Application-driven optimization of acquisition and reconstruction of SPECT/PET projection data

Abstract

A method for optimizing imaging dwell time during nuclear imaging. A fast pre-scan is conducted over the angular range of imaging, with equal dwell time at each position. A statistical sub-sampling is conducted over a region of interest to a desired fraction of the acquisition time. The statistical sub-sampling is used to simulate a varying dwell time tomographic dataset according to a profile pre-generated by an analysis of patient specific anatomic or functional information. Reconstructed data with equal dwell time and with simulated adaptive dwell time are constructed and compared to generate a difference image, a parameter of interest of which is evaluated. If the parameter of interest is within an acceptable level, the reduced dwell time for the given angular position is established; otherwise, the fraction is varied and the process is repeated. Once dwell times are computed for all view angles, a full scan is conducted.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

The invention will now be described in greater detail in connection with the associated drawings, in which:

FIG. 1 is a diagram illustrating the difference in projection data content for a given dwell time, depending on the view angle or position of a detector with respect to a region of interest of a patient;

FIG. 2 is a flowchart illustrating one possible method for calculating adaptive dwell times in accordance with the invention; and

FIG. 3 is a block diagram of a system for implementing the invention.

DETAILED DESCRIPTION OF EMBODIMENTS OF THE INVENTION

An embodiment 10 of one method in accordance with the concepts of the invention is illustrated in FIG. 2. According to this method, an initial “fast” pre-scan 12 is first conducted for total time T to accumulate statistical projection data. The fast pre-scan is conducted over a defined ROI in the projection image. The ROI may be defined automatically, such as through the use of various known methods such as automatic segmentation methods in heart studies. The ROI also may be defined semi-automatically, such as having a user define an ROI in one projection view, from which a software program can infer ROIs in all other projection views. Alternatively, the ROI may be identified manually by the user in each projection view.

During this initial pre-scan, the dwell time t_i per angular view is held constant over the entire scan range (either π or 2π radians, depending on the type of scan being conducted) such that:

T=Σt_i, i=1, . . . N, N=number of views, t=t_i, ∀i.

For each pixel in the ROI, an analysis of the number of counts acquired for the pixel is analyzed, such that the number of counts per pixel as well as the total counts over the entire ROI may be determined. Once the pre-scan is complete, at step 14 a statistical subsampling for a time T₁, which is a desired fraction ε of the total pre-scan acquisition time T is performed on the acquired pre-scan data, wherein T₁=εT, ε≦1.

Next, at step 16 the statistical subsampling is used to simulate a varying dwell time tomographic dataset according to a profile that has been pre-generated for the patient by an analysis of patient-specific anatomical dataset (e.g., μ-map from a CT scan) or a functional reconstructed dataset (e.g., from the pre-scan); alternatively, time increments may be assumed to be equal (t¹_i=t¹) such that t¹_i=ε_it₁≦t_i, and T₁=Σt¹_i.

Next, at step 18, the pre-scan acquired data is reconstructed to obtain a reconstructed image R, being based on the pre-scan using constant dwell time and at step 20 a second reconstructed image R₁ is obtained from the statistically sub-sampled dataset obtained at step 16. Next, at step 22 a difference image (R−R₁) is generated. Once the difference image is generated, at step 24 an objective value Φ is computed based on some measure of image quality (e.g., L₂), such that Φ=minarg∥(R−R₁)²∥.

It should be appreciated that in accordance with the invention, dwell time can be varied so as to optimize a variety of different image quality parameters, e.g., total image noise power, the probability of lesion detection, etc., and Φ will vary accordingly.

Once Φ has been calculated, at step 26 it is compared to a pre-selected, acceptable level of deviation δ. If Φ≦δ (or if a pre-selected maximum number of iterations has been reached), then ε_i for the given view angle, and hence the reduced dwell time ε_it_i¹ for the given view angle, has been determined. This reduced dwell time value for the given view angle is stored at step 28 and at step 30 the overall process repeats itself for the next view angle until all view angles have been analyzed. Otherwise, if Φ>δ, ε_i is varied at step 32 to a modified dwell time ε_i+Δε and the sub-process repeats at step 34, to obtain a new statistical subsampling of the pre-scan data with the modified dwell time (ε_i+Δε)t_i¹, and an updated or revised image set R₁ is reconstructed at step 20.

Alternatively, if Φ≦δ is determined to be true at step 26, the dwell time fraction ε_i may be reduced by a predetermined increment, and the statistical subsampling recalculated, to determine whether the dwell time can be even further reduced while still maintaining the objective measure of image quality Φ with the acceptable deviation range δ. As a further alternative, it is possible to use variable dwell times for different view angles in the pre-scan instead of a constant dwell time. In this case, the individual modifications to the dwell times as a function of view angle would simply be different than if a constant dwell time over all view angles were to be used.

Once all view angles have been processed in this manner such that a specific dwell time for each angle has been determined, a full scan is conducted using the dwell times calculated for each view angle.

