently equipped as standard with high-quality special sensing devices based on vision systems to detect these parameters. This article focuses on designing a reachable and fully functional vision system based on two standard CCD cameras usage, while the emphasis is on the RS 232C communication interface between two sites (vision and robotic systems). To this, we combine principles of the 1D photogrammetric calibration method from two known points at a stable point field and the available packages inside the processing unit of the vision system (as filtering, enhancing and extracting edges, weak and robust smoothing, etc.). A correlation factor at camera system (for reliable recognition of the sensed object) was set from 84 to 100%. Then, the pilot communication between both systems was proposed and then tested through CREAD/CWRITE commands according to protocol 3964R (used for the data transfer). Moreover, the system was proven by successful transition of the data into the robotic system. Since research gaps in this field still exist and many vision systems are based on PC processing or intelligent cameras, our potential research topic tries to provide the price–performance ratio solution for those who cannot regularly invest in the newest vision technology; however, they could still do so to stay competitive.

Evaluation of Advanced Bimanual Skills in Novices Using the Wrist-Like FlexDex™ Articulating Laparoscopic Needle Holder in 2D and 3D Vision: A Randomised Trial

AimThe aim of this study was to evaluate the novice performance of advanced bimanual laparoscopic skills using the articulating FlexDexTM laparoscopic needle holder in two-dimensional (2D) and three-dimensional (3D) visual systems.MethodsIn this prospective randomised trial, novices (n=40) without laparoscopic experience were recruited from a university cohort and randomised into two groups, which used the FlexDexTM and 2D or the FlexDex™ and 3D. Both groups performed 10 repetitions of a validated assessment task. Times taken and error rates were measured, and assessments were made based on completion times, error rates and learning curves.ResultsThe intervention group that used FlexDexTM and 3D visual output completed 10 attempts of the standardised laparoscopic task quicker than the control group that used FlexDexTM with standard 2D visual output (268 seconds vs 415 seconds taken for the first three attempts and 176 seconds vs 283 seconds taken for the last three attempts, respectively). Moreover, each attempt was completed faster by the intervention group compared to the control group. The difference in average time for the first three and last three attempts reached statistical significance (P < 0.001).ConclusionCombination of 3D visual systems and the FlexDexTM laparoscopic needle holder resulted in superior task performance speed, leading to shorter completion times and quicker learning effect. Although the 3D group demonstrated lower mean error rates, it did not reach statistical significance.Key Statement3D visual systems lead to faster task completion times when combined with an articulating laparoscopic needle holder compared to 2D vision. This effect however is not seen in error rates.

Comparison of heads up three dimensional visualization system to conventional microscope in retinopathy of prematurity related tractional retinal detachment

To report the outcomes, advantages and disadvantages of a heads-up three-dimensional (3D) visualization system compared to the conventional microscope in pediatric tractional retinal detachment (TRD) surgery secondary to advanced stage retinopathy of prematurity (ROP). Medical records of patients with ROP stage 4 or 5 who underwent surgery for tractional retinal detachment at King Khaled Eye Specialist Hospital between September 2017 and July 2019 were identified and reviewed. Eyes were divided into 2 groups, eyes that underwent surgery with a 3D heads-up platform (3D group) and eyes that underwent surgery with a conventional microscope (conventional group). Data were collected on neonatal history, visual acuity, intraoperative complications and success rates between groups.Eighteen eyes of 14 patients who underwent surgical repair of TRD related to ROP. Postoperative outcomes were compared between 10 eyes (7 patients) in the 3D group and 8 eyes (7 patients) in the conventional group There was no statistically significant difference in success rate between both groups (75% conventional group vs 70% 3D group). Partial or complete reattachment was achieved in 7 eyes in 3D group compared to 6 eyes in conventional group. Lower postmenstrual age at the time of the first surgery and presence of retinal breaks were associated with poorer surgical outcome. Heads up 3D visualization system is feasible in tractional retinal detachment related to ROP with similar success rate and no increased risk of complications when compared to conventional microscope. This system may be advantageous in advanced pediatric tractional retinal detachment surgeries.

