We appreciate the comments of Merin and Reeves (1) concerning our article (2). There is no doubt that image quality is the most crucial aspect of diabetic retinopathy screening. Both spatial and color resolution play an important role in both the acquisition and display devices.
In our study we used the fundus camera Canon CR5-45NM with a Sony video camera (DXC950). Unfortunately, at that time, the system mentioned by the authors of the letter (CR6-45 NM fundus camera with a Canon digital camera) was not available; indeed, though, it offers a relevant improvement over the older configuration.
The system we have used may not be able to provide enough image quality to detect every retinal lesion. It generated image files in JPEG format from a digital image of 800×600 pixels, which is lower than the recommended 1,300×1,000 and much lower than the new cameras that may reach 2,600×2,300 pixels. To minimize image degradation, we used the lowest JPEG compression available in our image grabber (1:3 ratio). However, despite these limitations, our data show that the grading of the retinopathy made after digital image inspection and after direct eye fundus examination shows agreement in the vast majority of cases and, therefore, an adequate follow-up or treatment of most patients could be made. Although JPEG compression may not have been the best choice, a recent article (3) shows that higher compression ratios than the one we used do not produce serious degradation of the images.
We mentioned in our article that we lost image features in high-grade diabetic retinopathy. This was not because of the resolution of our system but because hemorrhages and media opacifications deteriorated the visibility of the retina. We believe they would have been missed even with a higher resolution camera.
The review by Aiello et al. (4) suggests that, at the time their article was written, digital images obtained with nonmydriatic fundus cameras were not a replacement of standard seven-field stereoscopic 30° fundus photography. We did not compare our digital images with standard fundus photography; therefore, we cannot suggest such a replacement. Instead, we showed that there is agreement between digital image inspection and direct fundus examination in grading diabetic retinopathy. Thus, we believe that digital fundus images may be adequate for this purpose. We expect, as Merrin and Reeves mention, that new advances in technology will improve image quality, and will increase not only spatial and color resolution but also time resolution and stereoscopic viewing from which patients may benefit.
References
Address correspondence to Francisco Gómez-Ulla or Francisco Gonzalez, Department of Ophthalmology, Hospital Provincial de Conxo, C/Ramón Baltar s/n, E-15706, Santiago de Compostela, Spain. E-mail: [email protected] or [email protected].
