Four-Dimensional Imaging of the Heart Based on Near-Isotropic MDCT Data Sets

Abstract

Lawler LP, Ney D, Pannu HK, Fishman EK our-dimensional cardiac CT is a concept in evolution [1, 2]. In this paper, we define 4D cardiac CT as a processed study that assimilates a series of sequential, static, phase-specific, 3D volume helical data sets into a cine image that reflects the in vivo model of the 3D spatial information mapped to the sequential time and motion relationship of the cardiac cycle. It can be applied to derive morphologic and functional information, but it is distinct from existing tools, which deduce such information from static planar slices of end systole and end diastole alone. This technique will form an important facet of the comprehensive cardiac CT study. The initial relatively high-speed acquisition of electron beam CT (EBCT) [3] provided some of the earliest quantitative CT information on ventricular size and shape and on systolic function. However, this complex technology was never widely available, was limited to prospective gating, and is fast being replaced by the the more versatile mechanical CT where its acquisition parameters in terms of temporal resolution are rapidly approaching those of EBCT. Earlier work with single-detector helical scanning was able to produce animated 2D images of the heart with ventricular values that closely correlated with conventional ventriculography [4]. Materials and Methods Data Acquisition and Processing To generate a 4D data set, one must first acquire a density profile and positional information that characterize each voxel of the heart structure for a series of time points throughout the cardiac systolic and diastolic phases. The patient is scanned in a single breath-hold with simultaneous recording of the ECG signal. Current helical MDCT systems provide near-isotropic voxels from 0.75-mm detector systems (Sensation, Siemens Medical Solutions) giving 0.75to 1-mm slice collimation reconstructed at 50% overlap. Large detector arrays, fast gantry rotation, and segmental reconstruction contribute to a TR of 120–130 msec. Vascular lumen, endocardial surface, and chamber contrast resolution is achieved through a peripheral-venous upper-extremity power injection of 120 mL of nonionic iodinated contrast (Visipaque 320, Amersham Health) delivered at a rate of 3 mL/sec followed by a saline flush from a dual-head injector. Scanning acquisition begins 17 sec after the contrast administration. Unlike coronary CT angiography, which uses only diastolic data, 4D CT utilizes all data derived from irradiating the patient. Effective dose depends on heart rate but is approximately 7.0 mSv for a man and 10.2 mSv for a woman, with a CT dose index of 42.0 mGy. Axial planar data for 3D and 4D postprocessing are reconstructed from the raw data using retrospective gating, which defines a portion of the cardiac cycle as a percentage of the R-R interval. For 4D imaging, a contiguous sequence of nine or more separate “whole-heart” volumes, of equal duration intervals, is generated. These volumes represent the distinct phases (i.e., “time windows”) of cardiac motion (e.g., 10%, 20%, 30%, and so forth up to 90% of the R-R interval). This is a semiautomatic standard process of all cardiac software and is performed at the scanner. The user simply defines the percentage reconstruction desired, and the result is a series of separate stacks of axial images with each stack representing a particular period of the cardiac cycle. Within each time window, all voxels of the heart are represented. The total number of images may be more than 2,000 images, depending on the slice reconstruction used (0.75–1 mm). The data are sent to a 3D workstation that incorporates commercially available volume-rendering software and supports a work-in-progress version of 4D software.