Hyperband Versus Conventional Resting-State fMRI in Young Adults with High-Risk Substance Use: A Within-Participant Comparison of Default Mode Network Measures

Background Accelerated resting-state fMRI (rs-fMRI) may reduce scan time and participant burden, but whether shortened acquisitions preserve functional connectivity profiles and clinical associations remains unclear. This study compared hyperband and conventional acquisitions in characterizing default mode network (DMN) connectivity and associations with substance use. Methods A community sample of 108 young adults with recent heavy substance use (59.93% female; mean age = 20.58 ± 1.13) completed a 9-minute conventional scan and a 4.5-minute hyperband scan. DMN connectivity was assessed using ROI-to-ROI analyses and graph-theory metrics. Substance use measures included cannabis use severity (CUDIT), urine tetrahydrocannabinol positivity (THC+), and hazardous drinking (AUDIT). Results Quality-control metrics did not differ between acquisitions. Conventional scans showed higher connectivity between bilateral lateral parietal regions and between the posterior cingulate cortex and left lateral parietal cortex, although overall patterns were comparable. Agreement between acquisitions was low (mean ICC = 0.358 ± 0.123). Graph-theory metrics showed few differences, but most ICCs were non-significant. Nominal associations with substance use were inconsistent across acquisitions. Conclusions Hyperband and conventional rs-fMRI produced similar group-level DMN patterns, but hyperband acquisitions did not reliably preserve individual-level estimates or substance use associations. The data should therefore not be treated as equivalent or readily harmonizable.