To describe a novel, low-cost, and high-quality intraoperative video recording setup for oculoplastic surgery using a magnetic neck-mounted smartphone holder and iPhone with the Final Cut Camera app. A magnetic neck mount (Tianzhu Insta 360, China) and an iPhone 16 Pro Max were adapted for surgical recording by securing the holder around the binocular base of the operating microscope. The magnetically attached phone provided a stable and adjustable platform, allowing intraoperative view adjustments via the microscope's handles or foot pedal without compromising sterility. The Final Cut Camera app enabled 4K video capture with manual control of focus, exposure, and white balance. Its Live Multicam feature allowed simultaneous multi-angle recording and real-time monitoring via a compatible iPad, enabling an assistant to control framing and clarity intraoperatively. A multi-port adapter was used to support continuous power and external memory. Over six months, more than 50 oculoplastic procedures, including dacryocystorhinostomy, orbitotomy, and eyelid surgeries, were successfully recorded using this setup. All recordings were stable, centered, and of high image quality. Screenshots captured during surgery clearly depicted anatomical structures under standard operating room lighting. This smartphone-based recording method offers a simple, cost-effective, and ergonomically practical alternative to traditional surgical video systems in oculoplastic procedures. Its adaptability, ease of use, and compatibility with sterile environments make it an ideal solution for surgical documentation and teaching, especially in resource-limited settings.
暂无摘要（点击查看详情）
To validate a custom smartphone application for at-home visual acuity (VA) measurement in children. A total of 452 children aged 3-17.5 years participated. Certified examiners measured in-office test-retest VA (logMAR) using gold-standard Amblyopia Treatment Study HOTV (3-to-6-year-olds, younger cohort) or electronic Early Treatment of Diabetic Retinopathy Study (7-to-17.5-year-olds, older cohort) protocols at 3-4.5 m and app-based VA at 1.5 m. Caregivers measured at-home app-based VA at 1.5 m. Comparing at-home app-based with gold-standard VA, in eyes 20/40 or better, 95% (143/151) and 93% (91/98) of the younger and older cohorts were within 2 lines, respectively (mean differences: younger = -0.03, older = -0.04; 95% limits-of-agreement half-width (LOA): younger = ±0.26, older = ±0.22). In eyes 20/50 or worse, 66% (42/64) and 75% (76/101) of the younger and older cohorts were within 2 lines, respectively (mean differences: younger = 0.11, older = 0.13, LOA: younger = ±0.50, older = ±0.51). Comparing in-office app-based VA with gold-standard VA, in eyes 20/40 or better, 98% (160/164) and 94% (99/105) of the younger and older cohorts were within 2 lines, respectively (mean differences: younger = -0.03, older = -0.03; LOA: younger = ±0.22; older = ±0.24). In eyes 20/50 or worse, 85% (60/71) and 91% (101/111) of the younger and older cohorts were within 2 lines, respectively (mean differences: younger = 0.04; older = 0.04; LOA: younger = ±0.39; older = ±0.24). For gold-standard test-retest, in eyes 20/40 or better, 99% (163/164) and 99% (104/105) of the younger and older cohorts had retest within 2 lines, respectively (mean differences: younger = 0.00; older = 0.01; LOA: younger = ±0.17; older = ±0.11). For 20/50 or worse, 92% (66/72) and 100% (111/111) in the younger and older cohorts were within 2 lines, respectively (mean differences: younger = 0.01; older = 0.02; LOA: younger = ±0.35; older = ±0.15). Our app demonstrated good concordance with the gold standard at home and in the office for eyes with VA of 20/40 or better. However, concordance decreased considerably for eyes with VA 20/50 or worse, particularly at home.
