作者: Chauhan Bc , LeBlanc Rp , Blanchard Jw , Hamilton Dc
DOI:
关键词: Glaucoma 、 Statistical power 、 Biomedical engineering 、 Population 、 Pixel 、 Optic disc 、 Spatial correlation 、 Tomography 、 Optic disk 、 Mathematics
摘要: PURPOSE. To describe and evaluate a new statistical technique for detecting topographic changes in the optic disc peripapillary retina measured with confocal scanning laser tomography. METHODS. The 256 × 256-pixel array of height values obtained each image from Heidelberg Retina Tomograph (Heidelberg Engineering, Heidelberg, Germany) was divided into an 64 superpixels, where superpixel contained 16 (i.e., 4 4) pixels. An analysis variance developed to analyze three baseline follow-up images. performance tested without adjustment spatial correlation using computer simulations real data normal control subject patient progressive glaucomatous change. RESULTS. Computer simulation fixed population means variance, varying showed monotonically increasing number superpixels significant test results (false positives), 20% false-positives when 0.8 (the approximate median value data). false-positive similar (17%) serial images subject. When corrected correlation, independent level remained at expected less than 5% both data. Although glaucoma decreased after correction change readily apparent. Statistical power detect mean differences ranging 0.5 4.0 SDs low 1 SD or less, but increased dramatically larger changes. CONCLUSIONS. This has high sensitivity while maintaining specificity.