Subjective and Objective Assessment of Meibography Images

Abstract

Purpose : To assess the reliability of subjective grading and objective assessment of meibography in patients with symptoms of dry eye disease (DED)

Methods : Infrared meibography images (OCULUS Keratograph® 5M) of 100 right eyes of 100 patients (41 females and 59 males; age 47.5±1.5 years, range 11 to 82) were analyzed by two experienced observers. Meiboscore of the upper lid (UL) and the lower lid (LL) were evaluated by subjective grading of the proportion of meibomian gland (MG) loss (grade 0:≤25%; grade 1: 26-50%; grade 2: 51-75%; grade 3: >75%). Subsequently, objective image analysis using ImageJ was performed to evaluate MG loss (100-grade scale) in the UL and LL, as well as the MG length,and tortuosity in the UL. Intraobserver and interobserver intraclass correlation coefficients (ICCs) were calculated.

Results : The meiboscore and MG loss in UL demonstrated good intraobserver repeatability and interobserver reproducibility (ICC≥ 0.90). The MG length in the UL, meiboscore and MG loss in the LL showed good repeatability (ICC≥0.90) and reproducibility (ICC≥0.88). The MG tortuosity had lower intraobserver repeatability (ICC= 0.88) and interobserver reproducibility (ICC= 0.76). The meiboscore in the UL correlated strongly with MG loss (r=0.85, p=<0.001), MG length (r=-0.70, p<0.001), and tortuosity (r=-0.57, p<0.001) in the UL. Moreover, it correlated weakly with meiboscore (r=0.30, p=0.002) and MG loss (r= 0.296, p=0.003) in the LL. The meiboscore in the LL correlated strongly with MG loss in the LL, but only weakly with MG loss (r= 0.254, p=0.011), MG length (r=0.253, p=0.011), and MG tortuosity (r=-0.248, p=0.013) in the UL.

Conclusions : Subjective meiboscore grading and objective assessment of MG loss and MG length were highly repeatable and reproducible. The four-level subjective meiboscore correlated strongly with objective MG loss and is effective for quick MG assessment.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.