The Promise of Neuroimaging in Psychiatric Diagnosis
“Psychiatrists remain the only medical specialists that never look at the organ they treat.”- Daniel Amen
Interest in the neurobiological substrates of behavior and personality is growing with the explosion of new tools and technologies. One critical complaint about psychiatry is that there is too much subjectivity- it lacks definitive tests for many diseases. Researchers widely acknowledge the limitations of older methods such as self-report questionnaires/screens. The psychometric test gives us an idea of performance above/below the norm, yet gives no clue as to nature of the cognitive deficit and underlying neural substrate. For example, ADHD is currently diagnosed mainly by a child’s behaviour, and there are those who argue that aggressive drug companies, high-strung parents and overworked teachers have led to chronic overdiagnosis and overmedication. Is it possible to remove the subjectivity? To instead have an objective test based on neuroimaging?
A recent paper describes such a system–“Anatomical Brain Images Alone Can Accurately Diagnose Chronic Neuropsychiatric Illnesses”. The paper used data from earlier studies which analyzed the MRI scans of people already diagnosed with bipolar disorder, ADHD, schizophrenia, Tourette’s syndrome, and depression. The scientists divided scans randomly into two sets, one to build the diagnostic algorithm and the other to test it. The software organized the scans in the first set by groups according to similarity in shape of various brain regions. The groups were labeled by the disorder most common to those brain abnormalities. This enabled the system to build an algorithm based on brain structure. Next the scientists tried the new system against the second set. The system analyzed brain regions of each scan and assigned it to the group it most resembled. The scientists checked the algorithm’s work by comparing the new labels on the test scans with the original clinical diagnoses. When the system chose between two disorders or one ailment and a clean bill of health, its accuracy was close to perfect. It would seem neuroimaging is ready for implementation immediately.
However, there were some serious issues. Only chronically ill patients with a single, unambiguous diagnosis were used in the study. In the real world, people often describe an array of symptoms and receive multiple diagnoses. Though these are the exact people such a neuroimaging test would benefit most, the system unfortunately struggled when trying to assign people to one of three (rather than two) categories. In addition, brain scan tests may not be able to diagnose the early stages of a disease. The patients in this study had been ill for an average of more than 10 years. We know psychiatric illness induces long-term structural abnormalities. But could a neuroimaging test catch the biomarkers of a nascent psychiatric disorder? Not at the moment. And finally, the system used diagnostic categories which may not, ultimately, be biologically valid. In the real world, schizophrenia and bipolar disorder share many of the same symptoms and genetic risk factors. The lines are blurred in the world of biology, but not in the world of diagnosis. Therefore, the groupings used by the algorithm in this study may not truly reflect distinct neurobiological diseases.
Neuroimaging technology has led to enormous gains in research on the pathophysiology of psychiatric disorders. The impact of developing objective, biologically based tests for psychiatric illnesses would be enormous. Despite this, and efforts like the study discussed above, neuroimaging has yet to impact the diagnosis or treatment of individuals. There is still a ways to go in identifying all the underlying neural abnormalities and biomarkers of different psychiatric disorders before neuroimaging can be used as a viable diagnostic tool.