Gender benefit in laparoscopic surgical performance using a 3D-display system: data from a randomized cross-over trial

BackgroundThe use of 3D technique compared to high-resolution 2D-4K-display technique has been shown to optimize spatial orientation and surgical performance in laparoscopic surgery. Since women make up an increasing amount of medical students and surgeons, this study was designed to investigate whether one gender has a greater benefit from using a 3D compared to a 4K-display system.MethodsIn a randomized cross-over trial, the surgical performance of male and female medical students (MS), non-board certified surgeons (NBCS), and board certified surgeons (BCS) was compared using 3D- vs. 4K-display technique at a minimally invasive training parkour with multiple surgical tasks and repetitions.Results128 participants (56 women, 72 men) were included. Overall parkour time in seconds was 3D vs. 4K for all women 770.7 ± 31.9 vs. 1068.1 ± 50.0 (p < 0.001) and all men 664.5 ± 19.9 vs. 889.7 ± 31.2 (p < 0.001). Regarding overall mistakes, participants tend to commit less mistakes while using the 3D-vision system, showing 10.2 ± 1.1 vs. 13.3 ± 1.3 (p = 0.005) for all women and 9.6 ± 0.7 vs. 12.2 ± 1.0 (p = 0.001) for all men. The benefit of using a 3D system, measured by the difference in seconds, was for women 297.3 ± 41.8 (27.84%) vs. 225.2 ± 23.3 (25.31%) for men (p = 0.005). This can be confirmed in the MS group with 327.6 ± 65.5 (35.82%) vs. 249.8 ± 33.7 (32.12%), p = 0.041 and in the NBCS group 359 ± 52.4 (28.25%) vs. 198.2 ± 54.2 (18.62%), p = 0.003. There was no significant difference in the BCS group.Conclusion3D laparoscopic display technique optimizes surgical performance compared to the 2D-4K technique for both women and men. The greatest 3D benefit was found for women with less surgical experience. As a possible result of surgical education, this gender specific difference disappears with higher grade of experience. Using a 3D-vision system could facilitate surgical apprenticeship, especially for women.

Early corneal pachymetry maps after cataract surgery and influence of 3D digital visualization system in minimizing corneal oedema.

To describe the early topography of corneal swelling occurring after cataract surgery and to evaluate the impact of the three-dimensional (3D) digital visualization system in minimizing corneal oedema. Prospective observational, single-centre, consecutive case series of 134 patients undergoing cataract surgery performed by the same surgeon, with either 3D or conventional visualization systems. Eyes were assigned to two groups based on their anterior chamber depth (group ACD ≤3 mm and group ACD >3 mm). Optical coherence tomography was performed to evaluate postoperative corneal swelling. Three corneal swelling profiles were identified on the first postoperative day type 1, limited corneal oedema near peripheral corneal incisions; type 2, dome-shaped corneal swelling spreading from the principal corneal incision and reaching the paracentral cornea; type 3, continuous oedema spreading from the principal incision to central cornea, with a generalized oedema predominating in the upper part of the cornea. On the first day after surgery, in group ACD ≤3 mm, visual acuity was significantly better in patients undergoing surgery with 3D visualization (0.023 vs 0.072 logMar, p = 0.014) with reduced central corneal thickening 17.3 µm (±3.2) in comparison with conventional visualization 44.0 µm (±9.3) (p = 0.0082). In group ACD >3 mm, no significant association was found between the use of the 3D system and pachymetry changes and early visual rehabilitation. On day 21 after surgery, no significant differences in corneal pachymetry values were observed between the two surgical approaches in both groups. We describe early postoperative corneal map profiles providing insight into the pathogenesis of postoperative corneal swelling and possible prevention strategies. By improving visualization of the narrow surgical space in patients with shallow anterior chambers, the 3D system could help to minimize postoperative corneal oedema.

Research and application of 3D GIS in the visualization and information management of power grid construction projects

The research direction and application scope of transmission line 3D visualization aided design system include design data retrieval of transmission line project, preliminary path selection, feasibility study, and preliminary design. Actively explore the in-depth application of line 3D construction drawing design and digital transfer. Based on the relevant work experience of the author engaged in the application of GIS technology for many years, and taking the application of 3D GIS technology in power grid planning and design as the research object, this paper designs and implements a 3D GIS visualization system for power grid planning and design, decomposes the system function with a case, and analyzes the operation flow of the system in detail, which is believed to be beneficial to colleagues engaged in related work.

Changes in tail posture detected by a 3D machine vision system are associated with injury from damaging behaviours and ill health on commercial pig farms.

