Digital infrared reflex testing and asynchronous teleconsultation for infant eye screening

Abstract

Early detection and documentation of anterior segment opacities in infants (aged 1–6 months) is limited by resource constraints and specialist availability. This study aimed to evaluate the usefulness of digital infrared reflex testing (IRRT) and to estimate the feasibility of asynchronous teleconsultation between primary care paediatricians and ophthalmologists for infant eye screening. From April 2024 to March 2025, 189 infants aged 1–6 months underwent both digital IRRT and standard red reflex testing (RRT) in primary care settings. Digital images and clinical data were evaluated using teleconsultation between paediatricians and ophthalmologists. The feasibility endpoints included the proportion of successful teleconsultations, user acceptability, and technical barriers. IRRT versus RRT diagnostic agreement and the IRRT inconclusive examination rate were also assessed. Teleconsultation was feasible in 84% of the cases. The principal barriers were operational issues (time and setting) and staff training needs. Digital IRRT had an inconclusive result rate of 5.3% and high agreement with RRT (AC1 = 0.994, 95% CI: 0.983–1.000). Digital IRRT combined with teleconsultation is a feasible and accurate approach for infant eye screening in primary care settings, supporting image documentation, integration into telemedicine workflows, and potential future AI applications.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-40517-3.

Introduction

Early detection of ocular abnormalities in infants is critical to prevent visual impairment and support timely intervention. Traditional screening methods, such as red reflex testing (RRT), require skilled examiners and can be limited by resource availability. Digital infrared reflex testing (IRRT) combined with teleconsultation offers a promising approach for expanding effective infant eye screening in primary care settings.

Digital imaging of the eye’s anatomy and optical properties form the foundation for screening and diagnosing ocular abnormalities. Paediatric photo- and videoscreeners utilize near-infrared light (IR, 700–950 nm) as the primary source because of its deeper tissue penetrance properties, increased patient comfort during examination, and reduced effects such as pupil miosis and accommodative changes¹. IR light is a safe alternative to visible light (400–750 nm) for performing digital IRRT, an instrument-based method that assesses the retina’s reflection of IR in newborns and infants and is used to screen for pathologies that significantly reduce image transmission to the retina, including congenital cataracts, corneal opacities, and persistent foetal vascularization^2–4. The device simultaneously projects an IR-light beam into both eyes while an IR camera captures the reflected retinal light, producing a monochromatic gray reflex within the pupil. Normal IR images exhibit varying luminance saturation profiles due to retinal pigmentation and refractive state of the eye: uniform gradient in emmetropia, reversed (opposite the light source) in myopia, normal (same side as light source) in hyperopia, and meridian-dependent in astigmatism. Abnormal IR images display media opacities as dark shadows against the gray background.

The American Academy of Paediatrics currently recommends traditional visible-light, ophthalmoscope-based RRT prior to hospital discharge and subsequent routine health visits^5,6; however, instrument-based IR imaging now outperform direct ophthalmoscopy and can be used by non-specialist in community settings. These devices are portable, noninvasive and potentially operable by health personnel in routine screening programs^7,8, enabling collaboration between primary eye care providers and ophthalmologists to improve diagnostic accuracy and ensure high-quality patient data documentation for follow-up⁹. Despite these advantages, large-scale teleconsultation implementation remains hindered by the need for careful alignment of technology, organizational structures, policies, and funding frameworks¹⁰, representing a major barrier to systematic pediatric vision screening.

In this context, this study aimed to assess whether digital IRRT performed by primary care paediatricians and asynchronously evaluated by ophthalmologists provides a feasible and scalable approach for systematic screening of anterior segment opacities in infants, directly addressing current organisational and expertise barriers.

Methods

Study design and ethics

This pilot prospective observational study was conducted in accordance with the Declaration of Helsinki principles and received ethical approval from the local health care trust—Azienda Provinciale per i Servizi Sanitari (APSS) di Trento—ethics committee (protocol No. A957) on March 20, 2024, as part of the “Digital Health and AI” project funded by the Autonomous Province of Trento (resolution no. 2475 of 22/12/2022).

