Loading...
The psychiatric disorders grouped as schizophrenia are known for their clinical heterogeneity and highly diverse patterns in brain abnormalities on neuroimaging. To explore discrete patterns linked to the diagnosis, and not simply overall averages, an international team of investigators employed a machine-learning algorithm using neuroimaging data for comparisons between 307 patients with schizophrenia aged <45 and 364 healthy controls.
The algorithm accounted for age, sex, and imaging protocols. Analyses further controlled for medication, illness duration, symptom severity, and other variables. Two distinct neuroanatomic subtypes emerged. In subtype 1 (63% of patients), the algorithm showed widespread lower gray-matter volumes (as typically…