Abstract
Imaging genomics is an interdisciplinary field that integrates information from imaging modalities alongside genomic data such as single nucleotide polymorphisms (SNPs). By leveraging neuroimaging alongside genomic data, researchers can investigate how specific genetic variants influence structural brain phenotypes. Using new methods from the causal machine learning domain, it is now possible to investigate causal effects relating genetic variants to brain phenotypes using generated counterfactual phenotypes. While prior work has shown promising first feasibility results using causal generative models to detect these relationships, existing approaches often suffer from scalability issues and added complexity, limiting their practical application in larger-scale studies. This study investigates genotype-phenotype relations using the masked causal normalizing flow model, which allows for deterministic counterfactual generation instead of approximations, to assess the causal impact of selected SNPs on structural brain regions. For a first feasibility analysis, we specifically investigated brain regions that were shown to be related to two SNPs, rs1080066 and rs2164950, as identified in a previous genome-wise association study using UK Biobank data. Therefore, counterfactual imaging derived phenotypes (IDPs) for each dataset were generated by altering the allelic state of the chosen SNPs. The analysis revealed significant differences in IDPs between SNP alleles when assuming an alternate allele. These findings provide further evidence of the utility of counterfactual approaches in imaging genomics, providing a deeper understanding of the causal pathways linking genetic variant to IDPs. Overall, the results offer valuable insights for future research into the genetic underpinnings of mental and neurological disorders.
Links to Paper and Supplementary Materials
Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/AIIG2024_001.pdf
SharedIt Link: Not yet available
SpringerLink (DOI): Not yet available
Supplementary Material: Not Submitted
Link to Open Review
Open Review Page: Not Available
BibTex
@InProceedings{DagGab_Counterfactual_MICCAISAT2026,
author = { Dagasso, Gabrielle AND Vigneshwaran, Vibujithan AND Winder, Anthony J. AND Ohara, Erik Y. AND Wilms, Matthias AND Forkert, Nils D.},
title = { { Counterfactual analysis of genotype variant effects on imaging-derived phenotypes } },
booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
year = {2026},
publisher = {Springer Nature Switzerland},
volume = {LNCS 17252},
month = {pending},
page = {pending}
}