Gram staining provides rapid microbiological information that may assist in empirical antimicrobial selection; however, the results are often interpreted by microbiological specialists who are not always available. Therefore, we developed a computer-aided diagnosis system using artificial intelligence trained on microscopic images of Gram-stained urine, captured with an iPhone, using the Bartholomew and Mittwer method. The system interprets Gram-stained urine samples and classifies bacterial morphology (Class 1: 7 predefined morphology categories) and 17 predefined species-level categories (Class 2). In this retrospective observational study, five imaging devices and two staining methods (Bartholomew and Mittwer, Favor) were compared. Urine specimens were collected from two hospitals between 1 April and 31 December 2022. Validation images were generated using five devices (four smartphones and one microscopic camera). We used a micrometer with microscopy with all smartphones; some iPhone images were taken without a micrometer. Favor staining was only imaged using an iPhone without the micrometer. Image data sets were generated from 433 clinical and 17 spiked samples. The overall accuracy was 0.804 for Class 1 and 0.640 for Class 2. Images taken by the microscopic camera had the highest accuracy and kappa coefficient, whereas the AQUOS smartphone had the lowest accuracy and kappa coefficient. The accuracy of images created without a micrometer was 0.885 for Class 1 and 0.666 for Class 2. The Bartholomew and Mittwer method had better accuracy and a better kappa coefficient. Overall, accuracy depended on the staining method used in the training data, not on the imaging device.IMPORTANCEGram staining provides rapid information on both the site of infection and likely pathogens, guiding empirical antimicrobial selection. However, interpretation requires infectious disease expertise, which is not always available. We developed an artificial intelligence-based diagnostic support system trained on iPhone images of Gram-stained urine using the Bartholomew and Mittwer method to classify bacterial morphology (Class 1) and inferred species (Class 2). To provide essential baseline data on factors influencing accuracy and reliability, we compared Gram-stained urine images from two hospitals obtained with five imaging devices and two staining methods. Microscopic camera images showed the highest accuracy, whereas an AQUOS smartphone showed the lowest. Images without a micrometer performed better, and the Bartholomew and Mittwer method outperformed the Favor method. Accuracy increased when confidence levels were higher. Our findings suggest that using the same staining method as the training data and avoiding micrometer noise are critical, while device differences are less influential.
The risk of genitourinary toxicity during radiotherapy for prostate cancer was found to be lower for bladder volumes ≥200 ml. An app that reminds patients daily to drink water might be helpful. Before being investigated in patients, an app should be tested in healthy volunteers. Thirty healthy volunteers were included in this prospective study and asked to test the app and affirm (=satisfaction) or negate nine statements. These statements belonged to the sections 'Download and installation' (two statements), 'Navigation' (two statements), or 'Content/functions' (five statements). If a satisfaction rate was <60%, the app was to be considered not useful. If it was <80%, the app needed optimization. iPhone users (n=18) were compared to Android users (n=12). Satisfaction rates (participants affirming a statement) were 90.0% (27 out of 30 participants) and 86.7% (26 out of 30 participants) regarding the two statements belonging to the Download and installation section. Regarding the two statements of the Navigation section, satisfaction rates were 100% (28 out of 28) and 96.6% (28 out of 29), respectively. For the Content/functions section, satisfaction rates were 79.3% (23 out of 29 participants) for the statement: "The app reminded me at the selected times". For the other four statements, satisfaction rates were each 100% (25 out of 25, 27 out of 27, 28 out of 28, and 29 out of 29 participants). Significant differences between iPhone and Android users were not observed. When looking at the subgroups of iPhone and Android users, two additional aspects were identified that needed modifications. Although the new reminder app was mainly rated usable, some modifications were required. Our findings highlight that a pre-study in healthy volunteers is important.
