University of Utah Research Foundation

General Electric Company

Picker International, Inc.

Other

Grant T Gullberg

Gengsheng Lawrence Zeng

Hugh T Morgan

Carl R Crawford

Chi-Hua Tung

Paul E Christian

Roman Basko

Norbert Joseph Pelc

Gary H Glover

Vladimir Y Panin

Chuanyong Bai

Yi Weng

Yu-Lung Hsieh

A method of constructing a non-uniform attenuation map ( ) of a subject for use in image reconstruction of SPECT data is provided. It includes collecting a population of a priori transmission images and storing them in an a priori image memory ( ). The transmission images not of the subject. Next, a cross-correlation matrix ( ) is generated from the population of transmission images. The eigenvectors ( ) of the cross-correlation matrix ( ) are calculated. A set of orthonormal basis vectors ( ) is generated from the eigenvectors ( ). A linear combination of the basis vectors ( ) is constructed ( ), and coefficients for the basis vectors are determined ( ) such that the linear combination thereof defines the non-uniform attenuation map ( ).

Method and apparatus for image reconstruction using a knowledge set

A method of modeling 3D first-order scatter, non-uniform attenuation, and 3D system geometric point response in an ML-EM algorithm to reconstruction SPECT data is provided. It includes performing an initial slice-to-slice blurring operation ( ) on a volume of estimated emission source data. The volume of estimated emission source data is represented by a 3D array of voxels. A voxel-by-voxel multiplying ( ) of the results from the initial slice-to-slice blurring operation ( ) by a volume of attenuation coefficients yields a volume of effective scatter source data ( ). The volume of effective scatter source data ( ) is voxel-by-voxel added ( ) to the volume of estimated emission source data to produce a volume of combined estimated emission and scatter source data. Finally, a secondary slice-to-slice blurring operation ( ) is performed on the volume of combined estimated emission and scatter source data.

Projector/backprojector with slice-to-slice blurring for efficient 3D scatter modeling

A diagnostic imaging system includes a Compton camera (14) disposed on a gantry (16). The camera (14) includes linear detectors (30a, 30b) for detecting radiation emanating from a subject to be imaged. A data processor (32) collects and processes radiation data in accordance with the detected radiation. Position and energy resolving circuitry (34) determines positions and energy deposited by photons striking the detectors. A cone projection generator (40) generates cone projection data or cone integrals based on the collected data which determine a possible location of a gamma source of the detected radiation. A conversion processor (41) converts the cone projection data into plane projection data. The conversion processor (41) includes a line integral processor (42) which determines line integrals representing the cone projection data and applies a spherical harmonic expansion to the line integrals. A plane integral generator (43) generates plane integrals such as Radon projection planes based on the line integrals. A reconstruction processor (44) reconstructs an image representation of a region of interest from the subject from the plane projection data using filtered back projection.

Image reconstruction for compton camera including spherical harmonics

A SPECT system includes a Compton camera (14) disposed on a gantry (16). The camera (14) includes linear detectors (30a, 30b) operating without mechanical collimation for detecting radiation emanating from a subject to be imaged. A data processor (22) collects and processes radiation data in accordance with the detected radiation. Position and energy resolving circuitry (24) determines positions and energy deposited by photons striking the detectors. A projection generator (34) generates divergent projections or V-projections based on the collected data which determine a possible location of a gamma source of the detected radiation. A conversion processor (36) converts the V-projections into parallel projection data such as a Radon transformation. A reconstruction processor (38) reconstructs an image representation of a region of interest from the subject from the parallel projection data using filtered back projection. Alternately, two Compton cameras (70, 72) disposed at a selected angle to each other collect the radiation data and V-projections are generated from both cameras. A combination of V-projections are selected from both cameras and are converted into parallel projections before reconstruction of an image representation.

Image reconstruction from V-projections acquired by Compton camera

A SPECT system includes three gam camera heads (22a), (22b), (22c) which are mounted to a gantry (20) for rotation about a subject (12). The subject is injected with a source of emission radiation, which emission radiation is received by the camera heads. Camera head (22a) has a fan-beam collimator (24a) mounted on a radiation receiving face and generates fan-beam data indicative of the received emission radiation. The camera heads (22b) and (22c) each have a cone-beam collimator (24b), (24c) mounted respectively on their radiation receiving face and generate cone-beam data indicative of the received emission radiation. A transmission radiation source (26) is mounted opposite the camera head (22a) having the fan-beam collimator (24a). The fan-beam detector head (22a) further receives transmission radiation and generates fan-beam transmission radiation indicative thereof. A transmission data reconstruction processor (50) reconstructs the fan-beam transmission data. An emission data memory (110) separately stores the fan-beam and cone-beam emission data. Attenuation correction processors (78, 86) correct the emission data in accordance with the reconstructed attenuation data. An emission data reconstructor (72) reconstructs the corrected emission data into a corresponding three-dimensional image representation which is selectively displayed on a display (76) in a human-readable form.

