HALF FIELD OF VIEW REDUCED-SIZE CT DETECTOR
CROSS-REFERENCE TO RE.LATED APPLICATIONS
This Application claims priority to and the benefit of the filing dates of provisional applications having Serial Nos. 60/129,398 and 60/166,500 filed on April 15, 1999 and November 19, 1999, respectively, which are incorporated herein by reference in their entirety.
BACKGROUND OF THE INVENTION
The present invention relates to method and apparatus for use in a volumetric computed tomography (VCT) system that utilizes a reduced-sized area detector that covers only half of the field of view, thereby enabling the size and costs of the area detector to be reduced without increasing, or substantially increasing, artifacts.
Computed tomography (CT) is a technique that generally involves subjecting a patient to X-rays, acquiring digital X-ray projection data of a portion of the patient's body, and processing and back-projecting the digital X- ray projection data to produce an image that is then displayed on a display monitor of the CT system. CT systems typically comprise a gantry, a table, an X-ray tube, an X-ray detector array, a computer and a display monitor. The computer sends commands to controllers of the gantry to cause the gantry to rotate the X-ray tube and/or the detector array at a particular rotational speed.
In third generation CT systems, relative rotational motion is produced between the gantry, partly comprised by the detector array and the X-ray
tube, and the patient's body. As this relative rotational motion is produced, the computer controls the data acquisition process performed by the X-ray tube and the detector array to acquire digital X-ray radiographs. The computer then processes and back-projects the digital X-ray radiograph data by performing a reconstruction algorithm and displays the reconstructed CT image on the display monitor.
Many CT systems in use today utilize a single row of detectors in the gantry, which is normally referred to as a linear array of detector elements. More advanced CT systems use two to four linear arrays of detectors to construct a multi-row detector. Although both detector arrangements can be used with a helical scanning protocol, the multi-row detector facilitates patient scanning since a specified axial coverage of the patient can be scanned in less time by increasing the helical pitch of the detector array. Helical pitch is typically defined as the ratio of the displacement of the table supporting the patient during one rotation of the gantry to the detector pitch. For example, a helical pitch of one refers to translating the patient table an amount equal to the detector pitch during one revolution of the CT gantry of the CT system.
Normally, the linear detector or the multi-row detector arrays cover the entire field of view of the X-ray fan beam emitted by the X-ray source. In other words, the X-rays that pass through or illuminate the area of the object being scanned, which may or may not be a patient, are absorbed by the detector array.
In some CT imaging systems, it is desirable and may, in some instances, be necessary to reduce the size of the detector array. For instance, a recent development in CT technology is the use of area detector arrays, composed of many rows of linear detector arrays, for CT data acquisition. At present, a detector panel that can cover the entire field of view
or extent of the patient being imaged is not yet available. In addition, some systems that use linear detector arrays support a very large field of view for the patient being scanned. It is desirable to reduce the size and costs of the detector array in this situation as well. One technique that has been utilized to overcome these limitations is to translate a detector array that is smaller in size by half of the width of the detector array. For example, suppose that the original size of the detector array should be 80 cm to cover the desired field of view of the patient. A smaller detector, that may have a width equal to half of the width of the original detector, 40 cm in this case, may be used. This detector is offset by half of its width, or 20 cm in this example, such that it covers approximately one half of the field of view of the CT imaging system. In this example, the same field of view in the patient is obtained by a detector that has a width equal to half of its original value.
It is also possible to increase the field of view of a system that has a detector with a fixed width. Normally the projection of the center of rotation of the CT imaging system is aligned with the center of the detector panel. The center of rotation in a CT imaging is the physical location of the point which the X-ray source and the detector array rotate about. However, the field of view (FOV) of this system can be increased by offsetting the detector by half of its width with respect to its original position.The projection of the center of rotation of the imaging system is near an edge of the linear or multi-row detector that has been shifted, although the detector still measures projection data from X-rays that pass through the physical center of rotation (i.e., at the ISO) of the imaging system. This arrangement, in turn, effectively doubles the field of view of the original imaging system configuration, which can result in a large increase in the field of view of the imaging system. The system configuration of shifting the detector by half of its
width is typically referred to half-detector shift.
