Objective Diagnosis of Fibromyalgia Using Neuroretinal Evaluation and Artificial Intelligence

Int J Clin Health Psychol. 2022 May-Aug;22(2):100294. doi: 10.1016/j.ijchp.2022.100294. Epub 2022 Feb 23.

Abstract
in English, Spanish

Background/objective: This study aims to identify objective biomarkers of fibromyalgia (FM) by applying artificial intelligence algorithms to structural data on the neuroretina obtained using swept-source optical coherence tomography (SS-OCT).

Method: The study cohort comprised 29 FM patients and 32 control subjects. The thicknesses of complete retina, 3 retinal layers [ganglion cell layer (GCL+), GCL++ (between the inner limiting membrane and the inner nuclear layer boundaries) and retinal nerve fiber layer (RNFL)] and choroid in 9 areas around the macula were obtained using SS-OCT. Discriminant capacity was evaluated using the area under the curve (AUC) and the Relief algorithm. A diagnostic aid system with an automatic classifier was implemented.

Results: No significant difference (p ≥ .660) was found anywhere in the choroid. In the RNFL, a significant difference was found in the inner inferior region (p = .010). In the GCL+, GCL++ layers and complete retina, a significant difference was found in the 4 regions defining the inner ring: temporal, superior, nasal and inferior. Applying an ensemble RUSBoosted tree classifier to the features with greatest discriminant capacity achieved accuracy = .82 and AUC = .82.

Conclusions: This study identifies a potential novel objective and non-invasive biomarker of FM based on retina analysis using SS-OCT.

Antecedentes/objetivo: Identificar biomarcadores objetivos de fibromialgia (FM) aplicando inteligencia artificial a datos estructurales de retina obtenidos mediante tomografía de coherencia óptica Swept Source (TCO-SS).

Método: Se evaluó una cohorte de 29 pacientes con FM y otra de 32 sujetos control, registrando los espesores de la retina completa, de varias capas de la retina [capa de células ganglionares (CCG+), CCG ampliada (CCG++, entre la membrana limitante interna y los límites de la capa nuclear interna) y capa de fibras nerviosas (CFNR)] y de la coroides, mediante TCO-SS. La capacidad discriminante se evaluó mediante el área bajo la curva ROC (AROC) y el algoritmo Relief. Se implementó un sistema de ayuda al diagnóstico con clasificador automático.

Resultados: No se observó diferencia significativa (p ≥ .660) en la coroides, pero sí en el sector inferior del anillo interno de la CFNR (p = .010) y en los cuatro sectores del anillo interno en las capas CCG+, CCG++ y retina completa. Utilizando un árbol de decisión ensemble RUSBoosted como clasificador de las características con mayor capacidad discriminante, se obtuvo una predicción alta (AROC=.820).

Conclusiones: Se identifica un potencial biomarcador objetivo y no invasivo para el diagnóstico de FM basado en el análisis de la neurorretina mediante TCO-SS.

Keywords: Artificial intelligence; Fibromyalgia; Neurodegeneration; Observational descriptive study; Optical coherence tomography.