Abstract #3563

Connelly A, Calamante F, Willats L

Combining the techniques of maximum-likelihood expectation-maximisation and wavelet thresholding for the deconvolution of bolus tracking data is shown to be an effective method for accurately characterising the distorted shape of the tissue residue function recovered using a global arterial input function. In particular, when delay and/or dispersion of the bolus occur in the feeding vessels, the proposed methodology enables the resulting distortion to be quantified and distinguished from a true perfusion abnormality. A dispersion index can be defined and used in conjunction with delay and perfusion estimations to improve the interpretation of perfusion data.