Vision Problems Research Articles

The Multitarget Action of Vitamins in the Ischemic Stroke.

A stroke, also known as a cerebral hemorrhage, occurs when there is an interruption in the blood supply to a part of the brain, resulting in damage to brain cells. This issue is one of the leading causes of death in developed countries, currently killing about 5 million people annually. Individuals who survive ischemic stroke often face serious vision problems, paralysis, dementia, and other sequelae. The numerous efforts to prevent and/or treat stroke sequelae seem insufficient, which is concerning given the increasing global elderly population and the well-known association between aging and stroke risk. In this review, we aim to present and discuss the importance of vitamins in stroke prevention and/or incidence. Vitamins from diet or dietary supplements influence the body at various levels; they are a relevant factor but are reported only in isolated articles. This review reports and updates the multitarget role of vitamins involved in reducing stroke risk.

Direct Comparison of Treatment Outcome Between the Botulinum Toxin and Calcitonin Gene-Related Peptide Monoclonal Antibody in Migraine Patients.

Migraine is a genetic disorder characterized by recurrent episodes of headache that are throbbing in nature. The objective of this study was to directly compare the efficacy and safety of anti-calcitonin gene-related peptide (anti-CGRP) and botulinum neurotoxin (BoNT) for the preventive treatment of chronic migraine. This quasi-experimental comparative study was conducted on 80 "chronic migraine patients" at King Fahad University Hospital, Dammam, KSA. Chronic migraineurs were divided into two groups (40 patients/group) and were treated with the standard doses of BoNT (group I) and anti-CGRP (group II). All the patients filled out the migraine pain scale, migraine disability assessment score, headache impact test (HIT-6), and adverse drug event questionnaire before the start and at the end of 9 months of treatment. Most of the patients were females (76.3% vs. 23.8%) and were suffering from migraine for more than 24 months (66%). The mean age of the participants was 39.07 ± 10.01 years. Both BoNT and anti-CGRP groups showed a statistically significant decrease in mean HIT-6 and pain scores after 9 months of intervention. A direct comparison between the two treatment groups showed that the anti-CGRP drug caused a higher decrease in HIT-6 and pain scores as compared to the botulinum drug, but the difference was not statistically significant (P = 0.075 and 0.07, respectively). The most common adverse effect was "headache", reported by 45% and 40% of patients, followed by "pain at the site of injection" reported by 27.5% and 32.5% of BoNT and anti-CGRP groups, respectively. The two groups did not differ significantly in the frequency of adverse effects such as nausea, vomiting, visual problems, etc., except "joint stiffness". A significantly higher number of anti-CGRP patients experienced joint stiffness as compared to the BoNT group (17.5% vs. 0%, P = 0.006). A direct comparison between the two treatments indicated that neither of the two interventions is statistically superior to the other in terms of efficacy and both are equally effective in the management of migraine. However, BoNT can be preferred over anti-CGRP because of its cost-effectiveness.

Vision-Centric BEV Perception: A Survey.

In recent years, vision-centric Bird's Eye View (BEV) perception has garnered significant interest from both industry and academia due to its inherent advantages, such as providing an intuitive representation of the world and being conducive to data fusion. The rapid advancements in deep learning have led to the proposal of numerous methods for addressing vision-centric BEV perception challenges. However, there has been no recent survey encompassing this novel and burgeoning research field. To catalyze future research, this paper presents a comprehensive survey of the latest developments in vision-centric BEV perception and its extensions. It compiles and organizes up-to-date knowledge, offering a systematic review and summary of prevalent algorithms. Additionally, the paper provides in-depth analyses and comparative results on various BEV perception tasks, facilitating the evaluation of future works and sparking new research directions. Furthermore, the paper discusses and shares valuable empirical implementation details to aid in the advancement of related algorithms.

Gaseous Object Detection.

