Code for Topological scale framework analysis of rs-fMRI networks of inhaled substance abuse

This dataset contains the full source code (and results) used in the study “Topological Scale Framework Analysis of rs-fMRI Networks of Inhaled Substance Abuse”. The aim of this research was to test the hypothesis that adolescents diagnosed with Inhaled Substance Abuse Disorder (ISAD) exhibit reduced structural complexity and robustness in their brain functional networks compared to healthy controls. To this end, we employed a topological scale framework (scale-space model) to extract multiscale algebraic descriptors from resting-state fMRI (rs-fMRI) correlation matrices.

The dataset includes three Python files: the functions module – implements the scale-space model, hypergraph construction, and descriptor computation; an analysis script – loads subject-specific correlation matrices, applies the scale-space framework, and computes descriptors across scales; a comparison script – performs statistical analysis of descriptors (MANOVA, permutation tests, and OLS regression) to evaluate group differences.

The correlation matrices used in the study are not included here due to data-sharing restrictions. However, they are publicly available from the dataset repository described in Mijangos et al. (2024), which must be downloaded separately. The present code is fully compatible with that dataset and allows complete reproduction of the reported results once the matrices are obtained. Our analyses demonstrated that ISAD subjects show increased network fragmentation, reduced incidence matrix rank, and overall lower mean vertex degree across scales. Furthermore, statistical testing confirmed diagnosis as the only significant explanatory factor, with 31 out of 48 descriptors strongly associated with ISAD. These findings support the hypothesis of reduced topological robustness in ISAD-affected brain networks and highlight the sensitivity of the scale-space model to scale-dependent degradation.

Researchers can reuse this dataset to apply the scale-space model to other rs-fMRI or neuroimaging datasets; extend the topological analysis framework to different networked data, such as social, biological, or computational systems; and reproduce or benchmark alternative topological descriptors against those used in this study.

The scripts are documented with comments and a README file describing dependencies and usage instructions.