In a fan-beam CT system, a CT system where the X-ray source is a point which emits an angular aperture of X-rays that illuminate only the detector panel and resemble a fan, it is necessary to acquire projection data for a portion of the full rotation of the CT gantry. Specifically, it is necessary to acquire projection data while the gantry rotates a angular region about the patient equaling 180° plus the fan angle. Once again: the fan angle is the measure of the angular aperture of X-rays that illuminate only the detector array in the axial plane of the imaging system. Clearly, since it is not necessary to measure projection during the full 360° rotation of the gantry about the patient, some of the projection data must be redundant.
In the half-detector shift configuration of the CT system, data is acquired for a full 360° rotation of the gantry. At each view angle or the gantry, only one half of the projection data is measured. Data from other views of the gantry are used to complete the projection data for any given view angle. It is known in the art the manner in which this process is accomplished. However, when the measured projection data covering half of the field of view of the imaging system is combined with the data generated from other views of the gantry, the resulting projection data may not match well near the center of the projection data. These mismatches, if not reduced or eliminated, may produced undesirable artifacts in the reconstructed images
A technique that is currently being used to mitigate the artifacts that are caused by the discontinuities of projection data within the field of view utilizes a weighting function to smooth the discontinuities in the data within a transition region. This technique requires that the detector has some additional detector elements that extend past the projection of the center of rotation of the imaging system on the detector. As the gantry rotates 360° about the patient, the region of the shifted detector panel that extends slightly
past the projection of the center of rotation on the detector in both directions is called the transition region. Actual data is measured by the detector in half of the transition region, and data can be generated from alternate views of the gantry in the second half of the transition regionData within the transition region are multiplied by a weighting factor that smoothes the discontinuities. In general, a larger transition region results in better image quality, but also leads to higher system costs since the field of view of this system configuration is slightly less than that produced with the half-detector shift configuration. A need exists to improve the integration of the measured data and the fabricated data within the transition region so that the full field of view the detector array can be realized.
In a volumetric CT system utilizing a half-detector shift configuration, projection data is measured within half of the field of view of the imaging system while the other half of the projection radiograph must be generated from opposing rays. Unfortunately, projection data measured at other projection angles of the CT gantry do not have the same orientation as the rays that would have been measured had the detector been twice its original width and not offset. Accordingly, a need exists for a VCT system that utilizes an area detector in a half-detector shift configuration and that realizes the benefits thereof, and which overcomes the aforementioned difficulties.
BRIEF SUMMARY OF THE INVENTION
A computed tomography (CT) system for obtaining projection data of an object, which comprises an X-ray source, and a detector. The detector is shifted by half of its width relative to a center position which corresponds to the projection of a center of rotation of the CT system onto the detector. In accordance with the method of the present invention, for each projection view, a detector element value, Va, is chosen of a detector element that is closest to the an ISO center of the CT system. Then, for the chosen detector element, a value, Nb, for the detector element is
estimated from an opposing direction or from a forward projection in the same direction. A smoothing function capable of eliminating a discrepancy between Va and Vb is then selected. The smoothing function is then applied to eliminate the discrepancy between Va and Vb. A weighting function is then applied to pave the step difference when true projection data and estimated projection data are combined to create a smooth transition region.
BRIEF DESCRIPTION OF THE DRAWINGS
Fig. 1 is a block diagram of the CT system of the present invention. Fig. 2 illustrates the detector offset utilized in accordance with the method of the present invention.
Fig. 3 is a block diagram illustrating the method of the present invention in accordance with the preferred embodiment.
DETAILED DESCRIPTION OF THE INVENTION
Prior to describing the method and apparatus of the present invention, a general discussion of the VCT system of the present invention will be provided with reference to Fig. 1. FIG. 1 is a block diagram of a volumetric CT scanning system that is suitable for implementing the method and apparatus of the present invention. The Volumetric CT scanning system will be discussed with respect to its use in reconstructing an image of an anatomical feature of a patient, although it will be understood that the present invention is not limited to imaging any particular object. The present invention may also be used for industrial processes, as will be understood by those skilled in the art. In addition, the present invention is not limited to medical CT equipment, but includes industrial systems where the X-ray source and detector geometry are held fixed while the object is rotated during the scanning time.
