Researchers at ETH and the University of Zurich have succeeded in identifying different subtypes of schizophrenia using mathematical models that analyze fMRI images of the active brain. Does this open the way for more precise diagnoses and thus more targeted treatment options?
(ag) How can mental illness be diagnosed more accurately in the future? Scientists from ETH’s Institute of Biomedical Engineering, in cooperation with Berlin’s Charité hospital, have presented an approach that makes it possible to transfer imaged brain activity of test subjects into a mathematical model that then indicates whether or not they have schizophrenia. If positive, further differentiation into subgroups is performed. Specifically, what the researchers say is a “simple” model calculates the coupling strength between three selected brain regions. It tests how strong the communication between these areas is and thus allows conclusions to be drawn about the severity and nature of the illness of schizophrenic patients.
Study on 83 subjects
Already, the model has been tested on 41 patients with schizophrenia and 42 healthy participants. They had to look at and remember pictures (working memory is often impaired in schizophrenics) while their brain activity was recorded. The coupling strength of the three brain areas differed significantly not only between the patient and control groups, but also within the schizophrenia group itself. Three subgroups emerged with different brain activity patterns that corresponded to or represented the respective severity of schizophrenia when matched with clinical symptoms.
Of course, the model, while encouraging, is far from ready for real-world use, according to the researchers. In particular, it would be interesting to study untreated patients over time to see if the model can be used to confirm progression predictions about the disease.
Source: Media release, January 7, 2014, Schaffner M: Using mathematical models to detect schizophrenia. ETH Life January 7, 2014.
Literature:
- Brodersen KH, et al: Dissecting psychiatric spectrum disorders by generative embedding. NeuroImage: Clinical 2014; 4: 98-111. doi:10.1016/j.nicl.2013.11.002.
InFo NEUROLOGY & PSYCHIATRY 2014; 12(3): 38.
