Linear quantification correction for myocardial perfusion imaging from x-ray coronary angiography

Abstract

Myocardial perfusion imaging from x-ray coronary angiography is important with clinical benefits because online real-time assessment of myocardial blood flow can promote the clinical outcomes of interventional treatments for coronary artery disease. In this paper, we aim at the nonlinearity problem of contrast image measurements for the perfusion estimation, since x-ray nonlinear responses of iodinated contrast agent is always an important concern when lacking of x-ray depth information on 2D angiography. A new approach is developed to perform linear quantification correction to angiographic measurements in terms of iodine concentration for estimated body thickness. We recognize the causes of nonlinear measurements from three different sources, that is, image processing artifacts of background subtraction, x-ray physics causes of beam hardening, photon scattering and detector glare if image intensifier applied, as well as clinical application issue of residual contrast agents in myocardium during cardiac catheterization heart procedure. Correspondingly, the developed approach involves three countermeasures to handle the three nonlinear sources. In order to compensate the registration artifacts of background subtraction, the technique of layer image processing is applied to compensate the different cardiac and breathing motions. A prior phantom-based calibration is implemented to make a lookup table of correction models. A polynomial model selected from the table is used online to correct the nonlinear measurements due to x-ray physics causes. For the effect of residual contrast agent, a new workflow of triple background subtractions is proposed by introducing an initial background image. Finally, the proposed approach is validated with pre-clinical studies of porcine models.