ImageNomer

ImageNomer integrates functional connectivity and omics data to detect demographic confounds and quantify associations between fMRI-derived functional connectivity (FC), single nucleotide polymorphism (SNP) features, and behavioral measures such as Wide Range Achievement Test (WRAT) scores.

Key Features:

• Functional connectivity (FC) analysis: Analyzes fMRI-derived FC features at subject and cohort levels.
• Omics (SNP) analysis: Analyzes single nucleotide polymorphism (SNP) features alongside imaging-derived features.
• Confound detection: Detects and quantifies demographic confounds, explicitly race, affecting FC–phenotype and SNP–phenotype associations.
• Variance quantification: Quantifies and compares variance explained by FC and SNP features (reported 10–15% of WRAT variance in the PNC) and assesses changes after controlling for confounds.
• Visualization and summarization: Produces subject-level and cohort-level visualizations and summaries of high-dimensional datasets to reveal correlations and quality-control issues.
• Demonstrated datasets: Applied analyses on the Philadelphia Neurodevelopmental Cohort (PNC) and the Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP) datasets.
• Implementation: Python-based implementation.

Scientific Applications:

• Confound identification in achievement prediction: Identifies race as a confounding factor in predicting WRAT scores from fMRI FC and SNP data in the PNC.
• Comparative feature evaluation: Compares predictive contributions of FC and SNP features to behavioral variance and evaluates their sensitivity to demographic covariates.
• Race–FC correlation exploration: Explores race–FC correlations in the BSNIP dataset to assess demographic influences on functional connectivity.
• Bias assessment in neuroimaging-genetics studies: Assesses the feasibility of identifying unbiased achievement-related imaging and genetic features in adolescent cohorts.

Methodology:

Subject- and cohort-level visualization and summarization of FC and SNP data; correlation analyses between FC/SNP features and WRAT scores; quantification of variance explained by FC and SNP features (reported 10–15% in PNC); and re-analysis including race as a covariate to assess confounding, applied to PNC and BSNIP datasets.