A methodologic issue for ophthalmic telemedicine: image quality and its effect on diagnostic accuracy and confidence

The possibility of ophthalmic telemedicine raises questions as to the quality of images displayed on a video display terminal and its effect on a doctor's accuracy and confidence of diagnosis. This study compares diagnostic accuracy and confidence levels of subjects observing two formats of retinal imaging.

Test images included 90 conventional 4- x 6-inch retinal fundus photographs and the same set of photographs after they had been scanned, digitized, modified in brightness and contrast, and displayed on a video display terminal. Fifty-six of the images demonstrated one or more retinal anomalies; the remaining 34 were of normal fundi. Abnormalities were selected on the basis of estimates of prevalence in the general population. Twenty optometrists affiliated with the Southern California College of Optometry participated as subjects. One group of 10 was shown the photographs; the remaining ten subjects were shown the digitized images. Both groups were asked if the pictures were "healthy" or "unhealthy," how confident they were of their decision, and, if abnormal, what was the ocular disease.

T-tests were used to compare the results of the two groups. There was a statistically significant difference in the confidence of a specific diagnosis between the two groups (p = 0.018), as well as tendencies toward significance on other measures.

Viewing digitally reconstructed photographic images of the retina on a video display caused doctors to lose confidence in making a diagnosis before they lost accuracy.