Three-dimensional SPECT reconstruction of combined cone-beam and

A subject (12) in an examination region is injected with a radioisotope that emits radiation. A detector head (18) receives emission radiation projections (28a) from the radioisotope and transmission radiation projections (28b) from a transmission radiation source (22) disposed opposite the subject from the detector head. A volume memory (50) stores an estimated volume image. For each actually collected image emission data projection set, a projector (52) reprojects a set of projection of the volume image from the image memory (50) along each of the same projection directions as the emission data projections. Each projection is rotated (80) and warped (82) such that rays which converge with the same angle as the convergence of the collimator on the detector head become parallel. The layers are each convolved with a point response function (84) weighted in accordance with a depth of the corresponding layer in the volume image and corresponding points are summed (92) to create a reprojected projection. A ratio of each collected projection and the reprojected projection is calculated (54) and backprojected (56) into a volume of correction factors (58). The backprojectioned correction factors for the set of ratios are summed. A memory updating algorithm (76) multiplies the estimated volume image in the image memory (50) by the sum of the correction factors (58). This process is repeated iteratively over a plurality of projection directions, each iteration further refining the volume image in the volume image memory (50).

Rotating and warping projector/backprojector for converging-beam

A SPECT camera system has three detector heads (12a, 12b, 12c). Each of the detectors have a fan-beam collimator (14) disposed toward an examination region (10). The detectors receive radiation travelling along a fan of rays from an apex (x.sub.f,y.sub.f) to a planar face of the detectors. The detectors generate electronic data indicative of a location (x.sub.s,y.sub.s) on the detector at which a radiation event is received along a ray q.sub..beta. (s). The detectors in a selected orbit R(.beta.) around a subject within the examination region (10). A backprojector (56) includes a backprojection trajectory calculating processor (60) that calculates a trajectory through image space, hence through an image memory (58), corresponding to each radiation ray q.sub..beta. (s). The backprojector weights (68) each data value (x.sub.s,y.sub.s) with a weighting function, preferably a Jacobian J(r,.phi.,.beta.), and adds (64) each weighted data value to a corresponding memory cell of the image memory (58). An indexer (62) indexes the memory addresses (66) in the image memory to index the memory cells to which the weighted data value is added and indexes the weighting function (54) such that the weighting function changes along the trajectory. With the preferred 1/r weighting function, where r is distance along a radial polar coordinate in image space, the resultant image is blurred. The blurring is removed by deconvolving (70) the blurred image representation in accordance with 1/r to generate a corrected image stored in a corrected image memory (80).

Displaced center-of-rotation fan-beam tomography for cardiac imaging

A SPECT camera (A) includes a plurality of detector heads (12a, 12b, 12c) which rotate around an examination region (10). A reconstruction processor (50) reconstructs output projection data from the detector heads into a three-dimensional image representation which is stored in an image memory (62). Selected data from the image memory (62) is converted to a human-readable display on a video monitor (66). To adjust automatically for detector head misalignment, a calibration phantom (B) which has two orthogonal lines of scintillators (20.sub.1, 20.sub.2, 20.sub.3, 20.sub.4, 20.sub.5) arranged orthogonal to each other with a common scintillator in both lines is positioned in the examination region. As the detector heads rotate around the phantom, a misalignment parameter generator (40) generates the displacement distance of the actual rotation from the theoretical rotation axis (.zeta.), the actual rotation radius (R), a projection of the location of the focal point on a detector plane of the detector head in the transaxial direction (C.sub..xi.), and a focal length of the collimator (F). A correction processor (42) corrects the reconstruction processor (50) in accordance with the generated misalignment parameters. In the preferred embodiment, the correction processor provides a revised weighting function to the reconstruction processor and modifies its backprojection algorithm.

Electronic calibration of single photon emission computed tomography

Radiation passing through a cone beam collimator is received by a radiation detector (10) such as a gamma camera head, as the gamma camera head is moved in a helical orbit. Data g(n,u,v) collected during the helical orbit is scaled (42) to scaled data G(n,u,v). A first partial derivative .differential.G (n,u,v)/.differential.u is taken (46u) with respect to a horizontal direction and a second partial derivative .differential.G (n,u,v)/.differential.v is taken (46v) with respect to a vertical direction. The partial derivatives are linearly combined (48) by being multiplied by sine and cosine values of an angle .alpha. between the horizontal direction u and an arbitrary direction p in the detector plane to form partial derivatives .differential.G(n,u,v)/.differential.p. The coordinate system of the derivatives is converted (60) from the (n,u,v) coordinate system to an (n,.alpha.,p) coordinate system. The first derivatives are projected (62) i.e. summed row by row, onto a q axis which is perpendicular to the p arbitrary direction. The one-dimensional projection arrays are rebinned (70) to form first derivative Radon domain data R'(.theta.,.rho.). A second derivative R' (.theta.,.rho.) is taken (74) of the Radon domain data. The second derivative Radon domain data is backprojected (80) into an image memory (82) and displayed on a video monitor (84).

