Voxel-wise temporal decomposition of hypoxia-powered BOLD MRI: method development and proof-of-concept application in glioblastoma

GLIODISSECT enables voxel-wise temporal decomposition of hypoxia-powered BOLD MRI, achieving median voxel-wise Q2 of 0.345 and reducing pooled prediction error by 16.0% in 20 healthy controls.

• Why it matters: Hypoxia-powered BOLD MRI can probe controlled hypoxic responses, but existing analyses often summarize timing and amplitude together, limiting component-specific interpretation and voxel-wise normative mapping.
• What they did: The method uses full-record cross-correlation and first-step 10% landmarks to estimate a refined voxel-wise lag, then separates arrival, descending/recovery transitions, and threshold-defined steady state amplitude using 5%/95% and 20%/80% thresholds.
• The result: Timing and steady-state maps were preserved across thresholds (median within-subject spatial r 0.9987–0.9998), numerical criteria were met in 97.789% and 97.682% of DTP estimates and all DTR estimates, and glioblastoma cases showed heterogeneous component maps beyond single amplitude summaries.

AI summary of the abstract · may contain errors · not medical advice · read the paper.