This study took place from April 2024 to March 2025 in the Autonomous Province of Trento, Italy.

Clinical team and coordination

The clinical team consisted of five primary care paediatricians and four paediatric ophthalmologists from the Ophthalmology Unit of Rovereto and Trento Hospitals, which are the two main hospitals of the Province of Trento healthcare trust—Azienda Provinciale per i Servizi Sanitari (APSS) di Trento. These professionals were responsible for conducting the infant eye screenings and evaluating the digital IRRT images. Oversight and coordination of the study were managed by a dedicated project manager from TrentinoSalute4.0, who supervised implementation and ensured adherence to the study protocol. A statistician from the Clinical and Evaluative Epidemiology Service (APSS Trento) conducted the data analysis and provided methodological support.

Study population

Eligible participants were infants aged 1 to 6 months whose families resided in the Autonomous Province of Trento and who were attending routine health visits with their primary care paediatrician. The inclusion criterion was written informed consent from a parent or legal guardian, including consent for data processing. The exclusion criteria included children with partial or complete blepharoptosis and those without a legally authorized caregiver present at the visit.

Screening procedure

Primary care paediatricians performed standard RRT using a direct ophthalmoscope on all infants. Additionally, paediatricians acquired digital IRRT images using the 2Win videoscreener (Adaptica Padua, Italy, software version 5.6.2), saving the images in JPEG format. The device was handheld by the operator in a dimly lit room at a distance of one metre from the child. Bright, colour fixation lights and audible stimuli could be switched on to attract the infant’s attention binocularly. Some infants underwent multiple assessments over time (e.g., at 1, 3 and 6 months of age). (See Supplemental Video)

Videoscreener description

The 2Win videoscreener is a portable, binocular infrared autorefractor and vision analyzer designed for rapid, non-invasive screening of refractive errors and ocular abnormalities, especially in infants and non-cooperative children.

Main features are:

• Binocular refraction: simultaneously measures both eyes at 1-metre distance in mesopic conditions, detecting myopia, hyperopia, astigmatism, anisometropia (range of refraction: sphere ± 15D, cylinder ±5D).

• Automated functions: fully automatic pupil detection (4–11 mm), pupillometry, dynamic pupil response to light stimuli, head tilt and gaze deviation measurement.

• Infant-friendly: no eye drops needed, uses bright fixation targets and auditory stimuli, saves JPEG images for teleophthalmology.

• Technical specifications: handheld (840 g), battery-operated, WiFi/USB connectivity, 3.5’’ display, operates in 3 s per exam.

Data collection and management

Anamnestic data (Table 1) were collected by paediatricians through the Infantia Electronic Health Record (EHR) System, which was integrated with the HealthMeeting telemedicine platform specifically for this project to facilitate seamless data transfer and teleconsultation workflows. For infants with negative (normal) outcomes, paediatricians completed the anamnesis and uploaded the IRRT images to the Infantia EHR, which then transmitted the data automatically to the local health care trust (APSS) repository via the HealthMeeting platform. In cases with doubtful or positive (abnormal) outcomes, paediatricians submitted teleconsultation requests, including clinical information and images. These were securely forwarded through interoperable features to the APSS HealthMeeting system, which ensures data integrity, confidentiality, and secure transmission. Paediatric ophthalmologists reviewed the submitted cases and issued formal reports. Infants with clearly abnormal findings were referred for in-person ophthalmology evaluation within ten days. Figure 1 illustrates the organizational workflow supporting this teleconsultation process, while Fig. 2 presents infrared reflex testing (IRRT) digital images demonstrating normal, abnormal, false-positive, and inconclusive screening results.

Table 1.

Summary of anamnestic data collected via the Infantia Electronic Health Record System.

