Improved algorithm for statistical batch-by-batch analysis of corneal topographic data

J Cataract Refract Surg. Jul-Aug 1997;23(6):903-12. doi: 10.1016/s0886-3350(97)80252-0.

Abstract

Purpose: To describe an improved algorithm for topographic batch-by-batch analysis and evaluate the effect of artifact correction on the reproducibility of topographic records.

Setting: University Eye Hospital Vienna, Austria.

Methods: We twice examined 22 eyes of 22 patients (volunteers) using the TMS-1 computer-assisted videokeratoscope (Computed Anatomy, Inc.). Entry criteria were age more than 60 years and lack of corneal pathology. A batch-by-batch analysis of the paired differences was performed. The software consisted of special macro functions for a programmable computer database to eliminate small local artifacts, as well as rotate data, produce mirror images, calculate averaged differences for an entire group, and reduce data to 225 areas in 7 concentric rings. The reduced set of data was used for statistical analysis and correction of defocus artifacts. Reproducibility was evaluated by calculating the absolute deviations between the areas of paired measurements.

Results: The absolute deviations were below 0.9 diopter (D) in over 50% of the areas in 90% of the patients with the uncorrected data; after correction for local artifacts, defocus artifacts, and both local and defocus artifacts, these values were 0.9 D, 0.8 D, and 0.5 D, respectively. Absolute deviations were above 1.0 D in 10% of the patients in 66, 40, 29, and 0 of 225 areas, respectively.

Conclusion: Correction for both local and defocus artifacts yielded a two-fold improvement in the reproducibility of topographic records.

MeSH terms

  • Aged
  • Algorithms*
  • Artifacts
  • Cornea / anatomy & histology*
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Middle Aged
  • Reproducibility of Results