X-ray or gamma ray systems or devices – Specific application – Computerized tomography
Reexamination Certificate
2000-09-07
2003-02-18
Bruce, David V. (Department: 2882)
X-ray or gamma ray systems or devices
Specific application
Computerized tomography
C378S008000
Reexamination Certificate
active
06522712
ABSTRACT:
BACKGROUND OF THE INVENTION
This invention relates to generation of computed tomographic images. More particularly, the invention relates to reconstruction of such images through use of interpolated projection views generated by interpolating between existing projection views.
Computed tomography (CT) is an imaging technology in which an array of detectors generates data from energetic rays transmitted through or emitted from an imaged subject. For example, a transmission-type medical CT imaging system uses an array of x-ray detectors to detect an x-ray beam that has passed through the body of a human or animal subject. The attenuation of the beam by the subject causes the collected data, including the effects of the attenuation, to contain information about the interior structure of the subject. The detector data are processed by a computer system to generate image data representing a recognizable view of that interior structure.
CT techniques are valuable in a wide range of application areas where noninvasive and nondestructive examination of internal structures is needed. Medical applications include imaging of emissions from radioactive substances introduced into the subject (single photon emission CT, positron emission CT, etc.), as well as x-ray transmission CT. Non-medical applications include, for example, non-destructive testing and inspection, mineral deposit mapping (microseismic CT imaging), and three-dimensional image generation in electron microscopy.
A CT system generates a display image from data representing measurements of energetic signals transmitted through or emitted from a subject in a range of directions. This process is “tomographic” in that structural details of a subject are represented as a cross-sectional view along a given plane through the subject. The process is “computerized” because the raw detector data indirectly represent a view of the subject. Intensive data processing is required to convert the raw data into a recognizable view of the internal features of the subject.
Computer processing of the collected data is performed because the data correspond to mere projections of the subject structure along various different paths. The term “projection data” will be used herein for such CT data, irrespective of the application area in which the CT imaging procedure is used. The differences between the projection data for different paths through the imaged subject, in relation to the spatial separation of the paths, indirectly contain information about the interior structure of the object.
A CT reconstruction algorithm is applied to the projection data to generate image data that specify the image to be displayed. The image data are generated from the structure information indirectly represented by the projections. This is possible in view of the Fourier Slice Theorem, which states generally that a projection view of an imaged subject is related by Fourier transformation to the spatial structure of the subject. More specifically, and with reference to the canonical form of the Fourier Slice Theorem for parallel beam projection data, the Fourier transform of a parallel projection at a given view angle &thgr; is equivalent to a one-dimensional “slice” of the two-dimensional Fourier transform of the imaged subject, taken at the same angle &thgr; in the frequency domain.
The Fourier Slice Theorem allows the image to be reconstructed by Fourier transforming the projection views, assembling the transforms into a two-dimensional Fourier transform, and applying inverse Fourier transformation to the result. Of course, in practical applications this process may be implemented by some form of filtered back-projection or (for diffracting sources) filtered back-propagation. The salient point is that the reconstruction process relies upon the projection views that are available. It is a commonplace fact that the resolution of the generated image will depend on the spacing of the projection views in &thgr;. the more projection views that are taken, the greater the resolution of the resulting image.
This dependence in CT imaging on data resolution in &thgr; (angular direction) is the source of a persistent problem that will be termed here the “missing projection views” problem. An insufficient number of projection views for a desired level of image resolution can arise in many situations. For example, the data for some projection views may be corrupted, or the level of desired resolution may exceed the maximum resolution possible with the given discretization in &thgr;.
A typical solution for data corruption is to repeat the entire data acquisition process, i.e., to throw out the entire set of projection data. Further, where &thgr; discretization has been too coarse, the simplest solution has been to collect more projection views. More data, to provide a finer level of resolution in &thgr;, requires more intensive computation and also entails longer data acquisition cycles. In some situations, constraints imposed by the subject to be imaged, the imaging environment, or the computational resources of the reconstruction engine make denser data collection in &thgr; (i.e., finer resolution in &thgr;) an unattractive solution at best and frequently one that is unfeasible in practice.
The problem of missing projection views is particularly acute when the subject being imaged, or a relevant part thereof, is in motion. For example, CT x-ray imaging is currently being used for cardiac imaging to detect arterial calcification, which can be an indicator for coronary artery disease. X-ray CT is one alternative among various non-invasive techniques for generating images of a patient's beating heart. Other modalities include Doppler ultrasound, fluoroscopy, magnetic resonance imaging (MRI), and electron beam tomography (EBT). Each modality has its own advantages and disadvantages for a given application context.
For cardiac imaging, x-ray CT has the potential to provide finer spatial resolution in three dimensions, as compared with MRI (especially with multi-row spiral CT). However, current third and fourth generation CT scanners suffer from coarse temporal resolution because of limited gantry speed. This limitation on temporal resolution directly affects the available spatial resolution of the resulting images of the heart. More specifically, in a data acquisition session of practicable length, each phase or distinct configuration of the beating heart will correspond to a few projection views. For any given phase, therefore, the &thgr; discretization will be unduly coarse and will limit the resolution of the resulting image to an unacceptably coarse level.
Recent developments in image reconstruction algorithms have improved this situation somewhat by enabling CT techniques to achieve much better time resolution than with the standard operation mode. The basic approach of these improved techniques has been to use consistent projection data, collected from multiple heart cycles, and to perform retrospective data rebinning. Such data rebinning methods have been shown to improve significantly the time resolution of the collected data. In practice, however, existing rebinning methods have been found to introduce some image artifacts due to inconsistencies in the data.
It is therefore apparent that the problem of missing projection views, in cardiac imaging and in other applications of CT imaging techniques, has remained as a persistent obstacle to wider use of CT imaging. In the particular context of cardiac imaging, physicians desire a “freeze frame” capability for medical imaging systems. Such a capability would permit an image to be accurately reconstructed to represent the patient's heart at a selected part of the cardiac pumping cycle.
More generally, what has been needed is a method and system for processing CT projection data to overcome or minimize the effects of missing projection views. Such a method desirably would provide acceptable spatial resolution in the resulting images, without requiring extensive additional data acquisition. In imaging contexts where
Cesmeli Erdogan
Yavuz Mehmet
Bruce David V.
Cabou Christian G.
General Electric Company
Ingraham Donald S.
LandOfFree
Reconstruction of computed tomographic images using... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Reconstruction of computed tomographic images using..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Reconstruction of computed tomographic images using... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3142862