Family history of ocular conditions

Presence/absence of:

Amblyopia

Strabismus

Nystagmus/Low vision

Congenital cataract

Congenital glaucoma

Corneal opacities

Retinoblastoma

Drugs during pregnancy

Systemic history

Presence/absence of:

Dystocic birth

Prematurity (≤ 32 SG, weight ≤ 2000 gr)

Syndromes

Metabolic diseases

TORCHS infections

Current eye problems

Presence/absence of:

Vision problems

Strabismus

Anomaly on inspection (leukocoria, nystagmus)

Red reflex testing doubtful or positive

The table details the presence or absence of family history of ocular conditions, maternal use of drug during pregnancy, systemic history factors, and current eye problems observed in the infant (including red reflex testing doubtful or positive result).

Fig. 1.

Workflow diagram illustrating the teleconsultation process in infant eye screening. Data entry and image upload by the paediatrician occur via the Infantia Electronic Health Record System. Teleconsultation requests, including clinical data and IRRT images, are transmitted to ophthalmologist through the APSS HealthMeeting Teleconsultation Platform, enabling structured specialist review and determination of outcomes -negative report issuance or referral for in-person examination based on doubtful or positive findings-.

Fig. 2.

Infrared reflex testing (IRRT) images obtained with the 2Win videoscreener in infants. The device projects an IR-light beam into both eyes simultaneously while an IR camera captures the reflected light from the retina: a monochrome gray reflex appears in the pupil. The normal IR images display different luminance saturation profiles due to retinal pigmentation and refractive state of the eye. A normal IRRT image (A) shows a gray background in the pupil and a central white fixation point (corneal reflex). Abnormal images (B-D) display media opacities as dark lines, dots, or plaques against the gray background. (E) An inconclusive examination: blurred image and gaze deviation.

Statistical analysis

Statistical analysis was conducted using Sas System 9.4 software (SAS Institute Inc., Cary, NC, USA).

Agreement between IRRT and RRT was evaluated using Gwet’s agreement coefficient (AC1), a robust alternative to Cohen’s kappa, which is less sensitive to marginal probabilities. AC1 was originally introduced in 2001 by Gwet, who demonstrated that AC1 overcomes kappa’s limitations related to sensitivity to marginal probabilities and provides a more robust measure of agreement.

The analysis also included the rate of inconclusive IRRT results, such as blurred images or those with gaze deviation.

Paediatricians’ qualitative feedback

To assess feasibility, all five paediatricians completed a structured 29-item questionnaire covering training, device usability, teleconsultation workflow, technical challenges, clinical relevance and workflow impact. The English-translated items of the questionnaire are reported in Table 2.

Table 2.

English translation of the structured questionnaire completed by paediatricians, assessing the usability, training, workflow integration, technical challenges, and perceived impact of the digital IRRT and teleconsultation process.

Section	Question
2Win training	How would you rate your learning curve for using the videoscreener?
	Were the instructions and training for using the device clear and sufficient?
	How easy or difficult did you find the use of the device overall?
2WIN technical and operational issues	What technical or operational difficulties did you encounter when examining children aged 1–6 months? (Select all that apply)
	On average, how much time was needed to acquire a suitable image for a single patient after becoming familiar with the instrument and setting?
	Did you experience difficulties saving images in JPEG format on the device or PC?
	How satisfied are you overall with using the device?
Clinical value	Compared to the traditional red reflex test (RRT), how relevant do you find having digital imaging available?
	How useful is digital imaging for clinical documentation and patient follow-up?
	How useful is digital imaging for medico-legal purposes?
Preferences	Which method do you currently prefer for infant eye screening?
	Do you think future implementation of an AI algorithm to support exam interpretation would be relevant?
Teleconsultation experience	How easy or difficult was it to request a teleconsultation?
	Did you encounter difficulties transferring image/data files from your PC to the Infantia EHR?
	Did you encounter technical difficulties using the integrated teleconsultation module (e.g., uploading images, viewing fields, sending requests)?
	On average, how much time did the process of saving and transferring files for a single patient require?
Clinical impact	Were the specialist’s response times compatible with your clinical activity?
	How do you rate programming times for subsequent in-person paediatric ophthalmology visits for positive cases?
	Would you like to request teleconsultations for further specialist clinical areas?
Workflow impact	Did using the videoscreener increase the average paediatric visit time?
	Was the complete procedure (exam + teleconsultation management) compatible with your office’s timing and organization?
	Do you think the time cost of managing the entire procedure is justified by the clinical and preventive benefits?
	Do you think the project (device + teleconsultation) had a positive impact on vision prevention?
Acceptance and future use	How did parents receive the screening proposal?
	Would you like to continue using digital vision screening tools (videoscreener) in your daily practice?
	Do you think the digital IRRT screening + specialist teleconsultation could be extended to all paediatricians regionally?
	What would be needed, in order of priority, to make the project more effective and sustainable? List up to 3 points.
	Do you have other comments, suggestions, or concerns?