Cone beam reconstruction using helical data collection paths

A SPECT system includes three gamma camera heads (22a), (22b), (22c) which are mounted to a gantry (20) for rotation about a subject (12). The subject is injected with a source of emission radiation, which emission radiation is received by the camera heads. Transmission radiation from a transmission radiation source (30) is truncated to pass through a central portion of the subject but not peripheral portions and is received by one of the camera heads (22a) concurrently with the emission data. As the heads and radiation source rotate, the transmitted radiation passes through different parts or none of the peripheral portions at different angular orientations. An ultrasonic range arranger (152) measures an actual periphery of the subject. Attenuation properties of the subject are determined by reconstructing (90') the transmission data using an iterative approximation technique and the measured actual subject periphery. The actual periphery is used in the reconstruction process to reduce artifacts attributable to radiation truncation and the associated incomplete sampling of the peripheral portions. An emission reconstruction processor (112) reconstructs the emission projection data and attenuation properties into an attenuation corrected distribution of emission radiation sources in the subject.

Simultaneous transmission and emission converging tomography

A SPECT system includes three gamma camera heads (22a), (22b), (22c) which are mounted to a gantry (20) for rotation about a subject (12). The subject is injected with a source of emission radiation, which emission radiation is received by the camera heads. A reconstruction processor (112) reconstructs the emission projection data into a distribution of emission radiation sources in the subject. Transmission radiation from a radiation source (30) passes through the subject and is received by one of the camera heads (22a) concurrently with the emission radiation. The transmission radiation data is reconstructed into a three-dimensional CT type image representation of radiation attenuation characteristics of each pixel of the subject. An attenuation correction processor (118) corrects the emission projection data to compensate for attenuation along the path or ray that it traversed. In this manner, an attenuation corrected distribution of emission sources is generated.

Radiation passing through a cone-beam collimator is received by a radiation detector, such as a gamma camera head, as the gamma camera head is moved in a circular orbit and in a line orbit. Data collected during the circular orbit is stored (42c), transformed (50c) into the frequency domain and redundant data removed (52c, 52c), and transformed (56c) back to the spatial domain. The data collected during the line orbit is stored (42l). The line orbit data is transformed to the frequency domain and repeatedly filtered with a family of filter functions (52l, 54) to remove redundant data. Each filter function corresponds to a different row through the examination region. The filtered frequency domain slice data sets (62.sub.1, 62.sub.2, . . . , 62.sub.n) are transformed (56l) back to the spatial domain and transferred to a central portion of a spatial domain memory (58l). Empty memory cells of the memory (58l) are filled with zeros. The spatial domain data from memories (58c b, 58l) are filtered (72c, 72l), backprojected (74c, 74l), and summed (76) into a three-dimensional image representation that is stored in an image memory (78).

Cone beam reconstruction using combined circle and line orbits

A method and means is presented to estimate the contour of a body to compensate for attenuation in the production of a reconstructed image in fan beam computed tomographic systems having a detector system for taking a plurality of projections of a scanned body at a plurality of angles. The method and means defines a set of variable parameters representing an arbitrary ellipse and projects the arbitrary ellipse into fan beam projection space to determine the expected edges of the arbitrary ellipse at each angle in terms of the variable parameters. A series of fan beam projections is taken having sufficient information for reconstructing a cross-sectional image of the scanned body, and the fan beam coordinates of the apparent body edges in each of the projections are determined. The expected edges of the projected ellipse are fitted to the determined apparent body edge coordinates for each projection by iteratively optimizing all of the variable parameters of the arbitrary ellipse to determine a final set of parameters which defines an elliptical contour which best fits the scanned body. A cross sectional image of the body using the plurality of projections is then reconstructed and compensated by assigning attenuation values to the portion of the reconstructed image encompassed by the elliptical contour.

Method and apparatus for estimating elliptical body contours in fan beam

Method and means for determining and compensating for a shift in the center of rotation of a fan beam CT reconstruction apparatus. Known reconstruction algorithms for fan beam computed tomography are based on the assumption that the center of rotation of the source or detector is on the midline of the fan. In certain cases due, for example, to machine inaccuracies, the center of rotation can be shifted from the assumed center, and such shift can cause artifacts in the reconstruction. The present invention provides a procedure for estimating the magnitude of the shift and also accommodates for the shift by applying a series of weighting factors to the projection information which are dependent on the magnitude of the shift. The weighted projections are then processed by the usual convolution operation, and the convolved projection sets back projected using a geometry modified to account for the shift. Efficiencies are achieved in that all projection sets can be weighted and convolved before back projection, and existing back projection hardware or software can be used to map the modified projection sets into the pixelized image space.

Method and means of correcting for a shift in the center of rotation of

A method for shimming a magnet in a nuclear magnetic resonance (NMR) system employs measurements of the magnetic field obtained from chemical shift imaging (CSI) of a homogeneous phantom. The data are acquired, in one embodiment, in four planes parallel to, and rotated about, the axis of the magnet. A special volume subset of data are expanded in an orthonormal basis set (associated Legendre polynomials in one embodiment) for N+1 separate experiments, where N is the number of shim coils to be adjusted. The result is an N.times.N matrix of calibration coefficients for the shim coil set and a vector of shim current changes which must be applied to null out the measured field inhomogeneity.

Growing community of inventors

Salt Lake City, UT, United States of America

Grant T Gullberg