Temporal Resolution and Motion Artifacts in Dual-Source Cardiac CT and Single-Source CT with Iterative Reconstruction

The temporal resolution (TR) of an image has become an important property in computed tomography (CT) when examining organs whose motion can not be stopped for the duration of the CT scan, such as in cardiac CT. In this context, “higher TR” conventionally means that the raw data from which a given image was reconstructed was measured within a shorter time interval than the raw data of a similar image with a lower TR value. Since object motion during data acquisition will inevitably lead to artifacts in the reconstructed image, and “more motion” usually also means “more artifacts”, there are typically less motion artifacts visible in a high TR image of a moving object than in an image with lower TR of the same object. One essential property of TR is that it is not necessarily constant within a given image. The stated quantity therefore usually refers to the value in the isocenter (i.e., at the intersection point of the image plane with the rotation axis of the CT system) [1], [2]. For single-source CT (SSCT) and conventional filtered backprojection (FBP) reconstruction, the topic of TR in CT images has been extensively studied (see, e.g., [1], [2] and references therein). For the case of dual-source CT (DSCT) some practical experiments exploring the TR of reconstructed images either in the isocenter or in the full field-of-view (FOV) have been published [3], [4], but to our knowledge a theoretical analysis of raw data usage within the FOV and its implications on the resulting TR has never been performed so far. Another important point that has so far not been studied is the question if “same TR” also necessarily implies “same amount of motion artifacts in the reconstructed images” in the context of SSCT and DSCT. The same question can be raised in the context of iterative reconstruction methods with enhanced TR such as TRI-PICCS [5] and TRIM [6]. The purpose of this paper is to address these topics. We will begin our considerations with an introduction into cardiac CT and a comprehensive review of the TR properties in conventional SSCT. From there, we will discuss the extension of these properties to iterative reconstruction methods. Finally, we will explore the application of our results to DSCT. For brevity and simplicity, our discussions will be limited to two-dimensional data acquisition and reconstruction. However, all our results