In a volumetric CT scanning system, the gantry is rotated about an object, such as a human patient, and projection data are acquired. A computer 1 controls the operations of the volumetric CT scanning system. When referring herein to the
rotation of the gantry, that phrase is intended to denote rotation of the X-ray tube 2 and/or rotation of the detector 3. which preferably is a high resolution area detector. The X-ray tube 2 and the area detector 3 are comprised by the gantry. The controllers 4A and 4B are controlled by the Volumetric CT scanning system computer 1 and are coupled to the X-ray tube 2 and to the detector 3, respectively. The controllers 4A and 4B cause the appropriate relative rotational motion to be imparted to the X-ray tube 2 and/or to the detector 3. Individual controllers are not necessary. A single controller component may be used to rotate the gantry. It should also be noted that the computer 1 controls the variations in image scanning time, image resolution and/or axial coverage in order to implement the methods of the present invention.
The computer 1 controls the data acquisition process by instructing the data acquisition system 6 as to when to sample the detector 3 and by controlling the speed of the gantry. Additionally, the computer 1 instructs the data acquisition system 6 to configure the resolution of the radiographs obtained by the area detector 3, thereby allowing the resolution of the system to be varied. The data acquisition system 6 comprises the read out electronics, as shown.
The area detector 3 is comprised of an array of detector elements (not shown). Each detector element measures an intensity value associated therewith that is related to the amount of X-ray energy that impinges on the detector element. When the apparatus and method of the present invention are incorporated into a volumetric CT scanning system, a new volumetric CT scanning system is created. Therefore, the present invention also provides a new volumetric CT scanning system.
It should also be noted that the present invention is not limited to any particular computer for performing the data acquisition and processing tasks of the present invention. The term "computer", as that term is used herein, is intended to denote any machine capable of performing the calculations, or computations, necessary to perform the tasks of the present invention. Therefore, the computer
utilized to perform the control algorithm 10 of the present invention may be any machine that is capable of performing the necessary tasks
With respect to the present invention, it has been determined that the need for using additional detector elements to cover the transition region can be eliminated by an alternate data smoothing scheme. In addition, an alternative method utilizes an iterative algorithm to estimate the projection data that would have been measured if a large detector array that covered the entire field of view had been used to acquire the data. Both approaches will be discussed within the same context below since errors in the transition region are handle in a similar manner.
This technique forms a set of X-ray projections {P by either forward projecting the reconstructed data obtained from a prior iteration step or by interpolating the redundant detector data obtained from a set of opposing rays, which form another set of projection data in the reversed directions of {Pa}- The technique of forward projection is the process where rays are emitted from a hypothetical X-ray source; these rays traverse the reconstructed volume toward individual detector elements. Along the ray, the linear attenuation value of the reconstructed values along the rays are summed and denoted as the line integral of the line attenuation coefficient. The technique of forward projecting the reconstructed data(denoted as
FPT) is generally suitable for generating projection data corresponding to a larger cone angle (i.e., as is used in VCT systems) while the technique of interpolating redundant projection data(denoted as PDT) is more suitable for projection data that is closer to the mid-plane (i.e., that is closer to the ISO center). Similar to the fan angle, the cone angle refers to the angular extent of the X-ray emitted from the X-ray source in a direction orthogonal to the fan angle direction. The discrepancy between the estimated detector values obtained using the either FPT or PDT and the original values that would be obtained if the data had actually been measured may cause distortions in images close to the ISO center.
To alleviate such distortion, a technique has been developed that utilizes a smoothing function, which can be stated as follows with reference to
Figure 3: 1. For each projection view, choose the known detector element that is closest to the ISO center 21 , which will be referred to hereinafter as Va.
2. For the same detector element, obtain an estimated value 22 (i.e., via interpolation projection data from alternate views(PDT) or from forward projecting the reconstructed data(FPT)), which will be referred to hereinafter as Vb.
