RESUMO
PURPOSE: We aimed to evaluate the feasibility of using a novel device, the Smart Eye Camera (SEC), for assessing tear meniscus height (TMH) after fluorescein staining and the agreement of the results with measurements obtained using standard slit lamp examination. METHODS: TMH was assessed using both SEC and conventional slit lamp examination. The images were analyzed using the software ImageJ 1.53t (National Institutes of Health, Bethesda, MD, USA). A common measurement unit scale was established based on a paper strip, which was used as a calibration marker to convert pixels into metric scale. A color threshold was applied using uniform parameters for brightness, saturation, and hue. The images were then binarized to black and white to enhance the representation of the tear menisci. A 2 mm area around the upper and lower meniscus in the central eye lid zone was selected and magnified 3200 times to facilitate manual measurement. The values obtained using SEC were compared with those obtained with a slit lamp. RESULTS: The upper and lower TMH values measured using the SEC were not statistically different from those obtained with a slit lamp (0.209 ± 0.073 mm vs. 0.235 ± 0.085, p = 0.073, and 0.297 ± 0.168 vs. 0.260 ± 0.173, p = 0.275, respectively). The results of Bland-Altman analysis demonstrated strong agreement between the two instruments, with a mean bias of -0.016 mm (agreement limits: -0.117 to 0.145 mm) for upper TMH and 0.031 mm (agreement limits: -0.306 to 0.368 mm) for lower TMH. CONCLUSIONS: The SEC demonstrated sufficient validity and reliability for assessing TMH in healthy eyes in a clinical setting, demonstrating concordance with the conventional slit lamp examination.
RESUMO
In ophthalmology, the availability of many fundus photographs and optical coherence tomography images has spurred consideration of using artificial intelligence (AI) for diagnosing retinal and optic nerve disorders. However, AI application for diagnosing anterior segment eye conditions remains unfeasible due to limited standardized images and analysis models. We addressed this limitation by augmenting the quantity of standardized optical images using a video-recordable slit-lamp device. We then investigated whether our proposed machine learning (ML) AI algorithm could accurately diagnose cataracts from videos recorded with this device. We collected 206,574 cataract frames from 1812 cataract eye videos. Ophthalmologists graded the nuclear cataracts (NUCs) using the cataract grading scale of the World Health Organization. These gradings were used to train and validate an ML algorithm. A validation dataset was used to compare the NUC diagnosis and grading of AI and ophthalmologists. The results of individual cataract gradings were: NUC 0: area under the curve (AUC) = 0.967; NUC 1: AUC = 0.928; NUC 2: AUC = 0.923; and NUC 3: AUC = 0.949. Our ML-based cataract diagnostic model achieved performance comparable to a conventional device, presenting a promising and accurate auto diagnostic AI tool.
