Structural MRI may be effective at diagnosing mental illnesses such as bipolar disorder by applying a pattern recognition algorithm to anatomic data, according to a study published online June 4 in Psychological Medicine.
Most mental illnesses are diagnosed based on clinical assessment of symptoms, not imaging. Prior research has suggested morphological changes in the brain are associated with bipolar disorder. However, these studies were largely based on group-level data and could not make predictions about individual subjects.
“The data presented here demonstrate that these limitations may be surmounted with the aid of multivariate pattern recognition techniques,” wrote Sophia Frangou, MD, of the Ichan School of Medicine at Mount Sinai in New York City, and colleagues. “The application of [Gaussian process classifier (GPC)] analysis to anatomical scans in [bipolar disorder] provided diagnostic accuracy in the range 69–78 percent.”
Findings were based on analysis of structural MRI data from two independent samples of bipolar disorder patients. Cohort 1 featured 26 patients, cohort 2 featured 14 patients, and each cohort was matched with healthy controls.
GPCs were applied to gray and white matter MRI data, explained the authors. These classifiers are first trained to determine a predictive distribution from control cases, then the models are used to separate healthy individuals from those with bipolar disorder in a test sample.
Diagnostic accuracy of the GPC for gray matter was 73 percent in cohort 1 and 72 percent in cohort 2, reported Frangou and colleagues. Sensitivity and specificity of GM classification were 69 and 77 percent, respectively, in cohort 1, and 64 and 99 percent, respectively, in cohort 2. Diagnostic accuracy for white matter was 69 percent in cohort 1 and 78 percent in cohort 2.
“In both samples, [gray matter] and [white matter] clusters discriminating between patients and controls were localized within cortical and subcortical structures implicated in [bipolar disorder],” wrote the authors.
The level of accuracy demonstrated by the MRI analysis is comparable to other tests for bipolar disorder, noted the authors. This technique is not ready to replace a clinical exam, but Frangou and colleagues suggested it has potential to aid in diagnosis of bipolar disorder, which is often missed or misdiagnosed, “resulting in nearly a third of patients having to wait for approximately 10 years before they receive an accurate diagnosis.”