To establish whether pig tail posture is affected by injuries and ill health, a machine vision system using 3D cameras to measure tail angle was used. Camera data from 1692 pigs in 41 production batches of 42.4 (±16.6) days in length over 17 months at seven diverse grower/finisher commercial pig farms, was validated by visiting farms every 14(±10) days to score injury and ill health. Linear modelling of tail posture found considerable farm and batch effects. The percentage of tails held low (0°) or mid (1–45°) decreased over time from 54.9% and 23.8% respectively by -0.16 and -0.05%/day, while tails high (45–90°) increased from 21.5% by 0.20%/day. Although 22% of scored pigs had scratched tails, severe tail biting was rare; only 6% had tail wounds and 5% partial tail loss. Adding tail injury to models showed associations with tail posture: overall tail injury, worsening tail injury, and tail loss were associated with more pigs detected with low tail posture and fewer with high tails. Minor tail injuries and tail swelling were also associated with altered tail posture. Unexpectedly, other health and injury scores had a larger effect on tail posture- more low tails were observed when a greater proportion of pigs in a pen were scored with lameness or lesions caused by social aggression. Ear injuries were linked with reduced high tails. These findings are consistent with the idea that low tail posture could be a general indicator of poor welfare. However, effects of flank biting and ocular discharge on tail posture were not consistent with this. Our results show for the first time that perturbations in the normal time trends of tail posture are associated with tail biting and other signs of adverse health/welfare at diverse commercial farms, forming the basis for a decision support system.

3D shape sensing and deep learning-based segmentation of strawberries

Automation and robotisation of the agricultural sector are seen as a viable solution to socio-economic challenges faced by this industry. This technology often relies on intelligent perception systems providing information about crops, plants and the entire environment. The challenges faced by traditional 2D vision systems can be addressed by modern 3D vision systems which enable straightforward localisation of objects, size and shape estimation, or handling of occlusions. So far, the use of 3D sensing was mainly limited to indoor or structured environments. In this paper, we evaluate modern sensing technologies including stereo and time-of-flight cameras for 3D perception of shape in agriculture and study their usability for segmenting out soft fruit from background based on their shape. To that end, we propose a novel 3D deep neural network which exploits the organised nature of information originating from the camera-based 3D sensors. We demonstrate the superior performance and efficiency of the proposed architecture compared to the state-of-the-art 3D networks. Through a simulated study, we also show the potential of the 3D sensing paradigm for object segmentation in agriculture and provide insights and analysis of what shape quality is needed and expected for further analysis of crops. The results of this work should encourage researchers and companies to develop more accurate and robust 3D sensing technologies to assure their wider adoption in practical agricultural applications.

Vision-Based Approach in Contact Modelling between the Footpad of the Lander and the Analogue Representing Surface of Phobos.

Identifying solar system surface properties of celestial bodies requires the conducting of many tests and experiments in conditions similar to those found on various objects. One of the first tasks to be solved by engineers is determining the contact condition between the lander and the surface of a given celestial body during landing in a microgravity environment. This paper presents the results of experimental studies and numerical simulations of the contact phenomenon between the lander foot model and the Phobos analogue. The main goal of the experimental tests was to obtain measured deformation data of the studied analogues using 2D and 3D vision systems, which were employed to analyze the behavior of the lander foot and the surface of the studied analogue itself and to calibrate the numerical models. The analogue representing the Phobos surface was foam concrete. The variable parameters in the study were the analogue thickness and the lander foot velocity at the time of contact. Tests were conducted for three different contact velocities of 1.2 m/s, 3.0 m/s, and 3.5 m/s. Taking into account the mass of the lander foot model, kinetic energies of 30.28 J, 189.22 J, and 257.56 J were obtained. The results showed that at low contact velocities, and thus low kinetic energies, no significant differences in behavior of the material directly under the lander foot were observed, and similar values of forces in the lander foot were obtained. For higher contact velocities, the behavior of analogues with varying thicknesses was different, resulting in different values of analogue deformation and dynamics of increments and decrements of force in the lander foot itself. Although performed on a single material, the experiments revealed different behaviors depending on its thickness at the same impact energy. This is an essential guideline for engineers who need to take this fact into account when designing the lander itself.

