Computer-automated quantification of plus disease in retinopathy of prematurity.

BACKGROUND

In some cases of retinopathy of prematurity (ROP), it difficult to determine with certainty whether plus disease is present or absent. We have developed a computer program that captures digital images from a video-indirect ophthalmoscope, identifies and traces the major posterior pole blood vessels, measures the dilation and tortuosity of each vessel, and calculates whether or not an eye has plus disease. Our purpose was to determine the accuracy of the computer program in comparison with two masked examiners.

METHODS

A representative sample of posterior pole images from 20 premature infants, 10 normal and 10 representing various degrees of dilation and tortuosity, was extracted from our video database and analyzed by the computer program as well as by two masked examiners experienced in the diagnosis of ROP. The standard photograph from the Cryotherapy for ROP study, representing the minimum degree of dilation and tortuosity required for plus disease, was also digitized, analyzed, and used as a numeric comparison for the automated determination of plus disease.

RESULTS

Of the five images determined to have plus disease by both examiners, four were calculated to have plus disease by the computer program (80% sensitivity). Of the 11 images without plus disease, 10 were calculated not to have plus disease by the computer program (91% specificity).

CONCLUSIONS

Our computer program has very good sensitivity and specificity compared with masked examiners' determination of the presence or absence of plus disease. Automated analysis of dilation and tortuosity of posterior pole blood vessels has the potential to remove subjectivity from the determination of plus disease.