Abstract
Purpose :
<p style="margin:0cm;margin-bottom:.0001pt"><span style="font-family:helvetica,sans-serif; font-size:8.5pt; margin:0px"><font color="#000000">To determine Minimal Image Gain (MIG) in patients with vitreous hemorrhage (VH), and to compare variability between two non-experienced graders, and compare variability between two different ultrasound systems.</font></span></p>
Methods :
Retrospective and comparative study of patients with VH assessed at the Ocular Ultrasound Department. MIG was determined using two systems, the Aviso S and Cine Scan. A 10 mega Hz B-scan ultrasound probe was used to evaluate the macular region, longitudinal section was made at 9 o’clock meridian for right eyes and 3 o’clock for left eyes. To obtain MIG the following method was proposed: The image was observed without additional zoom, and with the highest amount of gain in db. The scan's gain is lowered in orden to obtain a completely “echo-silent” vitreous cavity. This point is termed MIG and the db registered. The procedure was done by the two graders and in the two systems for every patient.
Intraclass Correlation Coefficient (ICC) was performed to determine inter-observer variability between graders and between systems.
Results :
Sixteen eyes of 16 patients were analyzed. Mean MIG for Grader 1 in the Aviso: 61 db (47-85); Mean MIG for Grader 2 in the Aviso: 60 db (46-84); Mean MIG for Grader 1 in the Cinescan: 56 (32-78); Mean MIG for Grader 2 in the Cinescan: 57 (40-83). ICC for Grader 1 vs 2 in the Aviso: 0.876 for single measures (p<0.000), 0.934 for average measures (p>0.000), Cronbach’s Alpha 0.932. ICC for Grader 1 vs 2 in the Cinescan: 0.889 for single measures (p<0.000), 0.941 for average measures (p<0.000), Cronbach’s Alpha 0.941. ICC for Aviso vs Cinescan: 0.852 for single measures (p<0.000), 0.920 for average measures (p<0.000), Cronbach’s Alpha 0.949.
Conclusions :
Conclusions: MIG measurements can be accurately performed in patients with VH. There is high agreement even in non-experienced graders and between systems.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.