Validation of a Novel Automated System for the Quantification of Capillary Non-Perfusion Areas in the Retina of Rats With Oxygen-Induced Retinopathy (OIR)

Abstract

Abstract: : Purpose:To compare a novel automated objective computer system to a manual technique for quantitative evaluation of abnormal non-perfusion areas in the retinas of rats with oxygen-induced retinopathy (OIR). Methods:Brown–Norway newborn rats were exposed to hyperoxia (75% oxygen and 25% room air) from postnatal day 7 (P7) to P12 and then returned to room air until P17. At P12, they received a single intravitreal injection of Kringle-5 plasminogen peptide (K5) into the right eye (5 µg in 3 µl of phosphate buffered saline) and the vehicle alone into the left eye. At P17, retinal vascular changes were visualized after intra-cardiac injection of high molecular fluorescein dextran. Flat-mount retinas were performed and examined under a fluorescence microscope (Olympus, BX51) and photographed with a digital camera (SIS ColorView II). The size of capillary non-perfusion areas was measured in the retinas obtained from untreated and K5-treated animals exposed to hyperoxia. The ratio of capillary non-perfusion areas versus total retinal surface was calculated using two different approaches: i) by manual and ii) the novel automated quantification system. The first process consisted of manual delineation of the obvious ischemic areas and the total retinal surface. In the second process the estimation of the non-perfusion areas was achieved applying automated image analysis (AnalySIS 3.2 software, SIS, Germany). The new automated method used in this study has recently been developed (Doelemeyer et al., submitted to ARVO2003). Results:When K5 peptide was injected into the vitreous, the capillary non-perfusion areas were significantly reduced as compared to vehicle-treated eyes. Both manual and automated capillary non perfusion area quantification gave similar results. Data are shown as mean ± SEM. Manual method: OIR 42% ± 3 % (n=8); Vehicle 43% ± 3%(n=5); OIR + K5 34% ± 3% (n=5). Automated method: OIR 50% ± 2.8% (n=8); Vehicle 52% ± 2.5% (n=5); OIR + K5 39% ± 1.6% (n=5) Conclusions:Our results clearly demonstrate that the novel automated quantification system gives results comparable with those obtained manually, moreover, it is faster, and more importantly operator-independent. We strongly suggest that this new analytical system is more reliable for drug profiling.

Keywords: retinopathy of prematurity • animal model • imaging/image analysis: non-clinical