Authors

• Alfonso Parra Blesa (alfonsoparrablesa@gmail.com)
• Alfredo Sánchez Alberca (asalber@ceu.es)
• José J García-Medina (josegarciam@yahoo.com)

Objetives

Develop a staging system for glaucoma and a predictive model to classify new patiens in the glaucoma stages using optical coherence tomography (OCT) data.

Type of study

Analityc observational cross-sectional and comparative study.

Inclusion criteria

Only left eyes were considered.

Sample size

Total number of eyes: 1001 (766 Healthy and 235 with Glaucoma).

Variables

Data preprocessing

It is known that the thickness of the retinal nerve fiber layers depends on age and BMO area. To avoid this dependency a linear transformation provided by the OCT company was applied to every variable. This linear transformations was obtained fitting a linear regression model on a normative group of healthy eyes.

$$z_i=(x_i-\bar x-b_{xe}(e_i-\bar e)-b_{xa}(a_i-\bar a))/s_x$$ where:

• $x_i$ is the value of variable $x$ on eye $i$.
• $\bar x$ is the mean of $x$.
• $s_x$ is the standard deviation of $x$.
• $e_i$ is the age of individual $i$.
• $\bar e$ is the mean of the age in the normative database of healthy eyes.
• $b_{xe}$ is the slope of the regression line of variable $x$ on the age.
• $a_i$ is the BMO area of eye $i$.
• $\bar a$ is the mean of the BMO area in the normative database of healthy eyes.
• $b_{xa}$ is the slope of the regression line of variable $x$ on the BMO area.
• $z_i$ is the standardized value of variable $x$ on eye $i$.

Glosary

• OAG: Glaucoma de ángulo abierto
• OCT: Optical coherence tomography
• BMO: Bruch Membrane Opening