In aesthetic clinical trials, image self-capture using mobile devices may help reduce burden on clinic resources, increase data quality, and lower barriers to study participation. This study aimed to develop a mobile device app to help participants self-capture clinically usable images. The Allergan Aesthetic (an AbbVie Company) mobile image app was designed to auto-capture images while directing study participants on distance, head position, and expression to capture a high-quality clinical image. To assess resolution and optimal lighting conditions, images captured using the app in office, at home, and in outdoor settings were compared with those from a Canfield VISIA-CR system (Canfield Scientific). Objective image quality assessment of facial images captured using the app with an iPhone XR (Apple Inc) and iPhone 12 (Apple Inc), compared with images captured using the Canfield VISIA-CR with a digital single-lens reflex camera and the Canfield mobile image capture app with a variety of Android (Google) and iOS (Apple Inc) devices, was conducted using the Blind/Referenceless Image Spatial Quality Evaluator (BRISQUE). Clinical utility was assessed by calculating inter- and intrarater variability for severity ratings of participants' lateral canthal lines (LCL) or forehead lines (FHL) obtained from app-captured images compared with ratings based on in-person evaluations performed by a physician. Usability was assessed according to the ISO (International Organization for Standardization)/IEC (International Electrotechnical Commission) 250101 standard. The Allergan Aesthetic mobile image app was found to perform best under natural light and had image resolution insufficient for assessing minor facial structures, but appropriate for larger structures (eg, FHL). A total of 3968 images were assessed using BRISQUE. Images captured with the Allergan Aesthetic mobile image app had better image quality than those captured using other modalities, as indicated by lower mean BRISQUE scores of 14.05-19.81 compared with Canfield VISIA-CR with a DSLR (34.47) and the Canfield mobile image capture app (23.43). LCL and FHL were rated both in person and digitally in 68 and 71 participants, respectively (median age 52-56 y; 48% to 52% female; 75% to 78% White). Interrater reliability between clinician live evaluations and independent photo review of self-captured photos based on intraclass correlation coefficients (ICCs) was substantial (0.61-0.80) to almost perfect (0.81-1.00) for all raters (LCL: ICC 0.75-0.91 at rest and 0.79-0.89 at maximum contraction; FHL: ICC 0.77-0.93 at rest and 0.70-0.89 at maximum contraction). After 2 iterations of improvements, mean usability ratings of the app experience (out of 5) were as follows: easy to complete=3.2, enjoyable=3.1, satisfied with the level of guidance provided=3.2, and likely to complete a full session without exiting=4.1. The Allergan Aesthetic mobile image app delivers consistent, high-quality images that allow for assessment of LCL and FHL in good agreement with in-person evaluation. Image self-capture using mobile devices may help reduce clinic costs and remove barriers to participation in aesthetic clinical trials.
Lymphedema, a chronic and incurable condition with limited therapeutic options, has limited options to quantitatively assess functional changes during its development; as a result, a deeper understanding of its pathophysiology remains hindered. To characterize lymphatic alterations and their association with disease pathology in a clinically relevant model in the rat, we developed a longitudinal iPhone-based volumetry method combined with non-invasive NIR analysis of lymphatic function. Secondary lymphedema was induced by surgery and single-dose irradiation. iPhone volumetry provided longitudinal measurements of hindlimb volume, while NIR imaging quantified the pumping function of major lymphatic vessels in the popliteal area. Among 30 rats, lymphedema developed in 80%, defined as interlimb volume differences exceeding 5% and persisting through 14 days. In all rats with lymphedema, disease persisted until the end of the study at postoperative day 42 (P = 0.0015). NIR imaging revealed lymphatic dilation, dye extravasation, and lymphangiogenesis in affected limbs. Lymphatics in limbs with lymphedema exhibited increased contraction frequency, reduced amplitude, and diminished transport compared to baseline and contralateral controls (all P < 0.05). In contrast, rats that did no develop lymphedema showed no postoperative functional changes, although at baseline they displayed higher frequency and lower amplitude and transport compared with LE rats (all P < 0.001). Baseline transport values correlated negatively with swelling (r = -0.44, P = 0.002), as determined by ROC analysis, which yielded an AUC of 0.83, a sensitivity of 83.3%, and a specificity of 82.6%. Histopathology at day 42 confirmed significant dermal thickening and fat deposition in LE limbs (P < 0.001 and P = 0.002, respectively). Longitudinal volumetry and NIR imaging applied to a clinically relevant animal model suggest a strong association between swelling and lymphatic function, which could provide deeper insight into lymphedema pathophysiology and represent valuable tools for future research and therapeutic development.