Results

A total of 225 families provided informed consent to participate in the study. Thirty-six infants (16%) were excluded from the analysis because of the absence of JPEG images, images available only in PDF format, missing consent or incomplete anamnestic data, and infants falling outside the defined age range. Data from 189 eligible infants were included in the final analysis.

Study population characteristics

The descriptive characteristics of the study population are presented in Table 3. The mean age was 3.9 months (SD = 1.8), with a median age of 3 months. Males accounted for 56.6% of the cohort, resulting in a male-to-female ratio of 1.30. Ethnically, most infants were Caucasian (86.8%), with smaller representations from African (8.5%), Asian (3.7%), and South American (1.1%) groups. The family history of ocular conditions, systemic disease, and current eye problems was predominantly negative. Maternal use of drugs during pregnancy was reported in 1.1% of the cases.

Table 3.

Descriptive characteristics of the study population (N = 189).

Variable		Frequency n (%)
Gender	Male	107 (56.6%)
	Female	82 (43.4%)
M/F ratio	1.30
Age (months)	1	28 (14.8%)
	2	1 (0.5%)
	3	76 (40.2%)
	4	7 (3.7%)
	5	10 (5.3%)
	6	67 (35.4%)
Ethnicity	Caucasian	164 (86.8%)
	African	16 (8.5%)
	Asian	7 (3.7%)
	South-American	2 (1.1%)
Family history of ocular conditions	Positive	11 (5.8%)
	Negative	178 (94.2%)
Systemic history	Positive	2 (1.1%)
	Negative	187 (98.9%)
Current eye problems	Positive	6 (3.2%)
	Negative	183 (96.8%)
Maternal use of drugs	Yes	2 (1.1%)
	No	187 (98.9%)

The table summarizes demographic and clinical variables.

Agreement between RRT and IRRT

The agreement between the two screening methods (IRRT and RRT) is summarized in Table 4. The Gwet agreement coefficient (AC1) was 0.994 (95% confidence interval: 0.983–1.000), indicating almost perfect agreement. The inconclusive IRRT rate, comprising blurred images or images with gaze deviation, was also recorded (5.3%).

Table 4.

Agreement between digital Infrared reflex testing (IRRT) and Red reflex testing (RRT).

	RRT +	RRT -	RRT inconclusive
IRRT +	2	1	0
IRRT -	0	176	0
IRRT inconclusive	0	10	0
Statistical analysis	Estimate	Standard error	95% confidence intervals
Gwet’s agreement coefficient (AC1)	0.9943	0.0058	0.983	-1.000

The contingency table presents counts of positive, negative, and inconclusive results for both testing modalities. Gwet’s agreement coefficient (AC1) of 0.994 (95% CI 0.983–1.000) demonstrates almost perfect agreement between IRRT and RRT.

Qualitative feedback from primary care paediatricians

All five participating paediatricians completed a structured questionnaire assessing feasibility aspects of the digital IRRT screening and teleconsultation workflow; results indicated generally positive operator experiences while highlight operational challenges affecting feasibility.

