AS06-01 - Neuroimaging Markers of Genetic Risk, Disease Expression and Resilience in Bipolar Disorder: can They be Used for Diagnosis?

Abstract

Introduction

Althouhg Bipolar Disorder (BD) is amongst the leading causes of disability worldwide patients experience delays in accurate diagnosis ranging between 5–10 years.

Objectives

To develop a diagnostic aid for BD.

Aims

To evaluate the feasibility of applying pattern recognition techniques to structural magnetic resonance imaging (sMRI) data for the diagnostic classification of patients with BD.

Methods

Gaussian Process Classifiers (GPCs) were applied to gray (GM) and white matter (WM) sMRI data from 26 individuals with BD, 15 with Major Depressive Disorder (MDD) and 26 healthy controls.

Results

The best classification accuracy was obtained using GPC analysis of GM images which differntiated patients from controls with 73% accuracy (sensitivity 69%, specificity 77%). The GM classification accuracy obtained for BD compared to MDD was 66% (sensitivity 40%, specificity 93%). GM discriminative clusters were widely distributed within cortical and subcortical structures including the ventral prefrontal and cingulate cortex and parahippocampal gyrus, insula, thalamus and striatum.

Conclusions

Our results demonstrate the predictive value of neuroanatomical data in discriminating patients with BD from healthy individuals and MDD patients.