Purpose:To quantify, by a novel software image analyzer, the tortuosity of retinal blood vessels in digital retinal images of infants at risk of retinopathy of prematurity (ROP); the value was compared to the ROP stage as clinically assessed. Methods:In one eye of 31 premature newborns with several ROP stages (11 with stage 1, 7 with stage 2, 12 with stage 3 requiring treatment) and of 18 premature newborns without ROP, followed at the Neonatal Intensive Therapy Unit of the University of Catania, the tortuosity of retinal arteries and veins was measured by a semi-automated method. In digital retinal images, obtained during routine visits for ROP with RetCam (RetCam130; Clarity Medical Systems Inc., Pleasanton, CA) and a 1300 lens, the vessel tortuosity was measured until a distance of 10 optic disk size (200 pixels) from the margin of the optic disk. Operator generated tracing of the vessels, and a specific software calculated a tortuosity index (as the ratio between the length of the vessel and that of the underlying chord minus 1) in superior and inferior temporal arteries and veins. The stage of ROP was evaluated by an expert ophthalmologist. Results:An increased tortuosity was found in all retinal vessels of patients with stage 3 ROP (ANOVA p<0.001, Tukey-Kramer p<0.05 vs. other groups). Retinal tortuosity index was < 0.05 in controls, <0.08 in stage 1, 0.16 in stage 3 ROP. Multivariate analisys showed that weight (p=0.004), tortuosity of inferior artery (p=0.016) and of superior vein (p=0.021) were related to the ROP stage. Factors allowing the prediction of the stage 3 ROP were established through a decision tree induction (with Chi-Squared Automatic Interaction Detection -CHAID- segmentation technique). These are: a tortuosity index value greater than 0.9 in superior vein (all patients with this value had stage 3 ROP, chi square p=0.000), and, in this group, a weight lower than 650 grams (10 of 13 patients, chi-square p=0.005). For stage 3 ROP, sensitivity was 0.769, specificity was 1.000, and AUC of ROC curve was 0.984. Conclusions:Patients with stage 3 ROP have increased tortuosity index of retinal vessels. This semiautomatic analysis can allow to identify patients with higher ROP stages requiring retinal treatment.
New method of analysis of tortuosity of retinal vessels in Retinopathy of Prematurity
LONGO, ANTONIO;FRANCO, Livio;Andrea Russo;REIBALDI, MICHELE
2012-01-01
Abstract
Purpose:To quantify, by a novel software image analyzer, the tortuosity of retinal blood vessels in digital retinal images of infants at risk of retinopathy of prematurity (ROP); the value was compared to the ROP stage as clinically assessed. Methods:In one eye of 31 premature newborns with several ROP stages (11 with stage 1, 7 with stage 2, 12 with stage 3 requiring treatment) and of 18 premature newborns without ROP, followed at the Neonatal Intensive Therapy Unit of the University of Catania, the tortuosity of retinal arteries and veins was measured by a semi-automated method. In digital retinal images, obtained during routine visits for ROP with RetCam (RetCam130; Clarity Medical Systems Inc., Pleasanton, CA) and a 1300 lens, the vessel tortuosity was measured until a distance of 10 optic disk size (200 pixels) from the margin of the optic disk. Operator generated tracing of the vessels, and a specific software calculated a tortuosity index (as the ratio between the length of the vessel and that of the underlying chord minus 1) in superior and inferior temporal arteries and veins. The stage of ROP was evaluated by an expert ophthalmologist. Results:An increased tortuosity was found in all retinal vessels of patients with stage 3 ROP (ANOVA p<0.001, Tukey-Kramer p<0.05 vs. other groups). Retinal tortuosity index was < 0.05 in controls, <0.08 in stage 1, 0.16 in stage 3 ROP. Multivariate analisys showed that weight (p=0.004), tortuosity of inferior artery (p=0.016) and of superior vein (p=0.021) were related to the ROP stage. Factors allowing the prediction of the stage 3 ROP were established through a decision tree induction (with Chi-Squared Automatic Interaction Detection -CHAID- segmentation technique). These are: a tortuosity index value greater than 0.9 in superior vein (all patients with this value had stage 3 ROP, chi square p=0.000), and, in this group, a weight lower than 650 grams (10 of 13 patients, chi-square p=0.005). For stage 3 ROP, sensitivity was 0.769, specificity was 1.000, and AUC of ROC curve was 0.984. Conclusions:Patients with stage 3 ROP have increased tortuosity index of retinal vessels. This semiautomatic analysis can allow to identify patients with higher ROP stages requiring retinal treatment.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
