Related Experiment Video
Updated: May 2, 2026
Software-Assisted Quantitative Measurement of Osteoarthritic Subchondral Bone Thickness
Published on: March 18, 2022
Reproducibility and within-subject variability of HRK-9000A meibography in normal young participants
Liat Gantz1, Judith Wilks1, Reut Ifrah1
1Department of Optometry and Vision Science, Hadassah Academic College, Jerusalem, Israel.
Purpose:
Reliable assessment is critical for diagnosing and managing meibomian gland dysfunction. Multi-functional diagnostic devices, such as meibographers, streamline clinical workflows by integrating multiple ocular assessments. Ensuring reproducibility across examiners is vital for accurate diagnosis and monitoring of treatment. This prospective study determined the inter-examiner reproducibility (IER), inter-rater reproducibility (IRR) and within-subject variability (WSV) of the Huvitz HRK-9000A meibographer.
Methods:
Meibomian glands (MGs) of both eyelids of healthy participants were captured during the same session by Examiner 1 (E1) and Examiner 2 (E2) in a counter-balanced design. The images were rated offline by E1, E2 and an Independent rater (R1). The inter-examiner reproducibility (IER between E1 vs. E2) and inter-rater reproducibility (IRR between E1 vs. R1 and E2 vs. R1) were determined based on non-parametric Bland-Altman plots, intraclass correlation coefficients (ICCs) and weighted Kappa (κ) values. Within-subject variability (WSV) was determined using ICCs.
Results:
Mean MG loss of the upper (E1: 1.0 ± 0.8 vs. R1: 0.9 ± 0.8, E2: 1.2 ± 0.8 vs. R1: 0.9 ± 0.7) and lower eyelids (E1: 1.9 ± 0.9 vs. R1: 2.1 ± 1.1, E2: 1.5 ± 1.0 vs. R1: 1.8 ± 1.0) of 35 participants (mean age 22 ± 3 years, range 19-30) was significantly correlated (p < 0.001) for all pairs of examiners (E1-E2, E1-R1 and E2-R1), but was significantly different for the upper eyelids of E2 vs. E1 (p < 0.002) and E2 vs. R1 (p < 0.003). Median differences between pairs of raters for both eyelids were close to zero, with >74% of comparisons falling within the interquartile range, except for the upper eyelids of E2 vs. E1 and E2 vs. R1 (63% and 66%, respectively). IRR demonstrated moderate agreement (0.43-0.57) with good reliability (ICC range 0.76-0.86) for the upper eyelids and good agreement (0.60-0.65) with good reliability (ICC range 0.85-0.88) for the lower eyelids. The WSV was excellent (ICCs for repeated measurements > 0.90).
Conclusions:
The HRK-9000A meibographer demonstrated excellent reliability, with good inter-examiner and inter-rater reproducibility. It is suitable for meibographic assessment, follow-up or treatment.
