International Journal of Artificial Intelligence and Neural Networks
Author(s) : TAI-YUAN SU, TSUNG-YEN TSAI
Tear film unstable is one of the major features of dry eye syndrome. One diagnostic method is fluorescein tear film break-up time test, this test is limited by its subjective and necessity to use manually reading the fluorescent image to identify the break-up area and thus variable result. The previously study use a deep convolution neural network to perform an automatic method to detect the fluorescent tear film break-up detection. However, the method is complex and required a lots of computer power. In this study we use a real time data segmentation method You only look once (YOLO) to screen the tear film break up pattern in real time mater. We demonstrate that YOLO is effective method and comparatively fast for recognition and localization in fluorescent tear film break up test. Experimental verification proves their high detection ability, location precision and real time processing speed using modern graphics processing unit.