The “banding artifacts” in the coronary artery computed tomography (CT) (quasi-periodic horizontal shifts in multiplanar reformatted images) smear or distort the vulnerable plaque degrading the image quality to a non-diagnostic level (Fig. 1.1). We have developed a cone-beam ECG-gated reconstruction algorithm with a corrective method (Ref. 1, Fig. 1.2) but its corrective performance is still limited. Figure 1.2: The cardiac banding artifact correction (CBC) proposed by Taguchi, et al. (Ref. 3) improves the banding artifact by using a simple feathering technique.

Currently, many studies used strict exclusion criteria (<65 bpm and <+-5 bpm) to avoid the problem. The detector used in CT is not large enough to image the entire heart at one shot; thus, data are acquired during 5-10 heart beats, where data from one heart beat cover a limited range of the heart. An algorithm has to “connect” or “assemble” them to image the entire heart or coronaries. Unfortunately, with the presence of non-periodic heart motion (NHM), the shape of the heart may be different in subsequent heart beats even at the same ECG phase. Thus, each connection may generate the banding artifact (Fig. 1.3) as the algorithms currently available assume a perfect, periodic motion

Figure 1.1: Motion artifacts makes segments missing (left, arrowhead) and coronaries smeared (left, arrows) or disconnected (right, arrows). (Refs. 1-2)

Figure 1.2: The cardiac banding artifact correction (CBC) proposed by Taguchi, et al. (Ref. 3) improves the banding artifact by using a simple feathering technique.

Figure 1.3: Motion artifacts makes segments missing (left, arrowhead) and coronaries smeared (left, arrows) or disconnected (right, arrows). (Refs. 1-2)

The goal of this research is to develop a robust corrective method to overcome the NHM issue. We want to find a very similar state of the coronary artery in subsequent heart beats by optimizing the gating condition (the ECG phase, width and height of the gating window) for each heart beat. By “connecting” data from such conditions, we want to minimize the banding artifact, thus, substantially improve the robustness of the examination.
Our specific aims are: 1) to develop realistic simulation tools that generates 4D data with NHM; 2) to develop cardiac image reconstruction methods with NHM compensation; and 3) to evaluate the robustness of the methods using simulated and clinically acquired patient data. We plan to modify the 4D NCAT phantom, which accurately models the anatomical structures (including the heart and coronaries) and cardiac motion, and generate cardiac CT projection data sets that incorporate various types and degrees of the NHM.
 

We will develop methods to calculate the degree of equivalence between images from different heart beats and to estimate the amount of heart motion at the time of interest. An algorithm will be developed to uniquely define the gating condition for each heart beat to maximize the agreement between heart beats with the least motion.

We will evaluate the algorithm in terms of the accuracy of the depicted anatomy using both projection data generated from the NCAT phantom and the patient data acquired clinically. The robustness against NHM will be statistically evaluated to assess the potential of relaxing the exclusion criteria.