This study aimed to assess the diagnostic accuracy of digital intraoral photographs obtained using smartphones and a macro camera in evaluating oral health among adults. A total of 200 adult patients underwent clinical and radiographic examinations using the Decayed, Filled Teeth (DFT) Index, Caries Assessment Spectrum and Treatment (CAST) Index, Plaque Index (PI), and Modified Gingival Index (MGI). Intraoral photographs were taken using three devices: Samsung S23 Ultra, iPhone 14 Pro, and Canon EOS 400D with macro lens. Following the clinical recording of DFT, CAST, PI, and MGI scores by two calibrated examiners as the reference standard, intraoral photographs were captured by a third dentist and independently evaluated by two separate blinded examiners to compare the diagnostic accuracy of the devices against the clinical findings. Non-parametric analyses were conducted using the Friedman test with Dunn's post hoc test, Wilcoxon test and agreement between clinical and photographic methods was evaluated via the Bland-Altman method (p < 0.05). The macro camera demonstrated the highest inter-rater reliability for FT scores (ICC = 0.886), while iPhone-derived MGI scores showed the lowest reliability (ICC = 0.624). Statistically significant differences were found among all imaging devices for all indices (p < 0.001), except for MGI. Bland-Altman analysis showed that most values fell within the 95% limits of agreement, indicating good concordance with clinical data. Smartphone and macro camera photographs provided comparable diagnostic results for caries and restorations. However, limitations remain in the assessment of periodontal parameters via photographic methods. Smartphone-based intraoral photography can serve as a practical diagnostic tool in teledentistry.
Clinical studies have shown that aortic arch pulse-wave velocity (PWVaa), a measure of local aortic stiffness, is a strong independent predictor of subsequent white matter hyperintensity volume and white matter integrity, both associated with cognitive decline, elevated stroke risk, vascular dementia, and neurodegenerative diseases. Total arterial compliance (TAC), a measure of global arterial stiffness, has been recognized as a marker of preclinical vascular disease. This study introduces a smartphone-based method for the noninvasive measurement of PWVaa and TAC using carotid pressure waveforms acquired via smartphone. This method uses intrinsic frequency analysis of smartphone-acquired (iPhone) carotid pressure waveforms to assess PWVaa and TAC. The method was trained, validated, and blind-tested on a cohort of 132 participants aged 20 to 90 years, including both healthy individuals and those with cardiovascular disease, all of whom underwent cardiac magnetic resonance imaging, tonometry, and iPhone waveform measurements. In the blind test set, our method achieved Pearson correlations of 0.81 and 0.80 for PWVaa and TAC, with biases of -0.20 m/s and -0.06 mL/mm Hg and limits of agreement of -4.09 to 3.68 m/s and -0.52 to 0.40 mL/mm Hg, respectively. In the heart failure population, correlations were 0.81 for both, with a PWVaa a bias of -1.07 m/s and TAC bias of -0.06 mL/mm Hg. Our smartphone-based method enables accurate assessment of local and global arterial stiffness metrics (PWVaa and TAC). It offers easy-to-use monitoring of vascular aging and arterial health, with important implications for identifying patients at higher risk of neurodegenerative and cardiovascular diseases. URL: clinicaltrials.org; Unique Identifier: NCT02240979.
Dental photography plays a key role in diagnosis, documentation, and communication in the field of dentistry. The digital single-lens reflex (DSLR) cameras are the gold standard for image quality, but advances in modern smartphone technology necessitate evaluating their performance in resolution, distortion, magnification, color accuracy, and overall image quality. The purpose of this study was to compare DSLR cameras and smartphone cameras regarding resolution, distortion, magnification, color accuracy, and overall image quality in intra- and extraoral dental photography. A Nikon Z5 full-frame DSLR with a 105 mm macro lens and twin flash (control) was compared with three flagship smartphones: iPhone 15 Pro, Google Pixel 8, and Samsung S24. Ten participants were photographed in five standardized dental views: maximum intercuspation, right lateral occlusion, maxillary anterior with black contrastor, mandibular arch (mirror view), and front profile. Images were analyzed using the GNU Image Manipulation Program (GIMP 2.1); statistical analysis was performed with Statistical Package for the Social Sciences (SPSS) 27. The Samsung S24 closely matched the DSLR in shade accuracy and distortion. Google Pixel 8 produced an acceptable resolution (<300 dots per inch). The iPhone 15 Pro demonstrated superior performance in reducing distortion and maintaining clarity (P < 0.05). Despite the convenience and affordability of smartphone cameras, DSLR systems demonstrate superior performance in terms of magnification, image resolution, and color accuracy. These advantages render DSLRs more suitable for clinical applications that demand high precision and diagnostic reliability in dental photography.
