Automated Analysis of Anterior Chamber Inflammation by Spectral-Domain Optical Coherence Tomography

Purpose This study was designed to determine the feasibility of anterior segment optical coherence tomography (AS-OCT) to objectively image and quantify the degree of AC inflammation. Design Prospective evaluation of a diagnostic test. Participants Patients with anterior segment involving uveitis. Methods Observational case series of patients with uveitis. Single-line and 3-dimensional (3D) volume AS-OCT scans were manually graded to evaluate for the presence or absence of cells in the AC. Clinical grading scores were correlated to the number of cells seen in each line scan. An automated algorithm was developed to measure the number of cells seen in the 3D volume scan and compared with manual measurements and clinical grading scores. Main Outcome Measures Degree of anterior segment inflammation. Results A total of 114 eyes from 76 patients were imaged, 83 eyes with line scans and 31 eyes with volume scans. The average number of cells on line scans was 0.13 for grade 0, 1.2 for grade 1/2+, 2.6 for grade 1+, 5.7 for grade 2+, 15.5 for grade 3+, and 41.2 for grade 4+. Spearman correlation coefficient comparing clinical grade with the individual AS-OCT line scans was 0.967 ( P < 0.0001). The range of cells in the automated cell count of 3D volume scans was 13.60 to 1222; the range for manual cell counts was from 9.2 to 2245. The Spearman correlation coefficients were r = 0.7765 ( P < 0.0001) and r = 0.7484 ( P < 0.0001) comparing the manual and automated cell counts with the clinical grade, respectively. Spearman correlation coefficient comparing the automatic cell counts with manual cell count in the 3D volume scan was 0.997 ( P < 0.0001). Conclusions Anterior segment OCT can be used to image and grade the degree of AC inflammation. Clinical grading strongly correlates with the number of cells on AS-OCT line scans and volume scans. The automated algorithm to measure cell count had a high correlation to manual measurement of cells in the 3D volume scans. This modality could be used to objectively grade response to treatment.