Age-stratified quantitative ultrasound biomicroscopy image analysis of iris thickness using multivariable analysis to optimize reliability

Abstract

Purpose : Normative data for human iris parameters is not well established, resulting in a lack of understanding of the association between quantitative iris features and presence of intraocular disease. The purpose of this study was to quantify iris thickness of pediatric subjects in ultrasound biomicroscopy (UBM) images using an optimized image processing protocol with high repeatability and reliability of measurements.

Methods : Eligible participants’ parents were offered consent and subjects were enrolled prior to imaging. 86 images from 26 eyes in 14 healthy pediatric subjects (mean age 2.0+/-1.1yrs, median 1.8yrs, range 0.2-4.1yrs) were analyzed according to a prospective protocol. Two trained observers measured iris thickness on two axial and two longitudinal UBM images per eye in raw format and with image processing using edge detection. Image quality was graded by a trained ultrasonographer. Logistic regression was used to determine association between image features and consistency of measurements. Intraclass correlation coefficient (ICC) between observers was the outcome of interest. Covariates of interest were orientation (axial vs longitudinal), quality (good vs excellent), and processing (raw vs edge detection). Covariates contributing to improved ICC were used in UBM analysis to determine iris thickness stratified by age. Mean values were determined for ages <1yrs, 1-2yrs, and >2yrs. ANOVA was performed to see if there were statistically significant differences in thickness measurements based on age.

Results : Normative data for iris parameters were found to differ significantly by age strata (0-1yrs, 1-2yrs and 2-5yrs). Maximal iris thickness increased with age (p<0.001) while minimal and central iris thickness did not vary by age. ICC for these parameters ranged from 0.70-0.97. Preliminary results of multiple linear regression suggest that orientation and image processing were associated with improved ICC. Image quality was not significantly associated with ICC.

Conclusions : Iris measurement consistency is improved by use of longitudinal image orientation and edge detection in image processing. Normative data based on the optimized image analysis protocol suggests that iris thickness varies significantly with age.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.

 

Figure 1: Minimal (ID1), central (ID2) and maximal (ID3) iris thickness

 

Figure 2: A) Axial and B) longitudinal UBM with C,D) edge detection