Object detection, a fundamental and challenging problem in computer vision, has experienced rapid development due to the effectiveness of deep learning. The current objects to be detected are mostly rigid solid substances with apparent and distinct visual characteristics. In this paper, we endeavor on a scarcely explored task named Gaseous Object Detection (GOD), which is undertaken to explore whether the object detection techniques can be extended from solid substances to gaseous substances. Nevertheless, the gas exhibits significantly different visual characteristics: 1) saliency deficiency, 2) arbitrary and ever-changing shapes, 3) lack of distinct boundaries. To facilitate the study on this challenging task, we construct a GOD-Video dataset comprising 600 videos (141,017 frames) that cover various attributes with multiple types of gases. A comprehensive benchmark is established based on this dataset, allowing for a rigorous evaluation of frame-level and video-level detectors. Deduced from the Gaussian dispersion model, the physics-inspired Voxel Shift Field (VSF) is designed to model geometric irregularities and ever-changing shapes in potential 3D space. By integrating VSF into Faster RCNN, the VSF RCNN serves as a simple but strong baseline for gaseous object detection. Our work aims to attract further research into this valuable albeit challenging area.

Surface Reconstruction From Point Clouds: A Survey and a Benchmark.

Reconstruction of a continuous surface of two-dimensional manifold from its raw, discrete point cloud observation is a long-standing problem in computer vision and graphics research. The problem is technically ill-posed, and becomes more difficult considering that various sensing imperfections would appear in the point clouds obtained by practical depth scanning. In literature, a rich set of methods has been proposed, and reviews of existing methods are also provided. However, existing reviews are short of thorough investigations on a common benchmark. The present paper aims to review and benchmark existing methods in the new era of deep learning surface reconstruction. To this end, we contribute a large-scale benchmarking dataset consisting of both synthetic and real-scanned data; the benchmark includes object- and scene-level surfaces and takes into account various sensing imperfections that are commonly encountered in practical depth scanning. We conduct thorough empirical studies by comparing existing methods on the constructed benchmark, and pay special attention on robustness of existing methods against various scanning imperfections; we also study how different methods generalize in terms of reconstructing complex surface shapes. Our studies help identity the best conditions under which different methods work, and suggest some empirical findings. For example, while deep learning methods are increasingly popular in the research community, our systematic studies suggest that, surprisingly, a few classical methods perform even better in terms of both robustness and generalization; our studies also suggest that the practical challenges of misalignment of point sets from multi-view scanning, missing of surface points, and point outliers remain unsolved by all the existing surface reconstruction methods. We expect that the benchmark and our studies would be valuable both for practitioners and as a guidance for new innovations in future research.

Learning From Human Attention for Attribute-Assisted Visual Recognition.

With prior knowledge of seen objects, humans have a remarkable ability to recognize novel objects using shared and distinct local attributes. This is significant for the challenging tasks of zero-shot learning (ZSL) and fine-grained visual classification (FGVC), where the discriminative attributes of objects have played an important role. Inspired by human visual attention, neural networks have widely exploited the attention mechanism to learn the locally discriminative attributes for challenging tasks. Though greatly promoted the development of these fields, existing works mainly focus on learning the region embeddings of different attribute features and neglect the importance of discriminative attribute localization. It is also unclear whether the learned attention truly matches the real human attention. To tackle this problem, this paper proposes to employ real human gaze data for visual recognition networks to learn from human attention. Specifically, we design a unified Attribute Attention Network (A 2Net) that learns from human attention for both ZSL and FGVC tasks. The overall model consists of an attribute attention branch and a baseline classification network. On top of the image feature maps provided by the baseline classification network, the attribute attention branch employs attribute prototypes to produce attribute attention maps and attribute features. The attribute attention maps are converted to gaze-like attentions to be aligned with real human gaze attention. To guarantee the effectiveness of attribute feature learning, we further align the extracted attribute features with attribute-defined class embeddings. To facilitate learning from human gaze attention for the visual recognition problems, we design a bird classification game to collect real human gaze data using the CUB dataset via an eye-tracker device. Experiments on ZSL and FGVC tasks without/with real human gaze data validate the benefits and accuracy of our proposed model. This work supports the promising benefits of collecting human gaze datasets and automatic gaze estimation algorithms learning from human attention for high-level computer vision tasks.