3. Generate an appropriate smoothing function 23.
4. A smoothing function reduces the discrepancy between Va and Vb and gradually smoothes this within a region near the center of the field of view of the imaging system 24.
This can be seen from the following discussion: Letting d = Va - Vb which is the amount of discontinuity in the projection data at the center of the field of view. As an example, a possible smoothing function that could be used to gradually smoothen the discrepancy is an exponential function defined as: V = 0.5d e"ax 0
(Equationl) where x0is the absolute value of the distance from the detector location corresponding to the detector element value Va and a is a factor controlling the slope of the curve associated with the smoothing function. The exponential function is added/subtracted to projection values located to one side of the central ray location(the detector location corresponding to the projection of the center of rotation in the imaging system onto the detector) to bring the lower/higher estimated values up/down and is subtracted/added to projection values to the alternate side of the central ray location to reduce/increase the higher/lower original values. In other words, it provides a
method to reduce the inconsistency of the projection data at the central ray when the true projection data and the estimated projection data are combined together; hence, the process provides a smooth transition region in the data which reduces or eliminates artifacts 25. Presently, is not known in the prior art how to use an area detector within a
Volumetric CT (VCT) system in a half-detector shift configuration. Having previously described a VCT system and the different known techniques for eliminating discontinuities of projection data within the field of view of the imaging system by creating a transition region, another aspect of the present invention will now be described.
The variables fθ and fθ. will be used to denote two functions representing the signal strength of the forward and opposing rays(rays at the same angular orientation but traversing in opposite directions), respectively, obtained at source angle θ where fθ(n) = 0 for N2 < n < N (Equation 2) f (n) = 0 for 1 < n < N2 (Equation 3)
Ideally fθ(N2) should be exactly the same as fθ (N?) because both traverse the same portion of the object. But this can never happen due to the following reasons:
(a) The actual shape of each ray is a shallow tetrahedron originating at the source and ending at the detector element. There are no identical rays traversing exactly the same part of the object unless the object is totally homogeneous and circularly symmetrical.
(b) The object/patient motion during the scan cycle may introduce additional errors in each forward/opposite ray pair. (c) No perfect, efficient interpolation scheme has, to date been developed.
That is, the process of interpolation may introduce errors.
Assume that d(θ) is the difference between fθ(N2) and fθ-(N2) at the angular source position θ. If d(θ) is totally random, then an error introduced in the reconstructed image probably will be obscured by the quantum noise associated with
other CT random errors. However, if the error is somewhat systematic, it will introduce obvious artifacts in the reconstructed image. For this reason, a smoothing process must be utilized at the transition region. In other words , the smoothing function was developed to make the step error of fθ and fθ- smaller around the central ray location denoted by detector element N2.
W and W' will be used to represent the smoothing function for fθ and ffi-, respectively. When deriving W and W', certain criteria must be considered, which will be understood by those skilled in the art. Also, a variety of smoothing functions are suitable for this purpose, other than those specifically set forth herein, as will be understood by those skilled in the art. For example,
W(n) + W' (n) = 1 for all n (Equation 4) δW = δW = 0 at n = N2 ± Δn (Equation 5) δn δn
where Δn sets the extent of smoothness of W and W and δ is a differential operator. It should be noted that the conventional smoothing function for this type of application is also generally known as a feathering function. W and W' are used to find a transition region between the forward and opposing rays. It should be noted that, in order to make the smoothing function work, Δn must be an integer larger than zero. In fact, the larger Δn is, the better the smoothing function will work. However, increasing Δn too much could require that additional detector elements be used to extend beyond N2, which is the detector element at the central ray location. Therefore, Δn should be selected so that it is sufficiently large, but not so large as to require that additional detector element be added to the transition region. As such, fθ and fθ> have the following limits: fθ(n) = 0 for N2 + Δn < n < N (Equation 6) fθ.(n) = 0 for 1 < n < N, - Δn (Equation 7)
The actual detector signals being used in the back projection process are Wfθ and Wfθ . It should also be noted that, since a broader transition region tends to eliminate a lot of mis-matching errors between the forward and opposing rays, there is no need to compose opposing rays for each half field of view(FOV) projection data. In other words, each half FOV data (plus additional Δn detector values) are padded with zeroes to obtain detector data of length N, followed by the conventional filtered projection procedure. No interpolation is involved.
