Coronary artery disease: improved reproducibility of calcium scoring with an electron-beam CT volumetric method.

PURPOSE: To assess the variability and reproducibility of a volumetric calcium score calculated with electron-bean computed tomographic (CT) scans of coronary arteries. MATERIALS AND METHODS: Two sets of electron-beam CT scans were obtained in patients with coronary calcification (group A) or known risk factors for coronary arterial disease (group B). The second set or scans was obtained after a brief interval (group A, n = 52) or after 1 year with no risk modification (group B, n = 27). Traditional (plaque area and attenuation) and volumetric calcium scores were calculated for each patient and lesion. RESULTS: The median percentage change for individual lesions in group A was 13% for the volumetric and 19% for the traditional score. The overall reduction in error with the volumetric score was 40% (P < .001). The median percentage change for group A patient totals was 9% for the volumetric and 15% for the traditional score (P < .001). In group B patients, the median volumetric score increased by 44% after 1 year. CONCLUSION: The volumetric score showed better reproducibility than the traditional score, and its variability was considerably less than the score increase in untreated patients after 1 year. The reproducibility of the volumetric method makes it useful for assessing the progression of coronary arterial disease on serial electron-beam CT studies.

A PC-based 3D imaging system: algorithms, software, and hardware considerations.

Three-dimensional (3D) imaging in medicine is known to produce easily and quickly derivable medically relevant information, especially in complex situations. We intend to demonstrate in this paper, that with an appropriate choice of approaches and a proper design of algorithms and software, it is possible to develop a low-cost 3D imaging system that can provide a level of performance sufficient to meet the daily case load in an individual or even group-practice situation. We describe hardware considerations of a generic system and give an example of a specific system we used for our implementation. Given a 3D image as a stack of slices, we generate a packed binary cubic voxel array, by combining segmentation (density thresholding), interpolation, and packing in an efficient way. Since threshold-based segmentation is very often not perfect, object-like structures and noise clutter the binary scene. We utilize an effective mechanism to isolate the object from this clutter by tracking a specified, connected surface of the object. The surface description thus obtained is rendered to create a depiction of the surface on a 2D display screen. Efficient implementation of hidden-part removal and image-space shading and a simple and fast antialiasing technique provide a level of performance which otherwise would not have been possible in a PC environment. We outline our software emphasizing some design aspects and present some clinical examples.

Shape-based interpolation of multidimensional objects.

A shape-based interpolation scheme for multidimensional images is presented. This scheme consists of first segmenting the given image data into a binary image, converting the binary image back into a gray image wherein the gray value of a point represents its shortest distance (positive value for points of the object and negative for those outside) from the cross-sectional boundary, and then interpolating the gray image. The set of all points with nonnegative values associated with them in the interpolated image constitutes the interpolated object. The method not only minimizes user involvement in interactive segmentation, but also leads to more accurate representation and depiction of dynamic as well as static objects. The general methodology and the implementation details of the method are presented and compared on a qualitative and quantitative basis to the existing methods. The generality of the proposed scheme is illustrated with a number of medical imaging examples.

Low-level segmentation of 3-D magnetic resonance brain images-a rule-based system.

A rule-based, low-level segmentation system that can automatically identify the space occupied by different structures of the brain by magnetic resonance imaging (MRI) is described. Given three-dimensional image data as a stack of slices, it can extract brain parenchyma, cerebro-spinal fluid, and high-intensity abnormalities. The multiple feature environment of MR imaging is used to comput several low-level features to enhance the separability of voxels of different structures. The population distribution of each feature is considered and a confidence function is computed whose amplitude indicates the likelihood of a voxel, with a given feature value, being a member of a class of voxels. Confidence levels are divided into a set of ranges to define notions such as highly confident, moderately confident, and least confident. The rule-based system consists of a set of sequential stages in which partially segmented binary scenes of one stage guide the next stage. Some important low-level definitions and rules for a clinical imaging protocol are presented. The system is applied to several MR images.