The effects of irreversible JPEG compression on an automated algorithm for measuring carotid artery intima-media thickness from ultrasound images.

Our ultrasound practice has begun to investigate automated measurements of carotid artery intima-media thickness (IMT) as an indicator of subtle atherosclerosis. Since our clinical ultrasound images are irreversibly compressed, we investigated the effects of this compression on our IMT measurements. We obtained 10 ultrasound images of normal carotid arteries. These were compressed using JPEG to ratios of 5:1, 10:1, 15:1, 20:1, and 30:1. IMT measurements made from all compressed and uncompressed images were compared. For compression ratios ?10:1, IMT deviations between compressed and uncompressed images were ?0.03 mm. Higher than 10:1, the overall IMT deviations were small (0.01 +/- 0.04 mm), although one 25% deviation was measured. Comparison of other parameters yielded similar results. This initial study indicates that compression at 10:1 using baseline JPEG should have little effect on IMT measurements made using the current algorithm, and that compression to 20:1 or 30:1 may be feasible.