Purchase this article with an account.
CR Swanson, KD Cocker, KH Parker, MJ Moseley, SM E Wren, AR Fielder; Semi-automated computer analysis of vessel growth in Preterm Infants with ROP . Invest. Ophthalmol. Vis. Sci. 2002;43(13):1250.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Purpose: To quantify morphology in the principal retinal vessels of preterm infants before and during the development of retinopathy of prematurity (ROP). Methods: 76 retinal images were obtained from the left eyes of ten preterm infants (gestational age range: 24-29 weeks) undergoing regular screening examinations for ROP with a Retcam 120 digital fundus camera (Massie Labs, Dublin, CA). An independent examiner recorded the presence, location and severity of ROP at each examination. A semi-automated digital computer analysis technique for quantification of adult vasculature (Martinez-Perez, PhD thesis, Imperial College, 2001) was used to measure vessel dimensions and morphology. Analysis was restricted to the most prominent arterioles and venules within an area of radius 4.5mm (150 image pixels) centered on the optic disc. Results: In general, since arterioles are narrower than venules, they were not as amenable to analysis, particularly in infants younger than 31 weeks post-menstrual age. Mean vessel diameter (std error) averaged across all examinations was 77.41.7 µm for arterioles and 70.02.9 µm for venules. Mean arteriolar tortuosity (true length of vessel/straight-line distance covered) for the whole group was 1.140.02. One infant in the group progressed from no ROP to threshold. From this individual two arterioles and two venules were measured in all exams. These increased in diameter from 53.0 µm to 75.0 µm (42%) and 71.0 µm to 92.1 µm (30%) respectively. Arteriolar tortuosity increased from 1.08 to 1.12 (4.2%). Conclusion: We have shown that it is possible to acquire measurements of both diameter and tortuosity of retinal vasculature from digital images of the premature retina. Several factors contributed to variability in the data: image blur, image illumination (evenness and overall level) and the precise location of the optic disc border. In a few instances it was difficult to differentiate retinal vessels from prominent choroidal vasculature. Further refinements in image pre-processing and analysis should provide quantitative information on specific changes in vessel morphology appropriate to studies of retinovascular development in preterm infants with and without ROP.
This PDF is available to Subscribers Only