TITLE: Toward time resolved 4D cardiac CT imaging with patient dose reduction: Estimating the global heart motion

 

AUTHORS: Katsuyuki Taguchi, W. Paul Segars, George S. K. Fung, and Benjamin W. M. Tsui

___________________________________________________________________

ABSTRACT

 

PURPOSE: Coronary artery imaging with multi-slice helical computed tomography is a promising noninvasive technique. The most important issues of the current cardiac applications include the insufficient temporal resolution and large patient dose. The temporal resolution is primarily limited to, e.g., 175 ms, by the rotation time of the gantry. It is further improved to ~100 ms by applying narrower time windows to the projection data from more than one heart beat per voxel. However, applying narrower time windows forces us to throw away more data, which leads to more patient exposure dose. The heart keeps moving even in the most "quiet" phases (e.g., mid-diastole); thus, employing a reconstruction algorithm without motion compensation always results in blurred spatial resolutions due to the (motion induced) inconsistency even within the narrow time windows.

 

METHODS AND MATERIALS: The method uses an iterative approach repeating the following four steps until the difference between the two projection data sets falls below a certain criteria in step-4: 1) estimating or updating the cardiac motion vectors, 2) reconstructing the time-resolved 4D dynamic volume images using the motion vectors, 3) calculating the projection data from the current 4D images, 4) comparing them with the measured ones. The algorithm allows us to use out of phase data to reconstruct the cardiac phase of interest in order to improve the temporal resolution while reducing the patient dose. In this paper, we describe the algorithm and perform a pilot study for the first step of the algorithm

      In this study, we obtain the first estimate of the motion vector. The halfscan reconstruction with the sliding time-window technique is used to generate cine images: f(t, r) from cone-beam projection data. Here, we use one heart beat for each position r so that the time information is retained. Next, the magnitude of the first derivative of f(t, r) with respect to time, i.e., |df/dt|, is calculated and summed over a region-of-interest (ROI). The minima in the time-|df/dt| curve indicate the motion is "quiet" at the corresponding time, which may be the end-systole (ES) and mid- or end-diastole (MD or ED). The initial estimations of the vector field are obtained using the block-matching algorithm with the mean-absolute difference (MAD) by using either of the following two approaches: Sequential method or sandwich method. The sequential method uses two adjacent cine frames at a time and sequentially changes those frames. The sandwich method uses two frames at the most quiet phases, the ES and MD or ED, to roughly estimate the "global motion vector".

      We use the 4D NCAT phantom, a realistic computer model for the human anatomy and cardiac motions, to generate the dynamic fan-beam projection data sets.

 

RESULTS: Two quiet phases of the heart motion, ES and MD, were accurately estimated (Figure 1). The accuracy of the motion vector field by either of the sequential and sandwich methods (Figure 2) was not too bad, but not satisfactory. The motion vectors of the myocardium or the heart wall were correct; the left anterior descending artery was tracked; and so as the left circumflex. The motion vectors of the right coronary artery (RCA) (Figure 2, yellow boxes) from the mid-diastole to the end-systole (right) was also correctly estimated, however, from the systole to the diastole, the vectors were pointing to wrong directions.

 

CONCLUSIONS: Even though the result of this pilot study is not perfect, it is certainly encouraging. We will further develop the algorithm.

 

FUNDING SOURCES: Start-up fund of Division of Medical Imaging Physics

 

 


 

Figure 1. The mean-absolute difference (MAD) provides two quiet phases, end-systole (ES) and mid-diastole (MD), which can be two end points of global motion vectors

 

Figure 2. Estimated motion vector field by the sandwich method using two frames (ES and MD).