A New Approach for the Estimation of MTT in Bolus Passage Perfusion Techniques.

Background In clinical practice MRI is now the investigation of first choice in the vast majority of cerebral disease. The ability to produce accurate cerebral perfusion maps on standard clinical MR scanners is therefore highly desirable. These techniques use the area under the contrast concentration curve as an estimate of blood volume within the voxel (RCBV) and the width of the contrast bolus as an estimate of the mean transit t ime (MTT). This allows calculation of relative cerebral blood flow (RCBF=RCBV/MTT) and the production of parametric maps of each parameter. The production of accurate CBV measures is possible if inflow effects and other non-linear MR variables can be compensated for. Production of accurate measurements of MIT is far more difficult and it is increasmgly clear that current methods for the measurement of MTT are flawed [l]. The standard approach is to assume that the width of the gamma variate time curve, generated by the passage of the bolus is representative of the time taken for the bolus to pass through a voxel. Clearly such an approach is highly dependent upon the shape of the active region and isotropic voxels would be preferred in order to eliminate net flow direction dependencies. Attempts to account for differences m cardiac output and administration of contrast are generally based on deconvolution. However, as the spatial resolution of the data improves the expected contribution to the width of the curve will decrease. An analysis of the expected accuracy of the ability to measure MTT for typical bolus width measurement accuracy (0.6 seconds) and RMS arterial width (4 seconds) is shown in table 1. This table illustrates that for typical blood flow velocities of 2 m m per second in grey matter and white matter capillaries MTT should not be reliably measurable in high resolution MR data sets (3mm voxels) with the RMS width of the bolus rising from 4.0 to a maximum of 4.02 seconds in the capillaries.