MIN–MAX NORMALISATION ENHANCES MORPHOMETRIC SIMILARITY NETWORK SENSITIVITY IN YOUNG FEMALES WITH SELF-HARM BEHAVIOUR

Self-harm is a major global health issue, and understanding its neural mechanisms is key for effective interventions. Morphometric similarity networks (MSNs) quantify individual-level similarity of multimodal structural MRI features, typically using Pearson correlations of sulcal phenotype vectors after z-score normalisation. This study proposes an improved MSN construction method that uses Min–Max scaling to map each morphometric feature to a common range while preserving its original distribution, as an alternative to z-score normalisation. We compared MSNs from two normalisation methods in a sample of 48 young females (24 healthy controls and 24 with self-harm behaviour). Our results show that normalisation choice significantly affects MSN topology. Min–Max normalisation enhances between-group differences at both group and individual levels. In the group-average MSNs, the self-harm group exhibits stronger right intra-hemispheric connectivity but weaker left intra-hemispheric and inter-hemispheric connectivity. At the individual level, these patterns are replicated and further characterised by significantly reduced hemispheric strength ratios (left-to-right and intra-to-inter) and slightly lower overall network integration, including mean connection strength and node versatility. This study highlights how normalisation methods impact MSN construction and interpretation. Min–Max normalisation offers a more robust and sensitive framework for analysing brain network organisation, particularly in distinguishing self-harm, which could serve as a potential biomarker. Future studies should validate these findings in larger, diverse samples and explore clinical applications.