Divert More Attention to Vision-Language Object Tracking.

Multimodal vision-language (VL) learning has noticeably pushed the tendency toward generic intelligence owing to emerging large foundation models. However, tracking, as a fundamental vision problem, surprisingly enjoys less bonus from recent flourishing VL learning. We argue that the reasons are two-fold: the lack of large-scale vision-language annotated videos and ineffective vision-language interaction learning of current works. These nuisances motivate us to design more effective vision-language representation for tracking, meanwhile constructing a large database with language annotation for model learning. Particularly, in this paper, we first propose a general attribute annotation strategy to decorate videos in six popular tracking benchmarks, which contributes a large-scale vision-language tracking database with more than 23,000 videos. We then introduce a novel framework to improve tracking by learning a unified-adaptive VL representation, where the cores are the proposed asymmetric architecture search and modality mixer (ModaMixer). To further improve VL representation, we introduce a contrastive loss to align different modalities. To thoroughly evidence the effectiveness of our method, we integrate the proposed framework on three tracking methods with different designs, i.e., the CNN-based SiamCAR (Guo et al. 2020), the Transformer-based OSTrack (Ye et al. 2022), and the hybrid structure TransT (Chen et al. 2021). The experiments demonstrate that our framework can significantly improve all baselines on six benchmarks. Besides empirical results, we theoretically analyze our approach to show its rationality. By revealing the potential of VL representation, we expect the community to divert more attention to VL tracking and hope to open more possibilities for future tracking with diversified multimodal messages.

Advances in Deep Reinforcement Learning for Computer Vision Applications

Deep Reinforcement Learning (DRL) has become very popular for computer vision (CV), solving mostly visually complex environments with decision making and dynamic adaption to different situations. This article provides an introduction to the basic concepts of deep reinforcement learning with a special focus on their applications in computer vision tasks, including challenging problems and emerging solutions. It reviews various DRL algorithms such as Q-learning, policy gradient methods, and Actor-Critic models, explaining much of the modifications that are done to make it work on high-dimensional visual problems. Different applications of DRL in major CV applications like object detection, image segmentation, target tracking and image generation are reviewed to demonstrate the power as well as limitations of DRL in practice. More importantly, the novel paradigms such as hierarchical policy learning, adaptive reward design multi-task reinforcement learning and domain adaptation can be viewed as promising premises to improve model efficiency and generalizability in multiple scenarios. Finally, this paper mentions a few of the existing challenges like computational power cost and sample efficiency; as well as future paths for enhancing that can widen devout reinforcement learning in computer vision. Through this comprehensive overview, we aim to shed light on the promising synergies between DRL and CV, while identifying key areas for future research and application.

SGR-AutoLap: Surgical gesture recognition-based autonomous laparoscope control for human-robot shared control in semi-autonomous minimally invasive surgical robot

Purpose This paper aims to investigate the problem of vision based autonomous laparoscope control, which can serve as the primary function for semi-autonomous minimally invasive surgical robot system. Providing the surgical gesture recognition information is a fundamental key component for enabling intelligent context-aware assistance in autonomous laparoscope control task. While significant advances have been made in recent years, how to effectively carry out the efficient integration of surgical gesture recognition and autonomous laparoscope control algorithms for robotic assisted minimally invasive surgical robot system is still an open and challenging topic. Design/methodology/approach The authors demonstrate a novel surgeon in-loop semi-autonomous robotic-assisted minimally invasive surgery framework by integrating the surgical gesture recognition and autonomous laparoscope control tasks. Specifically, they explore using a transformer-based deep convolutional neural network to effectively recognize the current surgical gesture. Next, they propose an autonomous laparoscope control model to provide optimal field of view which is in line with surgeon intra-operation preferences. Findings The effectiveness of this surgical gesture recognition methodology is demonstrated on the public JIGSAWS and Cholec80 data sets, outperforming the comparable state-of-the-art methods. Furthermore, the authors have validated the effectiveness of the proposed semi-autonomous framework on the developed HUAQUE surgical robot platforms. Originality/value This study demonstrates the feasibility to perform cognitive assistant human–robot shared control for semi-autonomous robotic-assisted minimally invasive surgery, contributing to the reference for further surgical intelligence in computer-assisted intervention systems.

