Optical coherence tomography (OCT) constitutes a non-invasive and well-tolerated imaging modality leveraging low coherence light reflected by ocular tissues to generate high-resolution images (1). This technique empowers clinicians to produce three-dimensional (3D) renderings of intraocular structures in vivo (2, 3). Beyond subjective qualitative assessment, these images lend themselves to objective analysis, facilitating quantitative measurements such as stratified retinal thickness across distinct retinal zones (4, 5). With advancements in image resolution, quality, and software capabilities, the quantitative scrutiny of OCT images has undergone significant refinement (6, 7). The faster scanning speed and higher imaging quality allow it to be performed also in patients with poor cooperation (e.g., children) (8–10). Although the accuracy and reproducibility of OCT applications in pediatric contexts have undergone scrutiny, empirical evidence validating its clinical utility across diverse demographics remains lacking (11).

Precise and highly reproducible quantitative measurements serve as indispensable diagnostic and monitoring tools, guiding optimal treatment decisions across routine medical care and clinical trials alike. However, the variance in measurement values among different OCT instruments underscores the need for standardized algorithms for image recognition and segmentation (12). The application of OCT in pediatric cohorts introduces additional intricacies, owing to factors such as limited patient cooperation, diminutive ocular dimensions, and the imperative of precise image alignment and interpretation (11, 13). Despite these hurdles, OCT holds immense promise as a non-invasive, objective tool for delineating ocular morphology and pathology in children. A nuanced understanding of the determinants of successful OCT imaging and the factors shaping image quality is imperative for refining screening protocols, augmenting diagnostic precision, and guiding public health initiatives geared toward pediatric ocular wellness.

The Lhasa Childhood Eye Study (LCES) represents a pioneering effort, aiming to investigate the clinical findings of subject characteristics and factors influencing image quality in Tibetan children using OCT within an epidemiological cohort setting. The unique demographic and environmental characteristics of the Tibetan population, coupled with the challenges inherent in conducting large-scale pediatric ocular screenings in resource-limited settings, underscore the significance of this study. Through a comprehensive approach encompassing clinical examination, OCT imaging, and systemic evaluation, this study seeks to elucidate the epidemiology of ocular conditions, identify risk factors for poor image quality in OCT, and inform strategies for improving the efficiency and effectiveness of pediatric ocular screening programs.

Focusing on a large cohort of minority children, this study shed light on the clinical findings of subjects characteristics and image quality related factors in Tibetan children by OCT. In this study, we used manual segmentation to correct errors in automated macular and optic-disc segmentation to better reflect the anatomy represented in each scan. This study found that: (1) high analyzable fraction (91.5%) after exclusion of incomplete protocols or examination procedures; (2) the most significant causes of image failure are image defects and shifts caused by blinking and poor fixation stability in children; (3) multivariate analysis identified suburban residence, lower near VA, and male gender as significant predictors of worse image quality; and (4) differences in retinal thickness and optic disc parameters were observed pre and post-adjustment, with more significant changes in the optic disc area.

This study found that the response rate and coordination of Tibetan children for completing the OCT examination were relatively higher compared to other previous studies. In the Anyang Childhood Eye Study, about 7% of children aged 12 years old had failed images (18). In the Sydney Childhood Eye Study, OCT scans failed in about 21% of children aged 6 years (19). Although our study was conducted among Tibetan children and most 7-year-old Tibetan children are still unable to use Chinese to communicate, we still obtained a pass rate of nearly 85% of the scanned macular area. While this attrition rate is relatively low, it underscores the challenges inherent in longitudinal studies involving young children. Previous studies of screening with the Topcon device have shown an ineligible rate of about 20% and a scan rejection rate of about 2%–5% (16). Also the optic disc has a higher rejection rate than the macula. The primary reasons for image exclusion included defects related to blinking or poor fixation stability, image blurring, and uncorrectable positioning errors. This highlights the importance of rigorous quality control measures in ensuring reliable data acquisition.