Operators rated their learning curve as relatively moderate, with clear instructions and overall good ease of use for the videoscreener device; satisfaction with the device was generally high, especially regarding ease of image acquisition and use (average time: 2–3 min). Common technical difficulties included dimmed lighting, child movement, fixation issues, and prolonged image acquisition times while saving and transferring images mostly posed no major issues, although some difficulties with the teleconsultation module and file transfer were reported.

The relevance and usefulness of digital IRRT imaging compared with traditional RRT were rated moderate to high, in fact paediatricians found the presence of digital images extremely important for clinical documentation, medico-legal validity, and follow-up; there was a mixed preference for traditional versus digital screening and moderate enthusiasm for the addition of AI interpretation support.

Teleconsultation processes were deemed somewhat compatible with clinical schedules but increased visit times and negatively affected workflow in some settings; specialist response times and scheduling were mostly adequate, but further workflow optimization was desired. Despite challenges, there was strong support for the continued use of digital screening devices and program expansion, conditional on improved device integration and software.

Parental acceptance of the screening protocol was consistently positive.

Discussion

This study aimed to evaluate the usefulness of digital infrared reflex testing (IRRT) and to estimate the feasibility of asynchronous teleconsultation between primary care paediatricians and ophthalmologists for infant eye screening. There is a growing need for an objective, user-friendly eye screening tool capable of detecting anterior segment abnormalities—including congenital cataracts—with high sensitivity and specificity. IRRT, which uses IR light already integrated into nonmydriatic fundus cameras and paediatric videoscreeners, offers a digital alternative to traditional ophthalmoscope-based red reflex testing (RRT)^11,12. Prior studies have demonstrated the superior diagnostic accuracy of IRRT, especially in settings with limited examiner experience or diverse patient populations. In 2019, Duret et al. demonstrated that smartphone-based IRRT has greater diagnostic accuracy than standard RRT when performed by inexperienced screeners, particularly when patients are non-Caucasian children⁷. An ongoing pilot study (clinicaltrials.gov: NCT05282147) by Allen et al.^13,14 investigated whether IR image evaluation using a prototype infrared digital camera (Neocam) improved sensitivity and specificity compared with those of conventional RRT in newborn eye screening. Consistent with these studies, we found near-perfect agreement between digital IRRT and standard RRT among infants aged 1–6 months in a primary care setting, with a low percentage (5.3%) of inconclusive results.

Different commercially available paediatric videoscreeners use IR light to assess anterior segment opacities (images of blocking of the IR light at the level of the cornea, lens, and anterior vitreous), refractive errors (dynamic IR retinoscopy), and strabismus (deviation of the corneal reflex) according to age-based screening criteria published in 2013¹⁵ which defined refractive risk factors (hyperopia, astigmatism, myopia, and anisometropia) and non-refractive risk factors (strabismus and media opacity). In 2021, the AAPOS Vision Screening and Research Committees updated instrument-based paediatric vision screening guidelines, including media opacities, pupillary and eyelid abnormalities stating that any media opacity greater than 1 mm is considered potentially amblyogenic and should be detected through photo- or video-refractive screening¹⁶. The 2Win infrared videoscreener used in the study integrates age-appropriate screening tables (AAPOS 2021 guidelines), enabling, at 6 months, the combination of digital IRRT with video screening for strabismus and refractive errors; several referrals were made from the five paediatricians to ophthalmologists for high refractive errors and strabismus based of these findings^17,18. Despite these benefits, the 2Win videoscreener showed limitations including low image resolution (JPEG image: 936 × 289 pixels, DPI 96, 32 bit), long acquisition distances (1 m), and operator learning challenges (camera focus); suggested improvements include alternative shorter imaging distances for younger infants (e.g., 45 cm for newborns and infants up to 3 months of age), faster image acquisition, higher resolution cameras, and an additional visible green light source to enhance posterior segment screening (screening for retinoblastoma).