Primary Visual Pathway Changes in Individuals With Chronic Mild Traumatic Brain Injury

Individuals with mild traumatic brain injury (TBI) often report vision problems despite having normal visual acuity and fundus examinations. Diagnostics are needed for these patients. To determine if a battery of assessments or machine-learning approaches can aid in diagnosing visual dysfunction in patients with mild TBI. This prospective, observational, case-control study was conducted between May 2018 and November 2021. The study setting was at a level 1 trauma research hospital. Participant eligibility included adult males and females with recorded best-corrected visual acuity and normal fundus examination. Individuals in the case group had a history of mild TBI; controls had no history of TBI. Exclusion criteria included a history of ocular, neurological, or psychiatric disease, moderate-severe TBI, recent TBI, metal implants, age younger than 18 years, and pregnancy. Cases and controls were sex- and age-matched. Data analysis was performed from July 2023 to March 2024. History of mild TBI in the case group. The single-session visit included the Neurobehavioral Symptom Inventory and measurements of oculomotor function, optical coherence tomography, contrast sensitivity, visual evoked potentials, visual field testing, and magnetic resonance imaging. A total of 28 participants (mean [SD] age, 35.0 [12.8] years; 15 male [53.6%]) with mild TBI and 28 controls (mean [SD] age, 35.8 [8.5] years; 19 female [67.9%]) were analyzed. Participants with mild TBI showed reduced prism convergence test breakpoint (-8.38; 95% CI, -14.14 to -2.62; P = .008) and recovery point (-8.44; 95% CI, -13.82 to -3.06; P = .004). Participants with mild TBI also had decreased contrast sensitivity (-0.07; 95% CI, -0.13 to -0.01; P = .04) and increased visual evoked potential binocular summation index (0.32; 95% CI, 0.02-0.63; P = .02). A subset of participants exhibited reduced peripapillary retinal nerve fiber layer thickness, increased optic nerve/sheath size, and brain cortical volumes. Machine learning identified subtle differences across the primary visual pathway, including the optic radiations and occipital lobe regions, independent of visual symptoms. Results of this case-control study suggest that the visual system was affected in individuals with mild TBI, even in those who did not self-report vision problems. These findings support the utility of a battery of assessments or machine-learning approaches to accurately diagnose this population.

Therapeutic BCG vaccine protects against long COVID: The BATTLE randomized clinical trial.

Bacillus Calmette-Guérin (BCG) injected during the COVID-19 convalescence period was safe and enhanced recovery from anosmia and dysgeusia in the acute phase. To report the long-term results of the BATTLE trial, BCG vaccine in adults with mild COVID-19. Design: Double-blind, placebo-controlled, randomized (1:1) clinical trial. BCG intradermal vaccine and placebo. A total of 157 BCG and 142 placebo recipients participated in the 6-month follow-up, and 97 BCG and 95 placebo recipients participated in the 12-month follow-up. Long COVID symptoms and mechanistic analyses. BCG reduced hearing problems at 6 months (odds ratio [OR]=0.26) and sleeping, concentration, memory, and vision problems at 12 months (OR=0.45, 0.36, 0.38, and 0.36, respectively). Sensitivity analyses confirmed that long COVID-19 symptoms were reduced at the 6- and 12-month follow-ups (p=0.010 and 0.031, respectively). BCG's crossover interaction paradoxically increased hair loss in women and decreased it in men at 6 months (p=0.032). BCG immunomodulation is likely mediated through inhibition of Fas ligand expression in the blood and increased induction of IL6, IL10, interferon-induced transmembrane protein 3, and angiotensin-converting enzyme 2 in cultured human macrophages. Long-term follow-up of the BATTLE trial participants revealed that BCG protects against long COVID development if administered within the COVID-19 convalescence period. The response to BCG was subject-specific, including a paradoxical crossover interaction based on sex. Not tested for previous mycobacterial exposure; loss to follow-up, particularly at 12 months.