The incentive is that when one starts to increase the number or detector rows in a CT system, i.e. in an area detector, the additional detector elements (Δn times the number of rows) becomes larger. Thus, it is necessary to improve the conventional approach by devising a method and apparatus where Δn can be minimized.
The approach in this invention reduces Δn to 1 while a computer simulation showed that the conventional smoothing method requires Δn to be around 20 to achieve comparable artifact level. The advantages are even more significant in VCT applications where the number of detector elements needed in the transition region for an area detector may be three orders of magnitude more than elements needed in a linear array.
There is a possible systematic error in the forward and opposing rays where the latter is obtained through interpolation or forward projection. A step error can be measured by finding the difference between fθ(N2) and fθ-(N2). Let d(θ) = fθ (N2) - fθ-
(N2). d(θ) is viewed as a step error between fθ( N2) and fθ (N2) where θ is the angular position of the x-ray source.
Our approach is to smooth out the step error using an exponential function as described in Equation 1 where a is a control factor controlling the smoothness of the exponential function. The forward and opposing ray functions, fθ(n) and fθ (n) are converted into two other respective functions. gθ(n) and gθ (n) as follows: gθ(n) = fθ(n) - pθ(N? - n) (Equation 8) gβ-(n) = fθ.(n) + pθ(n - N2) (Equation 9) Equation 8 and Equation 9 make gθ (N2) = gθ (N2). For each projection image, the following procedure is implemented:
1. Obtain the original half FOV projection data calling it fθ (n), with zero- padding according to Equation 2.
2. Obtain a set of opposing rays fθ (n) and do zero-padding according to Equation 3. 3. Apply a smooth function based on the step error d(θ) (where d(θ) = fθ
(N2) - fθ-(N2) )according to Equation 8 and Equation 9.
4. Integrate gθ(n) and gθ (n) to form a N-detector data set, say hθ(n), where hθ(n) = gθ(n) for N2 < n< N and he(n) = gθ-(n) for Kn< N2
5. Apply the conventional filtered-back projection to hθ(n). 6. Repeat steps 1 through 5 for all projection angles.
The above procedures are valid for any CT scanner where opposing rays can be obtained through interpolating other projection data out of different angles. A perfect case occurs in 2D fan beam case where another half FOV of data can always be approximately calculated from redundant fan beam projection data. WTien expanding this technique to 3D VCT, the perfect interpolation may occur only on the mid-plane when a circular orbit is used.
Our simulation shows that when the same technique is applied to VCT using a circular orbit, this technique outperforms the conventional smoothing technique (using 20 additional detectors beyond the position of the central ray) for cone angles within ± 1.5 degrees. This result is not true for large cone angles since the opposing rays have an angular orientation that significantly differs from the data that would have been measured if the complete detector had been used. To remedy this situation, an iterative approach can be employed to improve the image quality. The procedure is given as follows: 1. Obtain an initial 3D image according to the steps 1 to 6 presented above. 2. In the second iteration of the approach, use the forward projection method to obtain the "opposing rays" for each half FOV projection data set and combine them into one complete projection data set according to steps 3 through 6 outlined above.
3. Continue step 2 until the process converges, i.e. the quality of the image no longer improves.
It should be noted that the present invention has been discussed with respect to certain embodiments. However, the present invention is not limited to these embodiments. For example, the three scenarios discussed, are not meant to be all inclusive of the manner in which tradeoffs of the aforementioned parameters can be utilized to obtain the proper mode of operation of the VCT system. These scenarios are discussed in order to illustrate the concepts of the present invention and the manner in which these fundamental parameters can be traded off in order to achieve the proper scanning protocol. In addition, the trade-offs are not limited to a scanning protocol, i.e. they apply to both axial scanning (the patient table is not moved during the scanning period) and helical scanning protocols. Those skilled in the art will understand the manner in which these concepts can be utilized and extrapolated to achieve other area detector scanning protocols that are useful for particular applications.