Об операционных осложнениях высокотехнологичной факоэмульсификации с имплантацией ИОЛ

Purpose. To analyze the nature, frequency and effectiveness of correction of surgical complications during high-tech phacoemulsification (PE) with IOL implantation. Material and methods. A prospective non-randomized study is devoted to the analysis of surgical complications in patients who underwent outpatient FE surgery with IOL implantation. The results of 3000 consecutive operations (3000 eyes) performed in 2019-2020 at the Eye Surgery clinic were studied. The calculation of the IOL was carried out on the diagnostic system "VERION" (Alcon). All operations were performed in the conditions of 3D visualization using the digital system "NGENUITY® 3D Visualization System" (Alcon). Alcon IOLs were implanted: monofocal, multifocal, toric. Results. During the surgical intervention, 33/3000 (1.1%) surgical complications occurred. The main surgical complication was the rupture of the posterior capsule of the lens-26 (0.86%) of the eyes, mainly without loss of the vitreous body-22 (0.73%). Conclusions. 1. The number of surgical complications during high-tech phacoemulsification with IOL implantation was 1.1%. 2. The main one among them was the rupture of the posterior capsule of the lens-0.86%, mainly without loss of the vitreous body-0.73%. 3. High-tech surgical support of operations and correction of the arisen operational complications allowed to successfully eliminate all complications in all cases. Key words: cataract, phacoemulsification, surgical complications, posterior capsule rupture.

Comparison of novel digital microscope using integrated intraoperative OCT with Ngenuity 3D visualization system in phacoemulsification

ObjectiveTo compare surgical efficiency, visual and physical comfort, and safety profile of the ARTEVO 800 Digital Microscope (Carl Zeiss Meditec AG, Jena, Germany) and the Ngenuity 3D Visualization System (Alcon Laboratories Inc, Fort Worth, TX) in cataract surgery. DesignCross-sectional study. ParticipantsOne hundred consecutive phacoemulsification cataract surgeries performed by five surgeons from June 1, 2020, to November 1, 2020. MethodsFor each case, the surgeons answered a 2-section questionnaire (before and after intervention) to collect data on cataract severity or grade, surgical risk, chosen three-dimensional (3D) visualization system, surgical complications, and the visual or physical discomfort experienced during the procedure. ResultsEach of the 5 surgeons performed 20 surgeries (N = 100) using either the ARTEVO 800 Digital Microscope (N = 50) or the Ngenuity Visualization System (N = 50). Mean duration of the surgical procedure was 17.07 ± 4.80 minutes, and none of the surgeons ever switched to the classical microscope. In addition, 40% of surgeries were considered at low risk, 30% at intermediate risk, and 30% at high risk. The Zoom, Focus, and XY commands were used 1–3 times, respectively, during 76 (p = 0.34), 73 (p = 0.49), and 76 (p = 0.64) interventions. Surgical uncertainty and operative fluency were similar using both systems (p = 0.53 and p = 0.19). We observed 14 intraoperative complications, 9 using Ngenuity and 5 using the ARTEVO 800. Surgeon's visual comfort (p = 0.79), colour or brightness perception (p = 0.82), and visual impairment (p = 0.62) during surgery were similar for both systems. Headache, backache, and other musculoskeletal problems were reported, respectively, after 14 (p = 0.79), 11 (p = 0.99), and 8 (p = 0.44) procedures. ConclusionBoth the Ngenuity 3D Visualization System (Alcon Laboratories Inc) and the ARTEVO 800 Digital Microscope (Carl Zeiss Meditec AG) provided comparable operative speed and overall surgical comfort during cataract surgery.

Assessment of importance-based machine learning feature selection methods for aggregate size distribution measurement in a 3D binocular vision system

Aggregate size is usually measured by manual sampling and sieving. Machine vision techniques can provide fast, non-invasive measurement. However, the traditional imaging method using a single size descriptor to discriminate different sieve-size classes of coarse aggregates might not yield high-precision classification results. To determine the optimum supervised machine learning model for coarse aggregates sieve-size measurement, 17 methods were evaluated and compared. To train our model, a new dataset named MFCA27 (Multiple Features of Coarse Aggregate 27) was introduced, which contains 27 features of aggregates based on aggregate three-dimensional (3D) top-surface object. In addition, a feature selection approach for investigating how accuracy varied with the datasets under different feature sets was developed, where feature selection was performed according to the impurity-based feature importance score measured using an extremely randomized tree model. Experiments demonstrated that the Gaussian process classifier (GPC) was the best-performing method on the datasets with two- or three-dimensional (2D/3D) feature sets in terms of accuracy and robustness. The results also showed that, compared with the traditional aggregate sieve-size measurement method, which is based on a single size descriptor, GPC can achieve an accuracy of 95.06% on the test dataset of MFCA27 in the aggregate sieve-size class measurement task.