시야검사 기반 녹내장 진단의 인간공학적 개선 연구

Abstract

Glaucoma is an eye disease in which the optic nerve is damaged, permanently impairing vision and progressing to complete blindness if left untreated. Unfortunately, there is no optic nerve regenerative therapy. After early detection of glaucoma, continuous management and treatment is need to prevent blindness. In order to diagnose for glaucoma, it is needed the results which are founded optic nerve damages in structural and functional measurements. Visual field test is a key examination for testing the function of optic nerve through automated perimetry which is measured the damaged loaction and degree on the retinal nerve fiber layer (RNFL). To obtain precise testing results, it is important to fix examinee’s eye on fixation target for testing time, however the existing perimetries don’t have any gaze fixation induction methods. Diagnosis of glaucoma is needed comprehensive considerations of the results of visual field testing and structural measurements, it is possible by glaucoma specialists who have a lot of knowledge and experiences in the clinics. Therefore, it is necessary to develop ergonomic gaze fixation induction methods for better reliability of visual field test and glaucoma diagnostic model for precise diagnosis of glaucoma.This present study is intended to research ergonomic improvements of diagnosis of glaucoma based on visual field examination. First, in order to effective visual field testing we developed the gaze fixation induction methods by changing size, flash rate, and shape of fixation target and conducted experiments for performance and subjective evaluation. As the result, fixation loss rate(FLR) of medium(angular subtense = 0.43°) size of fixation target was less 41% than that of other size of ones. Then, FLR of flashing dot(10 Hz) is less 11% than that of medium one. The reason was identified less concentration caused by the size differences between the size of fixation target and that of visual field testing targets. Because the flashing dot provide constant stimuli to examinee’s eye, examinees can concentrate on the flashing dot during visual field testing.Second, new glaucoma diagnostic model(PD/G-SVM) was developed and validated by using pattern deviation (PD) as learning sample and applying Gaussian support vector machine (G-SVM) to classify test smaple, considering glaucoma diagnosic process in the clinic. The performance of ROC area (0.95) of PD/G-SVM is better than that of ROC area (0.91) of existing research diagnostic model which was used threshold sensitivity value(dB) as learning sample. Because the function of PD is to highlight localized visual field loss, it is being considered as more important measure than threshold sensitivity value(dB) in the clinic. Missclassified samples from applying PD/G-SVM was classified by using vertical cup-to-disc ratio( > 0.713) as cut-off. Consequently, the accuracy of final glaucoma diagnostic model has improved about 2% more than before. In summary, the reliability of visual field test result is increased by using flashing dot for fixation target. The performance of glaucoma diagnostic model developed by using important factors which is being considered in the clinic is better than otherwise.