Sources of variability in computed tomography perfusion: implications for acute stroke management

Abstract

Object Although dynamic, first-pass cerebral CT perfusion is used in the evaluation of acute ischemic stroke, a lack of standardization restricts the value of this imaging modality in clinical decision-making. The purpose of this study was to comprehensively review the reported sources of variability and error in cerebral CT perfusion results. Methods A systematic literature review was conducted, 120 articles were reviewed, and 23 published original research articles were included. Sources of variability and error were thematically categorized and presented within the context of the 3 stages of a typical CT perfusion study: data acquisition, postprocessing, and results interpretation. Results Seven factors that caused variability were identified and described in detail: 1) contrast media, the iodinated compound injected intravascularly to permit imaging of the cerebral vessels; 2) data acquisition rate, the number of images obtained by CT scan per unit time; 3) user inputs, the subjective selections that operators make; 4) observer variation, the failure of operators to repeatedly measure a perfusion parameter with precision; 5) software operational mode, manual, semiautomatic, or automatic; 6) software design, the mathematical algorithms used to perform postprocessing; and 7) value type, absolute versus relative values. Conclusions Standardization at all 3 stages of the CT perfusion study cycle is warranted. At present, caution should be exercised when interpreting CT perfusion results as these values may vary considerably depending on a variety of factors. Future research is needed to define the role of CT perfusion in clinical decision-making for acute stroke patients and to determine the clinically acceptable limits of variability in CT perfusion results.