Machine Learning-Based Brain Morphology Analysis in Schizophrenia and Unaffected Siblings

1,018 participants from five public databases were analyzed using voxel-based morphometry and ensemble SVM models. The study identified consistent volume reductions in frontal, temporal, insular, and thalamic regions in patients, and found unaffected siblings were 3.8 times more likely to show similar brain morphology. The research, authored by Ishida Manabu and published in 2026, highlights a novel imaging phenotype of increased ventral striatal volume in siblings.

Use Cases

• Train machine learning classifiers for schizophrenia risk based on regional brain volume features.
• Identify structural endophenotypes for high-risk individuals using multi-site neuroimaging data.
• Analyze potential compensatory brain mechanisms in unaffected siblings based on ventral striatal volume.
• Validate voxel-based morphometry standardization techniques across independent databases.

Strengths

• Data from 1,018 participants across five independent public databases.
• Ensemble SVM achieved a high classification performance (AUC = 0.99861).
• Identified a specific novel imaging phenotype: increased ventral striatal volume in siblings.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data may reflect geographic or source bias inherent to the five aggregated databases.

Provenance

Source

Aggregated from five independent, public databases.

Collection Method

Brain MRIs were standardized via voxel-based morphometry, and covariate-adjusted z-scores were calculated for regional volumes.

Freshness

Last updated 2026-05-20 05:47:17; freshness should be verified.