A proposed method to quantify vitreous hemorrhage by ultrasound

Abstract

Purpose: To propose a method for quantification of vitreous hemorrhages (VH) termed minimum image gain (MIG). Therefore, to obtain MIG measurements in patients with VH and to compare them to normal controls; to compare results between graders; and to obtain and compare MIG from two different ultrasound systems. Methods: Retrospective and cross-sectional, observational, and comparative study performed in two parts: Part 1) Retrospective comparison ofMIG in VH vs controls by two experienced graders, with intra/inter-observer variability:MIG technique is described.MIG is performed retrospectively in two groups, VH patients and normal controls. Groups are compared with independent samples t-test. Intra- and interobserver variability between graders was obtained. Part 2) Cross-sectional analysis of variability from nonexperienced graders and with different ultrasound systems: MIG is performed in cross-sectionalmeasurements ofVHpatients, by two unexperienced graders, andwith two ultrasound systems. Interobserver variability and Bland–Altman plot with levels of agreement (LoA) were obtained. Results: Part 1: 50 patients with VH resulted in mean MIG: 52.8 dB; 34 controls resulted in mean MIG: 77.97 dB. Independent samples t-test resulted in a statistical significant difference. Intra- and inter-observer variability resulted in an almost perfect agreement between experienced graders. Part 2: 63 patients with VH, mean MIG: 56.19 dB. Inter-observer variability resulted in a very high agreement between unexperienced observers. LoA resulted in a statistical difference between the two ultrasound systems. Conclusion: MIG may provide an objective and reproducible way to quantify vitreous hemorrhage density and potentially any vitreous humor opacity. Agreement is high even with unexperienced graders. However, the two ultrasound systems analyzed may not be interchangeable.