Uncertain features exploration in temporal moment localization via language by utilizing customized temporal transformer

Temporal moment localization in videos using natural language (TMLVNL) is challenging problem in computer vision. TMLVNL aims to determine the correct moment in a lengthy, untrimmed video that corresponds to the input query. In addition to its inherent complexity, TMLVNL encounters several additional difficulties that can degrade performance, such as rare object positions, occlusions, camera focus issues, and motion blurriness. To address these issues, this study proposes a novel solution called the Temporal Ziggurat Transformer Network. First, we developed a novel method rather than relying on various combinational approaches. Second, we proposed complicated scenarios, such as unusual object postures, object occlusions, camera focus issues, and motion blurriness, by incorporating specialized blocks into our Customized Ziggurat Transformer (ZT) to thoroughly explore visual features. Third, to facilitate the understanding of visual features associated with query words, we proposed a query word-specific transformer (QWST) as a submodule of ZT. QWST integrates query word feature representations with extensively investigated visual features. Fourth, in our module named STDF, we managed query-sentence representations along with query word attributes to extract semantic context from video chunks. The moment was then localized, with its start and end borders identified using the moment localization module. Comprehensive experiments on the Charades-STA, TACos, and Activity-Netcaption datasets demonstrated that our strategy outperformed existing state-of-the-art methods.

PREVENTION AND TREATMENT OF VISUAL IMPAIRMENT IN CHILDREN ADOLESCENTS AND YOUNG PEOPLE WITH UNCONTROLLED USING A COMPUTER AND ELECTRONIC GADGETS

The prevention of visual impairment in people with uncontrolled use of computers and electronic gadgets, the study of clinical manifestations of their negative effects on the body and, especially, on the organ of vision, is an urgent problem of modern medicine. The purpose of the work was to substantiate the need for the prevention and treatment of visual impairment in adolescents with uncontrolled smartphone use. The article describes the negative effects of irrational use of computer gadgets on users' eyesight, clinical manifestations of computer vision syndrome and its component – dry eye syndrome, and describes in detail approaches to the prevention and treatment of visual impairment associated with uncontrolled use of computers and electronic gadgets. The results of our own research revealed the general prevalence of problematic smartphone use in 15.2% of adolescents, its more frequent occurrence in girls (21.2%) than boys (5.0%), and a higher prevalence of vision problems (computer vision syndrome, dry eye syndrome) in smartphone-dependent adolescents compared with independent ones. The authors conclude that the approach to the prevention and treatment of pathological changes in the body (primarily from the organ of vision) caused by the irrational use of modern gadgets should be comprehensive, including recommendations for optimizing the regime and working conditions with electronic computer devices, as well as the use of various ophthalmological measures in a complex of preventive and therapeutic measures funds.

Impact of visual impairments on cognitive functions in older adults: insights from NHANES data.

Visual impairments (VI) are common in the older adults and may affect cognitive functions through mechanisms such as reduced sensory input and increased social isolation. Although current understanding of this association is incomplete, evidence suggests a potential link between poor vision and cognitive decline. This study aims to investigate the relationship between Subjective Visual Impairment (SVI), Objective Visual Impairment (OVI), and cognitive decline in the older adults, and assess whether these associations remain significant after controlling for multiple confounding factors. Data from the National Health and Nutrition Examination Survey (NHANES) for the years 1999-2002 were utilized, focusing on participants aged 60 and above. A total of 2,522 eligible participants were analyzed to assess their cognitive function and visual status. Weighted logistic regression models were used to explore the relationship between cognitive status and VI, progressively adjusting for confounding variables. Non-linear associations between cognitive score and VI were further explored using restricted cubic spline analysis. Visual impairments were significantly associated with worse cognitive function. Participants with SVI had a 3.437-fold higher risk of cognitive decline compared to those without SVI (95% CI: 2.648-4.461, p < 0.001). After adjusting for multiple covariates, the association remained significant (adjusted OR for SVI: 1.921, 95% CI: 1.346-2.742, p = 0.001; adjusted OR for OVI: 3.075, 95% CI: 1.740-5.433, p = 0.001). The restricted cubic spline analysis revealed a non-linear relationship between cognitive score and visual impairment, suggesting that the impact of visual impairment on cognitive decline varies across different levels of cognitive function. This study highlights a significant association between visual impairment and cognitive decline, even after controlling for multiple potential influencers. The findings support the integration of vision assessments into older adults care to identify and address vision problems early, potentially mitigating cognitive decline. The discovery of non-linear relationships further suggests that vision interventions may be particularly vital at certain stages of cognitive scores.