Poor sleep is common and detrimental to health. Smartphone use is often noted as a sleep disruptor, but evidence remains limited and inconsistent. This necessitates research focused on objective, longitudinal designs, as well as analytical approaches that can reveal lagged and reciprocal relationships that capture within- and between-person effects. To address these gaps, the current study investigated within- and between-person lagged and reciprocal effects of sleep duration and smartphone use of 68 participants through longitudinal and objective data donated from iPhones and Apple Watches across 14 consecutive days. Apple Watches objectively measured total sleep and sleep stage durations (REM, core, and deep sleep), while iPhones assessed total smartphone use duration and in-bed smartphone use. Two Dynamic Structural Equation Models (DSEMs), one with total sleep and one with sleep broken down into three sleep stages, were conducted. At the within-person level, more total smartphone use increased same-day in-bed smartphone use, β = .25 (95% CI .20, .31), which in turn led to more same-day overall sleep, β = .08 (95% CI .02, .14). Additionally, results indicated stable between-person habits, with strong day-to-day associations for each variable with its own next-day value, β = .53-.82 (95% CI .47, .88). Findings contradict the perspective of smartphones as sleep disruptors, despite leaving open whether this added sleep means poorer rest or a real benefit of in-bed smartphone use. Furthermore, the strength of the between-person results emphasizes the importance of habits in this relationship. In studying day-to-day smartphone use and sleep, these findings provide nuanced empirical insights supporting health and policy recommendations regarding smartphone use and sleep hygiene.
Limited access to cadavers necessitates the availability of digital resources for anatomy education. Smartphone-based photogrammetry offers a promising solution for creating three-dimensional (3D) and augmented reality (AR) models. This study compared two mobile photogrammetry applications (Qlone and Polycam) that have been used in modern anatomical education. Human cadaveric specimens were prepared and scanned using an iPhone 12 equipped with each application. Initially, a structured qualitative assessment of the applications and their outputs was performed by three experts using a Likert scale, considering image quality, medical utility, and technical factors. After selecting the superior application, diverse anatomical specimens were reconstructed into 3D/AR models. Nine clinical anatomy experts used a Likert scale to rate 20 selected models in four areas: realism, clarity, completeness, and educational value. The comparative analysis indicated that Polycam is significantly superior to Qlone in 3D realism, resolution, shape fidelity, and educational value, despite Qlone's strengths in cost and processing speed. Polycam was then used to create high-fidelity 3D models of complex structures, which were refined and uploaded to a web-based platform. Experts scored the models as "good" to "excellent" in all four evaluation domains, with particularly high scores for anatomical realism in bones and solid organs. In conclusion, the Polycam application is useful for creating high-quality 3D/AR models of human anatomy. These digital resources maintain anatomical accuracy and enable immersive learning, making them an invaluable supplement to traditional dissection in medical schools.
Equine lameness diagnosis largely relies on subjective visual assessments, which can be biased. Although marker-based methods, force plates and inertial measurement units (IMUs) provide objective measurements, they require specialized setups. Vision-based algorithms offer a portable, markerless alternative, but their accuracy needs thorough testing. To evaluate a custom vision-based algorithm for estimating the groundline across multiple camera angles, including handheld use in horses trotting on a treadmill. Experimental comparative study. Eight Standardbred trotter mares were recorded trotting on a high-speed treadmill using seven iPhones positioned at various heights and angles, including a handheld device. A trained deep neural network algorithm placed 2D keypoints on each video frame. Vertical Displacement Signals (VDS) for the eye, withers and croup (tuber sacrale) were computed relative to either an algorithm-estimated or a fixed treadmill groundline. Maximum (Maxdiff) and minimum (Mindiff) stride values were compared using Bland-Altman analysis, scatter plots and histograms. The effect of handheld use on variability and accuracy was assessed by comparing results from a handheld camera to those from a static camera. Groundline estimation closely matched the fixed reference, exhibiting near-zero mean angle error and low mean average error (MAE = 0.45°; n = 242.192). Maxdiff and Mindiff stride-level (n = 36.981) MAE were 0.5 mm, with clinically acceptable additional variability introduced by handheld use at the trial level (Maxdiff and Mindiff MAE < 1.8 mm; n = 357). Treadmill-based data and a single breed/coat colour may limit generalizability to other settings. The vision-based algorithm accurately estimates the groundline and stride VDS parameters from various camera setups, including handheld. Further validation in diverse environments and against other objective gait analysis systems is recommended.
