Lisa Willats1, Alan Connelly1,2, Soren Christensen3,4, Geoffrey Donnan2,5, Stephen Davis4,6, Fernando Calamante1,2
1Brain Research Institute, Florey Neuroscience Institutes , Melbourne, Australia; 2Department of Medicine, University of Melbourne, Australia; 3Department of Radiology, University of Melbourne, Australia; 4Royal Melbourne Hospital, Melbourne, Australia; 5Florey Neuroscience Institutes, Melbourne, Australia; 6Department of Neurology, University of Melbourne, Australia
Recent studies have suggested that in addition to perfusion deficits, delay and dispersion (D/D) of blood flow may also increase the risk of permanent tissue damage. We investigate the infarct risk associated with D/D using generalised linear models (GLM) formed with delay insensitive global-AIF and local-AIF perfusion estimates, together with independent D/D parameters. Including D/D parameters was found to improve the GLM predictions. The global-AIF models performed better summarised over all risk thresholds. However, for a single risk threshold (as would be used clinically), the local-AIF models would perform closer to optimum for a larger proportion of patients.