Can brain MRI help us in diagnosing bipolar disorder? Psych Med,Feb,2014.

12.02.2014

Bipolar disorder (BD)  is often not diagnosed early in the course of illness. Delay in diagnosis and treatment adds to the disease burden considerably. It is also thought that without interventions, the  neurobiological changes can drive further deterioration. Even though structural MRI studies have shown that BD is reliably associated with abnormalities in the ventral prefrontal cortex, the cingulate gyrus, amygdala/parahippocampal complex and the basal ganglia, clinical usefulness of such knowledge has been negligible. Conventional MRI compute mean group differences in spatial localised anatomical regions and DO NOT use information about the distributed pattern of relationships among regions or voxels. Would recent advancements in pattern recognition help to discriminate BD from others in usual clinical setting?

Most commonly used pattern recognition algorithm is the support vector machine (SVM) classifier. This yield binary  outcomes i.e. case or control.  Probabilistic predictions ( offered by the new algorithm GPC: Gaussian process classifiers )  are desirable  as they  provide accurate quantification of predictive uncertainty and allow adjustment of predictions to compensate for different frequencies of diagnostic classes within the general population.

V. Rocha-Rego J. Jogia A. F. Marquand, J. Mourao-Miranda, A. Simmons and S. Frangou report the results of the first study to address this issue.  They used GPCs to examine the predictive value of whole-brain gray (GM) and white matter (WM) anatomy in discriminating patients with BD from healthy individuals . 26 Bipolar 1 patients in remission, free of any other lifetime psychiatric co-morbidity were recruited and compared against 26 healthy controls. Two independent cohorts (cohort one with 26 patients, cohort two with 14 patients) were studied. Participants were scanned using a 1.5-T GE NV/i Signa MR system (GE ) at the Maudsley Hospital, London.

Results:

Classification accuracy using GPC analysis of Grey Matter images was 73% with a sensitivity and specificity of 77% and 69% respectively. If a participant had a clinical diagnosis of BD, the probability of correct classification was 0.77. Conversely, if a participant did not have BD, the probability of being correctly classified as a control was 0.69.  The GPC analysis using White Matter images  yielded an accuracy of 69% with a sensitivity of 69% and specificity of 69%. In summary, the classification accuracy range between 69% and 78%. 

This is the first study to evaluate the feasibility of using pattern recognition algorithms for the automatic classification of structural MRI data of patients with BD and healthy controls.The accuracy of the GPC classification for BD was determined against ‘gold standard’ diagnostic assessments ( expert clinicians ,using SCID1) i.e. in pure samples.The accuracy of predictions in real world clinical situations is not known.  Would it be of use where behaviour based case-finding instruments such as the Mood Disorder Questionnaire is used?  Would it be useful where uncertainty in diagnosis exists?  ( due to comorbidity or due to disorder not expressed fully or during different phases of illness). It is possible that a classifier that is trained to identify true positives BD cases might  assist clinicians when used in combination with other clinical measures. 

Conclusion: GPC- based neuroanatomical pattern recognition techniques may prove clinically useful in improving the timely diagnosis of BD, which currently relies entirely on clinical symptoms. 

Summary of the article: 

Examination of the predictive value of structural magnetic resonance scans in bipolar disorder: a pattern classification approach.Rocha-Rego V, Jogia J, Marquand AF, Mourao-Miranda J, Simmons A, Frangou S.Psychol Med. 2014 Feb;44(3):519-32.

 

 

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s