In our investigation, a noteworthy correlation emerged between children’s near VA and the quality of OCT scans. Notably, children with superior near VA exhibited higher OCT imaging quality. Healthy children typically demonstrate enhanced ocular accommodation, and the majority of subjects in our study boasted near VA within the normal range. This phenomenon might be attributed to the proximity of internal optic markers of OCT to the eye, whereby myopic acuity can influence children’s fixation on these markers, thereby affecting the speed and quality of OCT scans. Previous research often links fixation stability with anisometropia, high myopia, strabismus, and amblyopia, yet the association with near vision remains largely unexplored (20, 21). Our study’s distribution of near visual acuity largely fell within the normal range compared to past studies, suggesting that a child’s near vision may exert a substantial influence on the quality of OCT examinations, particularly when the fixation optic marker is situated nearby (22, 23). Despite the absence of standardized clinical guidelines for near visual acuity measurement, its significance in determining the rehabilitation of visually impaired children cannot be overstated, underscoring the necessity for further exploration of the association between fixation stability and near visual acuity.

Furthermore, our findings reveal a significant association between the quality of OCT scans and the educational locale and gender of the children. Notably, girls demonstrated significantly higher proficiency in completing OCT scans with qualified images compared to boys, aligning with common perceptions that boys tend to exhibit more rambunctious behavior and are less inclined to cooperate during examinations. Moreover, our study observed that children attending schools in suburban areas underwent more OCT scans necessitating deletion and adjustment. Interestingly, despite exhibiting greater discipline during examinations, children from suburban schools often face language barriers due to their lower proficiency in Chinese, potentially hindering their performance in examinations. This study also revealed that lower image quality is associated with higher intraocular pressure and heart rate. Non-contact tonometry, due to its air-puff mechanism, often yields elevated intraocular pressure readings in children, as blinking or eyelash obstruction commonly interferes with measurements. Similarly, higher heart rates may be linked to the heightened activity levels typical of children. Therefore, increased intraocular pressure and heart rate may indicate that children are more active, sensitive, and less cooperative during examinations, ultimately resulting in reduced image quality.

Recent research underscores the reproducibility and clinical utility of OCT in pediatric populations (24–26). Our study reveals that OCT examinations conducted post-dilation in school-age children, coupled with the software’s automated segmentation calculations, yield informative results. However, automatic positioning segmentation errors contribute to thinner measured retinal thicknesses, consistent with prior investigations (27–29). Literature suggests that erroneous positioning of the optic disc or macular region by OCT often leads to deviations from normal values (30, 31). Topcon’s normal population database highlights the necessity for manual adjustments to rectify automatic segmentation errors, including grid locations, boundary segmentations, and disc margins with Fastmap, executed by clinical experts (16). Such errors may stem from inherent flaws in the machine algorithm. By manually repositioning the measurement region to its correct anatomical location, more accurate measurement values can typically be obtained. Our study also underscores that in the absence of obvious artifacts or image loss, most images can yield relatively valid measurement values through proper repositioning of the measurement area. This underscores the importance of manual supervision and adjustment in epidemiological screening or clinical practice to ensure the accuracy and authenticity of OCT scan results.

This study possesses several strengths as well as limitations. On one hand, the sample size is considerable, the examination protocols are comprehensive, the dataset is comprehensive, and the sample demographics are robust, particularly targeting minority children. However, the investigation solely examined the overall retinal thickness in the macular and optic disc regions, without delving into the perfection of OCT’s layered functionality. Furthermore, there exists a degree of subjectivity in the manual segmentation process of OCT scans. Additionally, the study predominantly included healthy children, warranting further exploration of OCT scanning outcomes in diverse retinal pathologies.

In conclusion, completion rates of OCT examinations for large-scale pediatric eye screening can reach high levels, and their completion rates are related to educational location, gender, visual acuity, and amblyopia; worse near vision significantly affects OCT image quality and completion rate; OCT scans in pediatric subjects pose greater challenges for alignment in the optic disc area compared to the macular region, and retinal thickness in both regions tends to be thinner than normal in the automated segmentation computations of the OCT apparatus. These discoveries enrich our comprehension of the factors influencing image quality in pediatric OCT imaging and underscore the imperative of rigorous quality assurance measures in large-scale epidemiological investigations.