Shaolin Yang, PRESENTER: Nothing to Disclose

Abstract: HTML Purpose: Bolus tracking perfusion quantification can be obtained through deconvolution, which typically needs large computation cost and cautious treatment. In this study, we propose two new processing methods that do not require deconvolution. Methods and Materials: Assume q stands for tissue {Gd}, Ca for arterial {Gd}, f and v for regional blood flow and volume, respectively. The differential-type mass conservation equation, on which perfusion quantification is essentially based on, can be mathematically treated through integration transformation. Then we obtain (omitting the hematocrit and tissue density factors for simplication) v = òqdt/òCadt, f = òqdt/òt(q/v-Ca)dt (1) These parameters can also be estimated using linear regression, which leads to the following solution v = -(åqqåCadq/dt - åqdq/dtåCaq)/( (åqdq/dt åCaCa - åCaqåCadq/dt), f = (åqqåCadq/dt - åqdq/dtåCaq)/( (åqqåCaCa - åCaqåCaq) (2) For reference, we term Eq.1 as the integral method and Eq.2 the differential method. In this study, we evaluate the two new methods, in addition to two commonly used deconvolution methods, parametric deconvolution and non-parametric SVD-based deconvolution. The four methods were evaluated on both phantom (CA® cellulose acetate hollow fiber dialyzer) and human perfusion data obtained with gradient echo EPI. Results: The integral method seems less sensitive to the delay and dispersion effects of arterial input function (AIF) than the other methods in phantom experiments. The flow values obtained with the differential method are lowest due to the finite difference approximations of the derivatives. The human results of all the methods showed the same pattern for blood perfusion and volume, but there are variations in the regional perfusion values in which the integral method gives higher perfusion estimation while the differential method gives lower perfusion estimation than the other three methods. Conclusion: The parametric deconvolution method uses nonlinear least squares minimization that has a large computation cost. The SVD deconvolution needs additional threshold estimation to compress sensitivity to noise. In comparison, the integral method and differential method, with explicit mathematical expression of regional blood flow and volume, are direct, simple and fast. Our preliminary data suggest that the two methods can give perfusion quantification estimation comparable with standard methods. Further effort is needed to identify the cause of the discrepancy of the processed blood flow and volume among different methods.      

Yang, S, Bolus Tracking Perfusion Quantification Methods without Deconvolution.  Radiological Society of North America 2003 Scientific Assembly and Annual Meeting, November 30 - December 5, 2003 ,Chicago IL. http://archive.rsna.org/2003/3101513.html