According to another method to determine dwell time so as to minimize noise power, it can be shown that for an imaging acquisition of time T, the total noise power in a reconstructed image for a particular view angle can be minimized by making the dwell time t_i for that view angle proportional to the square root of the count rate r_i at that angle:

$t_i=\frac{\sqrt{r_i}}{\sum \sqrt{r_i}}T.$

FIG. 3 shows an example system which may be used to implement the concepts of the present invention. Camera detector 301 is orbitally positioned about ROI 101 of patient or object 102, and acquires projection data for a particular dwell time. The motion of camera 301 is controlled, for example, by gantry control motor 305. The (x, y, z) projection data (i.e., spatial coordinates and total energy) is inputted to acquisition computer 302. The data is supplied to a dwell time calculator, such as count rate estimator 303, which determines the dwell time for the view angle based on a dwell time calculation method such as that shown in FIG. 2 or explained above with respect to the total noise power. The calculated dwell time is inputted to a unit 304, which determines the speed of the gantry motor based on the calculated dwell time, and sends a control signal to the gantry motor control unit 305 to control the gantry motor accordingly.

It should be appreciated by those having ordinary skill in the art that while the present invention has been illustrated and described in what is deemed to be the preferred embodiments, various changes and modifications may be made to the invention without departing from the spirit and scope of the invention. Therefore, it should be understood that the present invention is not limited to the particular embodiments disclosed herein.

Claims

1. A nuclear medicine tomographic imaging apparatus, comprising: a detector capable of orbiting about a patient to acquire projection data at a plurality of view angles;a control unit that varies the amount of time that said detector remains at a particular view angle in response to a control signal; anda processing unit that develops said control signal in response to a desired measure of image quality.

2. The imaging apparatus of claim 1, wherein said desired measure of image quality is inputted by a user.

3. The imaging apparatus of claim 1, wherein said control signal is developed based on at least one data acquisition parameter associated with said patient.

4. The imaging apparatus of claim 1, wherein said desired measure of image quality is minimized total noise power.

5. The imaging apparatus of claim 1, wherein said desired measure of image quality is maximization of probability of lesion detection.

6. The imaging apparatus of claim 1, wherein said desired measure of image quality is optimization of object contrast in a region of interest.

7. The imaging apparatus of claim 1, wherein said desired measure of image quality is optimization of image quality for multi-modality imaging.

8. The imaging apparatus of claim 1, wherein said processing unit develops said control signal by: performing a pre-scan of said patient for a plurality of view angles over a period of time shorter than a time required for a clinical scan, and obtaining pre-scan data;subsampling said pre-scan data with a varied dwell time to obtain simulated varied dwell time data;comparing a reconstructed image of said pre-scan data with a reconstructed image of said simulated varied dwell time data;determining whether the result of said comparison falls within a predefined acceptable range;modifying said varied dwell time to obtain a modified set of simulated varied dwell time data if the result of said comparison is outside of said predefined acceptable range; andusing said varied dwell time in a clinical scan of said patient if the result of said comparison is within said predefined acceptable range.

9. The imaging apparatus of claim 8, wherein said comparison involves calculating a difference between the reconstructed image of said pre-scan data and the reconstructed image of said simulated varied dwell time data.

10. The imaging apparatus of claim 8, wherein said pre-scan uses a constant dwell time for each view angle.

11. The imaging apparatus of claim 8, wherein said predefined acceptable range is based on said desired measure of image quality.

12. A method of calculating variable dwell times for particular view angles of a nuclear medicine imaging detector with respect to a patient, comprising the steps of: performing a pre-scan of said patient for a plurality of view angles over a period of time shorter than a time required for a clinical scan, and obtaining pre-scan data;subsampling said pre-scan data with a varied dwell time to obtain simulated varied dwell time data;comparing a reconstructed image of said pre-scan data with a reconstructed image of said simulated varied dwell time data;determining whether the result of said comparison falls within a predefined acceptable range;modifying said varied dwell time to obtain a modified set of simulated varied dwell time data if the result of said comparison is outside of said predefined acceptable range; andusing said varied dwell time in a clinical scan of said patient if the result of said comparison is within said predefined acceptable range.

13. The method of claim 12, wherein said comparison involves calculating a difference between the reconstructed image of said pre-scan data and the reconstructed image of said simulated varied dwell time data.

14. The method of claim 12, wherein said pre-scan uses a constant dwell time for each view angle.

15. The method of claim 12, wherein said predefined acceptable range is based on a desired measure of image quality specified by a user.

16. The method of claim 15, wherein said desired measure of image quality is minimized total noise power.

17. The method of claim 15, wherein said desired measure of image quality is maximization of probability of lesion detection.

18. The method of claim 15, wherein said desired measure of image quality is optimization of object contrast in a region of interest.

19. The method of claim 15, wherein said desired measure of image quality is optimization of image quality for multi-modality imaging.