Tinnitus is a complex condition with significant heterogeneity in its presentation, and its risk factors remain poorly characterized, posing challenges for prevention, diagnosis, and treatment. This study aimed to assess the prevalence, characteristics, and risk factors of tinnitus in the United States (U.S.) using large-scale survey data. We conducted a cross-sectional analysis of 125,252 volunteer adults (≥ 18 years) enrolled in the Apple Hearing Study, a nationwide app-based cohort of iPhone users in the U.S. (November 2019–November 2022). The outcomes were the weighted prevalence of any tinnitus and bothersome tinnitus, measured using self-reported tinnitus frequency, duration, awareness, loudness, and interference with hearing. Age-adjusted and multivariable logistic regression models were applied to analyze the odds ratios of self-reported potential risk factors on tinnitus, and a weighted decision tree identified the strongest predictors of bothersome tinnitus. The estimated weighted national prevalence of any tinnitus was 30.8% (95% Confidence Interval [CI]: [30.3%, 31.2%]) and bothersome tinnitus was 11.6% (95% CI: [11.3%, 11.9%]). Controlling for age, sex, race/ethnicity, and other sociodemographic characteristics, self-rated hearing ability was the strongest risk factor for any tinnitus (odds ratios of 4.52 [95% CI: 4.03–5.06] and bothersome tinnitus 8.88 [95% CI: 7.52–10.49], comparing poor to excellent hearing). The odds of both types of tinnitus increased with age, peaking in the 60–64 age group (2.01 [95% CI: 1.77–2.28] for any tinnitus and 2.72 [95% CI: 2.24–3.92] for bothersome tinnitus) after adjusting for the same set of variables. Non-Hispanic Whites had higher odds of any and bothersome tinnitus compared to other race/ethnicities. A reported history of occupational noise exposure was associated with higher odds of any and bothersome tinnitus. Approximately 3 in 10 U.S. adults are estimated to experience any tinnitus, and about 1 in 10 affected by bothersome tinnitus. Tinnitus is associated with worse self-rated hearing ability, age, race/ethnicity, and a history of workplace noise. These results align with prior epidemiological estimates and demonstrate the feasibility of using app-based platforms to collect large-scale, high-quality hearing health data. The online version contains supplementary material available at 10.1186/s12889-026-27048-2.
Polychromia remains one of the most reproducible dermoscopic indicators of melanoma, yet its clinical assessment is predominantly subjective. Shannon entropy has been proposed as an objective measure of color heterogeneity in pigmented skin lesions. However, global entropy derived from grayscale or composite RGB histograms may primarily capture luminance dispersion rather than true chromatic complexity. This proof-of-concept study evaluated whether global Shannon entropy quantifies polychromia and whether channel-specific entropy metrics more accurately reflect chromatic heterogeneity. Smartphone photographs (iPhone 13 Pro Max, Apple Inc.) of a histopathologically confirmed superficial melanoma, a benign junctional nevus, and their respective perilesional skin were analyzed using ImageJ (National Institutes of Health). Intensity histograms were generated in an 8-bit grayscale, composite RGB mode, and separately for the red, green, and blue channels. Shannon entropy (H, log₂), inter-channel entropy differences (ΔR-G, ΔR-B, and ΔG-B), red-channel asymmetry (Aᴿ), and a composite Polychromia Index (Iᴾ) were computed for each region of interest, with all metrics normalized to perilesional skin to control for illumination and baseline heterogeneity. Grayscale and RGB-composite histograms yielded nearly identical entropy values for both lesions, confirming that global entropy primarily reflects luminance contrast rather than chromatic structure. By contrast, channel-specific analysis revealed marked divergence in the melanoma, with normalized inter-channel entropy differences showing substantial residual chromatic heterogeneity (ΔG-B_residual = +12.31; ΔR-G_residual = +9.71), representing 600-4000% increases compared with the nevus. The normalized Polychromia Index (Iᴾ) demonstrated an 8.22-unit separation between the melanoma (+6.84) and the nevus (-1.38), closely aligning with the visual impression of color variegation. These findings indicate that global Shannon entropy does not meaningfully quantify polychromia under real-world smartphone imaging conditions. Channel-specific entropy and inter-channel metrics, however, reliably discriminate chromatically heterogeneous lesions from uniform ones. This low-cost, reproducible framework offers a physiologically interpretable approach to objective color heterogeneity assessment and holds potential for teledermatology and automated melanoma-detection systems.
