Comparability Of Pixel Size In Images Of Different Cameras Of The Same Type Used For Multi-Center Trials At Different Study Sites Evaluated With A New Test Eye (SISPOT)

Abstract

Purpose: : Evaluation of possible differences in pixel size (PIXS) in images taken with one cameratype at different study sites to be able to compare the images for multi-center trials.

Methods: : All study sites involved in a multi-center trial were provided the SISPOT, a new test eye which consists of an anterior prime lens and a posterior focussing screen with a circle of exactly 4 mm diameter with 12 radial crossing lines. Each operator had to take 5 independent images of the circle on the focussing screen strictly following a protocol for mounting the test eye, setting the parameters of the camera and taking the images. The VRC evaluated the images by measuring the PIXS in every image after setting markers on the 24 circle-crossing line intersections. All calculated PIXS of the same cameratype (FF450 for color fundus (CF) images and HRAII and FF450 for fluorescence angiography (FLA) images) underwent a Kruskal-Wallis one-way analysis of variance to find out if the variations in PIXS are statistically significant and the images from the different cameras of the same type can be compared without further calculations.

Results: : 7 (4) sites used HRAII (FF450) cameratype for taking FLA images with 1536x1536 (1280x1024) pixels/image. The resulting 35 (20) images were not normally distributed and therefore underwent a Kruskal-Wallis analysis for independent samples (p=0.05 significance value). The significance value was p<0.001 (p<0.001), meaning there is a significant difference between the PIXS of images from each HRAII (FF450). The mean PIXS was 5.796±0.128 µm (6.478±0.125 µm). 3 sites used FF450 cameratype for taking CF images with 2256x2032 pixels/image. Kruskal-Wallis analysis lead to p=0.002, meaning there is also a significant difference between the PIXS of images from FF450 for CF. The mean PIXS was 3.816±0.100 µm.

Conclusions: : For multi-center trials it is important to be able to compare the images from different study sites. Pictures taken with the one cameratype may differ in PIXS as we see from our study. A reference measurement with a reliable tool like the SISPOT is therefore mandatory to have the opportunity to calculate the size of structures graded.

Keywords: clinical (human) or epidemiologic studies: systems/equipment/techniques • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • imaging/image analysis: clinical