PET imaging pattern analysis may allow schizophrenia diagnosis

MedWire News: Pattern recognition analysis of radiotracer 6-[18F-] fluoro-L-DOPA ([18F] DOPA) positron emission tomography (PET) brain images may allow the accurate diagnosis of schizophrenia, UK study findings indicate.

Previous neuroimaging studies have suggested that striatal dopaminergic overactivity is involved in the pathophysiology of schizophrenia, with [18F] DOPA PET imaging revealing increased striatal dopamine synthesis and storage.

Noting that conventional [18F] DOPA PET imaging analysis is not sensitive or specific enough to be used diagnostically, Subrata Bose, from Imperial College London, and colleagues applied an Artificial Neural Network (ANN) model to images obtained from 19 paranoid schizophrenia patients and 31 healthy controls using [18F] DOPA rate constants within the anterior–posterior subdivisions of the striatum and across the two hemispheres. In addition, the images were analysed using general linear analysis.

The patients had an average age of 35.2 years, compared with 31.6 years among controls. Positive and negative symptom scores were in the mild to moderate severity range. Seven patients were medication free, with four antipsychotic naïve.

There was a trend towards higher [18F] DOPA uptake in patients than in controls and this was unaffected by hemisphere or anterior–posterior subdivisions.

An ANN model using all 12 bilateral [18F] DOPA uptake rate constants had a sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for identifying schizophrenia patients of 84%, 74%, 67%, and 88%, respectively.

These values were significantly improved by a second analysis using the 6 laterality quotients, at 89%, 94%, 89%, and 94%, respectively. Classification was not affected by gender or handedness.

The researchers also report that general linear analysis produced a sensitivity, specificity, PPV, and NPV for identifying schizophrenia patients of just 63%, 58%, 48%, and 72%, respectively.

The conclude in the journal Schizophrenia Research that “the use of ANNs, possibly in conjunction with neuroimaging and other data, is a promising approach in designing a diagnostic test for multifactorial psychiatric conditions such as schizophrenia.”