Vision Transformers (ViTs) mark a revolutionary advance in neural networks with their token mixer's powerful global context capability. However, the pairwise token affinity and complex matrix operations limit its deployment on resource-constrained scenarios and real-time applications, such as mobile devices, although considerable efforts have been made in previous works. In this paper, we introduce CAS-ViT: Convolutional Additive Self-attention Vision Transformers, to achieve a balance between efficiency and performance in mobile applications. Firstly, we argue that the capability of token mixers to obtain global contextual information hinges on multiple information interactions, such as spatial and channel domains. Subsequently, we propose Convolutional Additive Token Mixer (CATM) employing underlying spatial and channel attention as novel interaction forms. This module eliminates troublesome complex operations such as matrix multiplication and Softmax. We introduce Convolutional Additive Self-attention(CAS) block hybrid architecture and utilize CATM for each block. And further, we build a family of lightweight networks, which can be easily extended to various downstream tasks. Finally, we evaluate CAS-ViT across a variety of vision tasks, including image classification, object detection, instance segmentation, and semantic segmentation. Our M and T model achieves 83.0%/84.1% top-1 with only 12M/21M parameters on ImageNet-1K. Meanwhile, throughput evaluations on GPUs, ONNX, and iPhones also demonstrate superior results compared to other state-of-the-art backbones. Extensive experiments demonstrate that our approach achieves a better balance of performance, efficient inference and easy-to-deploy. Our code and model are available at: https://github.com/Tianfang-Zhang/CAS-ViT.
The aim of this study was for the first time to evaluate of intraoral digital photography in assessing buccal surface of anterior composite restorations using a smartphone (iPhone 14 Pro), a smartphone with a lens (2IN1 Phone Macro Lens), and a digital camera with a macro lens (Canon Rebel XTi), compared to clinical examination, based on World Dental Federation (FDI) criteria. A total of 185 anterior composite restorations were evaluated by calibrated restorative dentistry specialists. Restorations were scored according to the FDI criteria as intact, requiring repair, or needing replacement. Clinical examination was considered the gold standard. Photographs were taken under standardized conditions, and inter-observer and inter-method agreement were analyzed using Cohen’s Kappa and intraclass correlation coefficient (ICC). High inter-observer agreement was observed across all methods (Kappa = 0.928–1.0). Good to excellent agreement was found between clinical examination and digital photography methods for FDI final scores (Kappa = 0.775–0.973, p < 0.001). Photographs taken with the lens-equipped smartphone and macro camera showed higher agreement with clinical examination (Kappa = 0.973). Digital photography, particularly with a lens-equipped smartphone and macro camera, offers effectiveness comparable to clinical examination in evaluating anterior composite restorations. This study demonstrates that intraoral digital photography, particularly with a lens-equipped smartphone and macro camera, achieves diagnostic outcomes closely aligned with clinical examination using FDI criteria. While smartphones alone are practical, lens-assisted imaging enhances accuracy, supporting its use as a cost-effective alternative to professional macro cameras.
