Appropriate methodology for automated scaling of DSC-CBF images for stroke evaluation

Introduction: DSC perfusion studies are routinely performed for the assessment of stroke and brain tumor patients and are being used with increasing frequency in large scale clinical trials. A persistent DSC problem in routine imaging and clinical trials is that DSC CBF images are difficult to scale with accuracy. Although CBF images derived from DSC are (in principle) quantitative, the absolute CBF values derived from patient studies often show wild global variation and some form of image calibration is needed to reliably interpret measured CBF values. Furthermore it is highly desirable to use a CBF rescaling method that uses a minimum of intelligent user input and which could be automated for the processing of large scale studies. The inability to produce reliable accurate quantitative CBF images across subjects is a major limiting factor that has prevented the use of CBF imaging in favor of less sophisticated Tmax, MTT or TTP assessment in stroke. This study tested alternative fully automated approaches to DSC-CBF image scaling. Two automated approaches to arterial input function (AIF) selection were evaluated and the value of automated venous outflow measurement followed by AIF correction was assessed.