Brain imaging in schizophrenia: what MRI and PET show — and what they don't

One of the most common questions families ask when schizophrenia is first suspected is, "Can a brain scan tell us for sure?" The honest answer is no. Brain imaging plays an important role in psychiatric care — but not the role most people imagine. This guide explains what MRI, CT, and PET scans can do, what they cannot do, and why imaging is still ordered as part of a thorough evaluation.

The kinds of brain imaging used

• CT (computed tomography) — fast, widely available, good for ruling out bleeding, large tumours, or major structural abnormalities. Limited soft-tissue detail.
• MRI (magnetic resonance imaging) — slower and more expensive but provides much better soft-tissue detail. The standard tool when imaging is ordered for first-episode psychosis.
• fMRI (functional MRI) — measures blood flow changes that track neural activity. Used heavily in research, almost never in clinical care.
• PET (positron emission tomography) — measures metabolism or specific receptor binding using injected radiotracers. Used in research and in selected diagnostic situations (such as suspected dementia or epilepsy).
• DTI (diffusion tensor imaging) — an MRI technique for examining white-matter tracts. Almost entirely a research tool.

Why imaging is ordered in first-episode psychosis

Most international guidelines, including NICE in the UK, recommend brain imaging — usually MRI — at first presentation of psychosis. The goal is not to "see" schizophrenia but to identify other conditions that can present with psychotic symptoms:

• Brain tumours, particularly frontal or temporal lobe
• Stroke or vascular malformations
• Multiple sclerosis lesions
• Autoimmune encephalitis (such as anti-NMDA receptor encephalitis)
• Neurodegenerative conditions in older patients
• Hydrocephalus
• Traumatic brain injury sequelae

The proportion of first-episode psychosis cases where imaging changes management is small (estimates range from 1% to 5%) but the consequences of missing one of these conditions are severe.

What research-scale studies have found

Decades of research comparing groups of people with schizophrenia to control groups have shown several reproducible findings:

• Slightly enlarged ventricles (the fluid-filled spaces inside the brain), on average
• Slightly reduced overall grey matter volume, particularly in frontal and temporal regions
• Subtle volume reductions in the hippocampus, thalamus, and prefrontal cortex
• White matter abnormalities in tracts connecting frontal regions to other brain areas
• Differences in functional connectivity between brain networks during rest and during cognitive tasks
• Increased dopamine synthesis capacity in the striatum on PET scans

These differences are robust at the group level but small in magnitude. The distributions of measurements heavily overlap with healthy controls — meaning that any individual person with schizophrenia might have completely normal-looking imaging, and many healthy people have measurements that fall within the "schizophrenia" range. This is why imaging cannot diagnose any individual.

The dopamine hypothesis and PET

One of the most consistent imaging findings in schizophrenia is increased dopamine synthesis capacity in the striatum, measured with PET tracers like 18F-DOPA. This finding is one of the strongest pieces of biological evidence supporting the dopamine hypothesis of schizophrenia and helps explain why dopamine-blocking antipsychotics work. Importantly, this signal appears even in people in the prodromal phase before full schizophrenia develops, which has fuelled hope that PET could one day help identify high-risk individuals — but it is not yet usable for individual diagnosis.

Machine learning and "neuroimaging biomarkers"

For the past decade, research groups have applied machine-learning algorithms to large MRI datasets to try to predict diagnosis, conversion to psychosis, or treatment response. Some of these models achieve accuracies of 70–80% in classifying schizophrenia versus controls in research samples. They have not been adopted clinically because:

• Performance drops when models are tested on independent samples from different scanners and populations
• Accuracy is not high enough for individual decisions in high-stakes situations
• The models do not reliably distinguish schizophrenia from related conditions (bipolar disorder, autism, severe depression)
• No biomarker has cleared regulatory approval as a diagnostic tool

Progress is real but slower than headlines often suggest.

What imaging cannot tell you

• Whether you, specifically, have schizophrenia
• Which medication will work for you
• How severe your illness will be
• Whether your symptoms will respond to treatment
• Whether someone is "really" psychotic or "faking it"
• Whether your child or sibling will develop schizophrenia

What imaging is being studied for

• Predicting conversion from prodrome to first episode — combining imaging, clinical, and genetic data is a focus of programs like NIMH-funded AMP SCZ
• Predicting treatment response — imaging features may eventually help match people to medications
• Distinguishing schizophrenia from bipolar disorder and other conditions at the individual level
• Tracking treatment effects over time in clinical trials

What to expect if you are sent for an MRI

A standard brain MRI takes 30 to 60 minutes. You lie still inside a long tube while the scanner makes loud knocking sounds. You will wear earplugs or headphones. If contrast dye is used (less common in psychiatric imaging), you will have an IV. The procedure is painless. People who experience claustrophobia can usually be accommodated with sedation, an open MRI, or a short-acting anxiolytic — ask in advance.

What the radiology report will say

If imaging is ordered to rule out other causes of psychosis, the report will most often read something like "no acute intracranial abnormality" or "unremarkable for age." This is the expected and reassuring result. It does not confirm schizophrenia, but it makes other diagnoses less likely.

Should I pay out of pocket for a research-grade scan?

Probably not. Commercial offerings that promise to "diagnose schizophrenia" or other psychiatric conditions from a brain scan are not supported by mainstream evidence. The genuine research advances are happening in academic centres and large publicly funded studies, not in private clinics. Money is generally better spent on a thorough clinical evaluation by a specialist.

This article is for educational purposes only and is not medical advice, diagnosis, or treatment. Always consult a qualified mental health professional. If you or someone you know is in crisis, call or text 988 in the US, or your local emergency number.