We present a case of pancake syndrome (PS) uniquely diagnosed using the macro photography function of a smartphone to detect mite contamination in flour. The patient was a 12-year-old girl with a history of allergic rhinitis who developed severe allergic symptoms, including wheezing and hypoxemia, immediately after consuming homemade Takoyaki-a Japanese dish made with a wheat flour-based batter and cooked in a molded pan-prepared with flour that had been stored at room temperature for several months. Emergency treatment with intramuscular epinephrine and intravenous corticosteroids led to symptom resolution. Based on her clinical history and elevated dust mite-specific IgE levels, PS was suspected. For diagnostic confirmation, the suspect flour was examined using the macro mode of an Apple iPhone 15 Pro, which successfully revealed live mites. This case highlights the potential utility of smartphones in emergency department settings for diagnosis, patient education, and preventive guidance.
The integration of markerless motion capture systems such as OpenCap with force platforms expands the possibilities of biomechanical analysis in low-cost environments; however, it requires robust temporal synchronization procedures in the absence of shared hardware triggers. To develop and validate an automatic synchronization algorithm based on heel kinematic events to align OpenCap data with force platform signals during lower-limb functional exercises. Thirty normal-weight adult women (18-45 years) were evaluated while performing between 11 and 14 functional tasks (60° and 90° squats, lunges, sliding variations, and step exercises), yielding 330 motion records. Kinematics were estimated using OpenCap (four iPhone 12 cameras at 60 Hz), and kinetics were recorded using BTS P6000 force platforms synchronized with an OptiTrack system (Gold Standard). The algorithm detected heel contact from the filtered vertical coordinate and aligned this event with the initial rise in vertical ground reaction force. Validation against the Gold Standard was performed in 20 squat repetitions (10 at 60° and 10 at 90°) using Pearson correlation, RMSE, and MAE of the time-normalized and amplitude-normalized (0-1) vertical ground reaction force (vGRF). The algorithm successfully synchronized 92.5% of the 330 records; the remaining cases showed kinematic noise or additional steps that prevented robust event detection. During validation, correlations were r = 0.85 (60°) and r = 0.81 (90°), with Root Mean Square Error (RMSE) < 0.17 and Mean Absolute Error (MAE) < 0.14, values representing less than 0.1% of the peak force. The heel-contact-based algorithm allows accurate synchronization of OpenCap and force platform signals during lower-limb functional exercises, achieving performance comparable to hardware-synchronized systems. This approach facilitates the integration of markerless motion capture in clinical, sports, and occupational settings where advanced dynamic analysis is required with limited infrastructure.
Due to academic pressures and irregular schedules, university students often face challenges in maintaining healthy movement behaviours (including sleep, physical activity, and screen time), which are interrelated and influence both physical and mental health. Smartwatch- and smartphone-based ecological momentary assessments (EMAs) and ecological momentary interventions (EMIs) offer real-time, context-aware strategies to promote movement behaviours. This pilot study aims to assess the feasibility and preliminary effectiveness of a hybrid approach that combines continuous digital monitoring of movement behaviours with sequentially embedded randomised controlled trials (RCTs) evaluating EMIs. MOVE@NUS pilot study employed a five-month hybrid design that combines continuous passive monitoring (primarily via Apple Watches, supplemented by iPhones) with three embedded RCTs targeting sleep (RCT-1), physical activity (RCT-2), and screen time (RCT-3). For each RCT, participants are randomised on a 1:1:1 schedule (control, intervention 1, intervention 2). Eligible participants are first-year undergraduates at the National University of Singapore, aged 18-25 years, who own or regularly use an iPhone and an Apple Watch. EMIs, delivered via the study app, comprise standard health messages or personalised reminders based on HealthKit data or participants' self-reported behaviours and preferences. Self-reported measures include eight EMA bursts (three-day periods every two weeks) and online questionnaires at baseline, midway (2.5 months), and endpoint (5 months). All EMIs and EMAs are text-based and can be completed in under two minutes. Feasibility outcomes include recruitment, engagement, and user experience assessed through quantitative surveys and semi-structured interviews. Preliminary effectiveness will be explored separately for each RCT, comparing movement behaviours between intervention and control groups. Findings from this study will inform the development of scalable and longer-term digital intervention cohorts for promoting healthier lifestyles among university students. Furthermore, as university students soon transition to the workforce, insights gained will inform scalable digital health interventions for broader populations. ClinicalTrials.gov ID NCT06597890 First Posted: 19 September 2024.