Prediction Of Contrast Enhancement In Left Atrial Appendage (LAA) CT Through A Numerical Modeling Approach
Qi Z, Wang D, Lee J, Song T, Pantelic M, Keimig T, Nadig J, Reeser N, Zemke D, Seger N, and Bevins N. Prediction Of Contrast Enhancement In Left Atrial Appendage (LAA) CT Through A Numerical Modeling Approach. Journal of Cardiovascular Computed Tomography 2020; 14(3):S36-S37.
Journal of Cardiovascular Computed Tomography
Introduction: Optimization of contrast protocol is crucial in LAA CT for the diagnosis of LAA thrombus; it ideally requires not only adequate LAA opacification, but also sufficient difference in opacification when a delayed phase is included to aid the diagnosis. Prediction of contrast enhancement with reasonable accuracy prior to the study may allow for patient specific adjustment of contrast protocol for improved outcome.
Methods: The proposed approach adopts a previously published model of the cardiovascular system with modifications made to include LAA. In this model, the cardiovascular system consisting of the heart, the vessels and various organs is simplified as a large group of interconnected compartments; the transfer of the iodinated contrast medium among different compartment is governed by a large group of differential equations. A patient’s clinical information, including age, gender, height and weight, are used to derive patient specific factors to adjust both the blood volume and the blood flow in the model. The iodine concentration of LAA at any time point is determined by solving the group of differential equations included in the model, and then the resulting HU enhancement is predicted by incorporating the physics of the CT scan process. A total of 20 LAA CT studies performed at our institution are included for evaluation of the proposed approach; the CT protocol included both a peak phase and a delayed phase. For all studies, no evidence of LAA thrombus was found by the interpreting physician teams. The predicted HU enhancements at both phases using the proposed approach are compared against the actual HU enhancements measured from the CT datasets.
Results: For the peak phase, the errors of the predicted HU enhancements, compared to the actual enhancements, have a mean value of -32 HU and a standard deviation of 32 HU. For the delayed phase, the errors of the predicted enhancements have a mean value of -31 HU and a standard deviation of 34 HU. In 75% of all comparisons, the deviation of the predicted enhancement from the actual enhancement is under 50 HU in magnitude.
Conclusions: A numerical modeling approach is proposed to predict contrast enhancement of LAA CT with the input of both patient specific clinical information and the contrast injection protocol. The proposed approach has potential values in achieving image quality improvement in LAA CT through contrast protocol optimization.