Cerebral microstructural alterations as an imaging biomarker for Post-COVID-condition.

Area of Science:

• Neuroimaging
• Biomarker Discovery
• Medical Diagnostics

Background:

• Post-COVID-condition (PCC) presents a significant challenge in diagnosis and management.
• Objective diagnostic tools are needed to identify individuals with PCC.
• Magnetic Resonance Imaging (MRI) offers potential for identifying neuroimaging biomarkers.

Purpose of the Study:

• To develop and validate an imaging biomarker-based approach for diagnosing Post-COVID-condition (PCC) at the individual patient level.
• To identify specific MRI metrics and brain regions that can differentiate PCC patients from healthy controls.
• To assess the diagnostic performance of a machine learning model trained on these imaging biomarkers.

Main Methods:

• Prospective cohort study comparing PCC patients (n=89) with unimpaired COVID-19 survivors (n=38).
• Acquisition and analysis of MRI data including macrostructure, diffusion tensor imaging (DTI), and multi-shell microstructure imaging.
• Training a linear support vector machine (SVM) using extracted atlas-based imaging metrics.

Main Results:

• Microstructural imaging parameters yielded the highest diagnostic performance.
• The optimal SVM model achieved an Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.95, with 94% sensitivity and 85% specificity.
• Key discriminatory regions included gray matter areas (cortical regions, putamen, thalamus) and white matter tracts (corpus callosum, frontal white matter).

Conclusions:

• Microstructural MRI data analyzed with SVM can effectively differentiate PCC patients from controls with high sensitivity.
• The findings represent a significant advancement toward a biomarker-based diagnosis of PCC.
• Further validation in larger, multicentric cohorts is necessary due to moderate specificity and monocentric design before clinical application.