Identifying patients with obsessive-compulsive disorder using whole-brain anatomy

Area of Science:

• Neuroscience
• Psychiatry
• Medical Imaging

Background:

• Obsessive-compulsive disorder (OCD) presents significant heterogeneity in discrete anatomical measures, hindering clear discrimination from healthy individuals using structural neuroimaging.
• The clinical applicability of current structural neuroimaging studies for OCD is limited by this patient variability.

Purpose of the Study:

• To assess the feasibility of identifying obsessive-compulsive disorder (OCD) patients based on whole-brain anatomical alterations.
• To develop a method for quantifying 'OCD-like' anatomical patterns in individual subjects.

Main Methods:

• Collected whole-brain magnetic resonance images from two independent samples of OCD outpatients and control subjects.
• Computed voxel-wise whole-brain structural differences between OCD patients and controls at the group level.
• Calculated a single 'OCD-like' anatomical alteration expression value for each subject and assessed classification accuracy.

Main Results:

• Cross-validation achieved a high classification accuracy of 93.1% for identifying OCD patients.
• A second assessment on new groups yielded a lower but significant overall accuracy of 76.6%.
• Individual expression values correlated significantly with OCD symptom severity (Y-BOCS scale).

Conclusions:

• Obsessive-compulsive disorder (OCD) patients can be identified using whole-brain structural alterations.
• The inherent variability in psychiatric populations may affect classification accuracy.
• Individual anatomical characterization could provide valuable biological information for clinical assessment.