再现性
IRIS(生物传感器)
医学
超声生物显微镜
核医学
生物医学工程
眼科
数学
人工智能
计算机科学
生物识别
青光眼
统计
作者
N Wang,Minling Lai,Weibin Zhou
出处
期刊:PubMed
日期:1997-03-01
卷期号:13 (1): 29-34
被引量:3
摘要
To evaluate intraobserver and interobserver reproducibility of real time measurement of iris morphology in living human eyes.Based on the platform of software (Autocad, version 12), we developed an ultrasound biomicroscopy (UBM) image assistant measuring system. By using the system, we can perform the iris configuration quantitative measurement in living eyes. The measuring parameters including: iris rest length, radius of iris curvature, and the thickness in different parts of iris. Ten anterior segment images of one normal individual were obtained by a single operator to evaluate the intraobserver reproducibility of image capture, and ten times measurement of one image were performed by a single operator to assess the reproducibility of image measurement. The measurement of three independent observers were compared to investigate interobserver reproducibility in quantitative measurement. Intraobserver and interobserver reproducibility of measurement were assessed by calculating the coefficient of variation for each individual observer and by using the F test to detect a difference among observers. The iris configuration of 96 subjects (192 eyes) were measured.Intraobserver reproducibility was ranged from 0.9-4.9% in all measured parameters. Interobserver reproducibility for some of measured parameters varied considerably and was affected by subjective interpretative of visualized anatomic landmarks. The preliminary measured parameters in Chinese Show that iris rest length is 3.699 +/- 0.397 mm, radius of iris curvature is 9.101 +/- 1.408 mm, the average thickness of iris is 0.406 +/- 0.042 mm.The intraobserver reproducibility and the measured accuracy can fit the requisition of the ocular biometry and of the ocular physiology, pathophysiology, and pharmacology study. The method supply a new assistance for UBM image measurement.
科研通智能强力驱动
Strongly Powered by AbleSci AI