Vertexwise Cortical Deviation Mapping (VCDM): A precision approach to cortical thickness assessment in non-affective psychosis

Schizophrenia is associated with structural brain abnormalities, particularly in cortical thickness and gray matter volume. However, despite consistent group-level findings, these measures have had limited clinical utility due to regional variability and the lack of individualized reference frameworks. Structural MRI data were obtained from 1,343 individuals across three cohorts: the Human Connectome Project Young Adult (HCP-YA), Early Psychosis (HCP-EP) and a similar mixed cohort (BRAINS). Cortical thickness was assessed in healthy control (CON, n = 1206) and non-affective psychosis (NAP, n = 137) subjects using FreeSurfer and Vertexwise Cortical Deviation Mapping (VCDM) across 298,000 cortical vertices to identify deviations exceeding ± 2 standard deviations from age-adjusted normative values. Group differences and clinical correlations were evaluated using ANOVA and regression analyses. Heritability estimates for deviation metrics were derived from monozygotic and dizygotic twin data. NAP participants showed significant global cortical thinning compared to controls, with moderate effect sizes (d = 0.53–0.58). Vertex-wise analyses revealed substantially higher percentages of very thin cortical (pTHIN) vertices in NAP (left: d = 1.09; right: d = 1.25), while the percentage of very thick vertices (pTHICK) did not differ by group. Regional analyses of frontal and temporal cortices did not yield improved group differentiation. Clinical symptom severity (e.g., PANSS-positive, SANS scores) predicted higher pTHIN in NAP, while cognitive performance (e.g., working memory, vocabulary) was associated with pTHICK in both groups. Vertex-wise deviation metrics were found to be highly heritable, with heritability estimates of 0.86 (pTHIN) and 0.80 (pTHICK). VCDM offers a sensitive, individualized method for detecting cortical abnormalities in psychosis, with stronger effect sizes and broader spatial resolution than traditional approaches. It may support more precise diagnosis, prognosis, and treatment monitoring in clinical settings.