A quantitative method for correlating observations of decreased apparent diffusion coefficient with elevated cerebral...

In patients presenting with cerebral ischemic injury, the outcome of injured brain tissue quantified as decreased apparent diffusion coefficient (ADC) may depend on associated alterations in cerebral blood perfusion (CBP). This study proposes a non-biased method to quantify associations between ADC and CBP in newborns with global or focal cerebral ischemia. The study population consisted of nine neonates (age: 0 to 3 days) presenting with clinical and imaging evidence of ischemia (seven with global hypoxic ischemia, and two with focal arterial ischemic stroke) with decreased ADC. Six newborns without diffusion abnormalities on magnetic resonance (MR) imaging served as a comparative cohort (age: 0 days to 4 weeks). All patients underwent MR imaging including diffusion weighted imaging (DWI) to determine ADC and axial arterial spin labeling (ASL) to determine CBP. An algorithm was developed that uses the B0 volume from the DWI raw data as a reference, co-registers the ADC and ASL-CBP data to the B0, generates mask filters, and finally performs a statistical analysis to automatically select regions of interest (ROIs) with ADC or ASL-CBP values that deviate significantly from the rest of the brain. If ROIs are identified in this analysis, the algorithm then evaluates correlation based on ROI location and volume. A significant correlation was found between decreased ADC and elevated ASL-CBP with regions of elevated ASL-CBP typically larger than the corresponding ADC abnormality. The association between decreased diffusivity and increased ASL-CBP suggests that, for this cohort, cerebral ischemia is associated with hyperperfusion.