Spectrum of Visual Dysfunction Detected by a Novel Testing Protocol Within a Special School Eye Care Service.

Children with special educational needs are more likely to have vision problems than peers in mainstream education. Reports focus on visual acuity and refraction, overlooking visuoperceptual difficulties, including cerebral visual impairment. This article reports on the feasibility and outcomes of visual function testing performed during in-school visual assessments. A retrospective chart review was undertaken of children participating in a special school vision programme. The testing strategy included acuity, fields, contrast sensitivity, eye movements, accommodation, stereopsis, visual attention, refraction and a parent-completed questionnaire. The testing method was chosen based on the child's ability to engage with testing. 78 cases were identified (mean age 9.6 years). Low vision (worse than 6/19) was identified in 31%. All six tests of visual function were completed by 44% (mean 5.1; range 2-6). The mean number of atypical responses was 1 (range 0-4). Almost half (49%) showed at least one atypical response, most commonly visual attention (35%) and 25% had atypical eye movements. Overall, only 20% of children in the special school setting demonstrated a normal response to each test they were able to complete. Acuity testing alone does not demonstrate the atypical visual function of a child with special needs. Using techniques which require minimal cognitive, speech or motor function, a range of visual functions can be elicited during in-school testing. This testing strategy has the advantage of demonstrating areas of visual (dys)function impacting on the child's ability to access learning which can be immediately fed back to teaching staff and carers.

Unilateral molar Incisor hypomineralization influences the chewing side? an observational study in children.

The aim of this observational study was to compare the chewing patterns of children with unilateral Molar Incisor Hypomineralization (MIH) to those of children without MIH and to assess how the MIH severity influences the risk of chewing deviation. The hypothesis was that children with unilateral MIH would prefer to chew on the side opposite the defect. A total of 121 children were included in this investigation and evaluated for their preferred chewing side (PCS). This study population comprised 45 children with unilateral mild MIH (G1), 42 with unilateral severe MIH (G2), and 34 children in the control group (CG) who had no MIH but presented with unilateral chewing complaints. The PCS pattern differed among the groups, with 31% of the children with MIH and 55.9% of the CG showing bilateral chewing. The chewing deviation prevalence was 24% in the CG and 52% in G2 (χ2 test; p = 0.03). Binomial logistic regression analysis revealed that both severities of MIH influenced the likelihood of chewing deviation (χ2 = 7.33, p = 0.026). Children with unilateral severe MIH were 3.57 times more likely to exhibit chewing deviation than were those without MIH. The null hypothesis was rejected, indicating that unilateral MIH affects children's masticatory patterns, increasing the risk of chewing deviation. Unilateral chewing may be associated with various health issues, such as temporomandibular disorders (TMDs), decreased hearing, and vision problems. Unilateral MIH is a potential risk factor for unilateral chewing.

Attention-enhanced multimodal feature fusion network for clothes-changing person re-identification

Clothes-Changing Person Re-Identification is a challenging problem in computer vision, primarily due to the appearance variations caused by clothing changes across different camera views. This poses significant challenges to traditional person re-identification techniques that rely on clothing features. These challenges include the inconsistency of clothing and the difficulty in learning reliable clothing-irrelevant local features. To address this issue, we propose a novel network architecture called the Attention-Enhanced Multimodal Feature Fusion Network (AE-Net). AE-Net effectively mitigates the impact of clothing changes on recognition accuracy by integrating RGB global features, grayscale image features, and clothing-irrelevant features obtained through semantic segmentation. Specifically, global features capture the overall appearance of the person; grayscale image features help eliminate the interference of color in recognition; and clothing-irrelevant features derived from semantic segmentation enforce the model to learn features independent of the person’s clothing. Additionally, we introduce a multi-scale fusion attention mechanism that further enhances the model’s ability to capture both detailed and global structures, thereby improving recognition accuracy and robustness. Extensive experimental results demonstrate that AE-Net outperforms several state-of-the-art methods on the PRCC and LTCC datasets, particularly in scenarios with significant clothing changes. On the PRCC and LTCC datasets, AE-Net achieves Top-1 accuracy rates of 60.4% and 42.9%, respectively.