In this study, compared with standard RRT, paediatricians identified several key strengths of digital IRRT, including objective photographic documentation supporting clinical and medico-legal needs, better performance across varying pigmentation and pupil sizes, seamless telemedicine integration, and suitability for computer-assisted diagnostics, highlight the promising future application of this technology. Teleconsultation—the synchronous or asynchronous digital exchange of clinical information—was feasible in 84% of the cases and was facilitated by the integration of the Infantia EHR and HealthMeeting telemedicine platform. Paediatricians valued faster diagnostic decisions, structured reporting, and coordinated in-person referrals, although the challenges included time-consuming data entry, consent procedures, and interoperability issues. Ongoing national-level investments are expected to address regional disparities and technical fragmentation.

Teleconsultation facilitates diagnostics or therapeutic collaboration between health care professionals (physicians and/or nurses); Kolsoum et al. 2016 systematic review¹⁹ focusing on decision-making, technologies, specialties, temporal/geographical distributions, medical fields, outcomes, and structures, found that fewer than 15% of studies involved paediatrics, and surgery, general medicine, cardiology, oncology, orthopaedics were the most frequent topics. Consultations were specialist–specialist or more commonly generalist–specialist, directly influencing diagnosis/ treatment plans in 45% of studies.

Overall, digital IRRT screening combined with teleconsultation demonstrated good feasibility as a clinical tool, with a positive impact on visual prevention and parent acceptance, but key improvements are recommended in device design, image transfer, environmental conditions (e.g., room lighting), and streamlined teleconsultation to enhance effectiveness and scalability in regional paediatric care.

Artificial intelligence (AI) holds promise for further enhancing paediatric ophthalmology screening by enabling the automated interpretation of images and risk scoring²⁰; in congenital cataract screening, machine learning (ML) and deep learning (DL) have demonstrated experts-level diagnostic accuracy in related domains like slit-lamp analysis²¹ including risk factors ranking/ quantification²², while Tripathi proposed a novel near-infrared algorithm for adult cataract in 2022²³.

Our ongoing research aims to combine digital IRRT images with clinical data to develop predictive AI models for the early identification of infants at risk of anterior segment pathologies.

Limitations

This study excluded newborns, the primary target population for congenital cataract screening. Additionally, only a small subset of primary care paediatricians in the region participated (only 5 out of 66 primary care paediatricians from the Autonomous Province of Trento participated in the teleconsultation pilot), potentially impacting generalizability. Although the results of this study strongly agreed between digital IRRT and ophthalmoscope-based RRT, full validation of IRRT against a paediatric ophthalmological examination, which is the gold standard, remains necessary.

Conclusion

Digital IRRT demonstrated high agreement with standard RRT and offers significant advantages, including reproducible image documentation, telemedicine workflows integration, and AI-assisted diagnostics suitability, while non-real-time teleconsultation between primary care paediatricians and ophthalmologists was feasible with appropriate training and technological support, paving the way for scalable infant eye screening programs. Overall, this study supports a paradigm shift towards digital, telemedicine-enabled models that improve access, accuracy, and efficiency, with significant implications for practice guidelines and health policy.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Racano Elisabetta, Spinelli Chiara, Sacco Giuseppina, Benedetti Giorgia, Romanelli Federica, Compagnone Adele, Betta Marta, Pasquini Roberta, Filippi Lorena, Pavanello Lucia, Malfatti Giulia, Pertile Riccardo. The draft of the manuscript was written by Racano Elisabetta and Malfatti Giulia and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Written informed parental consent for publication of infrared images in scientific journals and dissemination channels was obtained and documented for all participants whose images are included in Fig. 2.

Funding

The study is part of the “Digital Health and AI” project funded by the Autonomous Province of Trento (resolution no. 2475 of 22/12/2022).

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki. The study protocol (No. A957) received ethical approval from the APSS Trento Ethics Committee on March 20, 2024.

Consent to participate

Written informed consent was obtained from the parents (or legal guardian) of all children involved in the study.