Posters

Principal component analysis for quantitative evaluation of the progression of the clinical profile from early to advanced keratoconus

Poster Details

First Author: E.Chorro SPAIN

Co Author(s):    L. Bataille   J. Alio   A. Vega   M. Prieto   A. Cardona   D. Rivera     

Abstract Details

Purpose:

To obtain the profile of the keratoconus in the different states of progression: early, moderate and advanced, and to study how this profile changes with the progression of the severity in order to reduce number of relevant information for detection and classification.

Setting:

Vissum, Alicante, Spain. Division of Ophthalmology, Miguel Hernández University, Alicante, Spain.

Methods:

The data were obtained from 544 patients from the VISSUM centers. The analysis was performed on sets of 261, 166 and 117 cases with grade I, II and III, respectively. The technical statistics used for obtaining the pattern in each state is the principal components analysis taking into consideration the visual, refractive, anterior corneal topography and corneal aberrometry, a total of 16 clinical values. This statistic technique is used to perform a dimensional reduction by calculating new factors called principal components. The purpose of this is to correlate the changes of this pattern with the progression of the keratoconus stage.

Results:

The 16 parameters could be reduced to four factors, which explains approximately 80% of the total variability of the data, in all grades. These principal components were related to the corneal thickness, astigmatic information, aberrations and keratometric data, respectively, and they define the profile in each state of keratoconus progression. The results show that almost all factors reduce their explained variances when the grade increases. This means that some parameters, as thickness, are less relevant in advanced states. However, the factor related to the aberrations increases when the grade increases, becoming more relevant in more advanced grades.

Conclusions:

To conclude, the principal component analysis performs a reduction of data and defines a profile in each stage. In addition, a correlation between the changes of this pattern with the progression of keratoconus stage was found.

Financial Disclosure:

None