Untrained neural networks can demonstrate memorization-independent abstract reasoning

The nature of abstract reasoning is a matter of debate. Modern artificial neural network (ANN) models, like large language models, demonstrate impressive success when tested on abstract reasoning problems. However, it has been argued that their success reflects some form of memorization of similar problems (data contamination) rather than a general-purpose abstract reasoning capability. This concern is supported by evidence of brittleness, and the requirement of extensive training. In our study, we explored whether abstract reasoning can be achieved using the toolbox of ANNs, without prior training. Specifically, we studied an ANN model in which the weights of a naive network are optimized during the solution of the problem, using the problem data itself, rather than any prior knowledge. We tested this modeling approach on visual reasoning problems and found that it performs relatively well. Crucially, this success does not rely on memorization of similar problems. We further suggest an explanation of how it works. Finally, as problem solving is performed by changing the ANN weights, we explored the connection between problem solving and the accumulation of knowledge in the ANNs.

Correlation between choroidal thickness and the degree of myopia.

Myopia is a frequent visual problem, and the relationship between choroidal thickness (CT) and the degree of myopia has been a hot topic in myopia research. This work aimed to explore the correlation between CT and the degree of myopia, providing a reference for diagnosing and treating myopia. A cross-sectional study was conducted from September 2021 to December 2022, collecting data from 95 myopic patients aged between 18 and 50 years in the outpatient department. All subjects' CT in the macular center (MC), spherical equivalent (SE), and other ocular parameters were measured. Furthermore, the Pearson correlation coefficient (PCC) analyzed relationships between CT and various factors. The choroid was thickest in the MC and gradually became thinner towards the periphery, with the thinnest region located nasally in the healthy group. In the mild, moderate, and severe myopia groups, the choroid was thickest at 1,000 μm temporal to the fovea, becoming thinner towards the periphery, with the thinnest region located nasally. The MC's CT was correlated with a family history of myopia, SE, axial length (AL), and intraocular pressure (IOP). Meanwhile, there was a negative linear relationship between AL and CT in the MC (standard coefficient (SC) of -0.596, P-value of 0.000, tolerance of 0.217, and variance inflation factor (VIF) of 4.467), and a positive linear correlation between SE and CT in the MC (SC of 0.205, P-value of 0.013, tolerance of 0.257, and VIF of 3.792). This work provided clues for further understanding of the pathogenesis of myopic eyes and served as a scientific basis for early screening and treatment of myopia. Additionally, investigating the correlation between myopia and CT can also yield a reference for developing personalized myopia management strategies, which will help slow down myopia's progression and prevent related complications.

GETr: A Geometric Equivariant Transformer for Point Cloud Registration

AbstractAs a fundamental problem in computer vision, 3D point cloud registration (PCR) aims to seek the optimal transformation to align point cloud pairs. Meanwhile, the equivariance lies at the core of matching point clouds at arbitrary pose. In this paper, we propose GETr, a geometric equivariant transformer for PCR. By learning the point‐wise orientations, we decouple the coordinate to the pose of the point clouds, which is the key to achieve equivariance in our framework. Then we utilize attention mechanism to learn the geometric features for superpoints matching, the proposed novel self‐attention mechanism encodes the geometric information of point clouds. Finally, the coarse‐to‐fine manner is used to obtain high‐quality correspondence for registration. Extensive experiments on both indoor and outdoor benchmarks demonstrate that our method outperforms various existing state‐of‐the‐art methods.