Daniel Amen · Frontiers in Psychiatry 2021 · Retrospective case-control study · n=1135

SPECT Functional Neuroimaging Distinguishes Adult Attention Deficit Hyperactivity Disorder From Healthy Controls in Big Data Imaging Cohorts

Cited 18 times in the scientific literature.

Level 4 - case-series / case-control

Retrospective case-control diagnostic study with non-matched controls

OpenAlex W3217544718 · doi:10.3389/fpsyt.2021.725788 · record verified 2026-08-29

What was done

Researchers conducted a retrospective case-control study using a multisite psychiatric database to evaluate whether baseline resting-state brain perfusion SPECT imaging distinguishes adult ADHD from healthy controls. The sample comprised 1,135 participants: 1,006 patients with an ADHD diagnosis and no psychiatric comorbidities, and 129 non-matched controls with no history of psychiatric disorders, brain injury, or substance use. Region-of-interest (ROI) measures and visual readings were evaluated using binary logistic regression and Receiver Operating Characteristic analysis.

What was found

Visual reads demonstrated 100% sensitivity and >97% specificity for differentiating ADHD from controls, while post-hoc ROI analysis yielded 100% sensitivity and 100% specificity. Reduced baseline perfusion in ADHD patients was primarily observed in the orbitofrontal cortices, anterior cingulate gyri, prefrontal cortices, basal ganglia, and temporal lobes, along with predictive differences in cerebellar subregions, the medial anterior prefrontal cortex, left anterior temporal lobe, and right insular cortex.

Why it matters

The study identifies candidate regional cerebral blood flow patterns associated with uncomplicated adult ADHD compared to healthy controls.

Limits

The study is retrospective and relies on an unbalanced, non-matched control group (129 controls vs. 1,006 cases). Excluding all comorbid psychiatric conditions introduces severe spectrum bias, limiting real-world generalizability. Reported 100% sensitivity and specificity strongly indicate model overfitting and lack of independent out-of-sample validation.

Cited by