Peak detection calibration for gamma camera using non-uniform pinhole aperture grid mask
Point source responses of pinhole apertures in a non-uniform grid mask used to spatially calibrate a gamma camera can be modeled as a two-dimensional Gaussian function with a set of seven parameters. The Gaussian parameters can be measured using a surface-fitting algorithm that seeks minimum error in the least squares sense. The process is repeated for data from each pinhole location and the data are added together to generate a complete model of the flood image from the mask, which then can be used in a peak detection process for clinical images.
1. Apparatus for spatial calibration of a gamma camera including a scintillation crystal and an array of photodetectors optically coupled to the scintillation crystal, comprising:
a) a mask having a plurality of pinhole apertures arrayed in a non-rectangular distribution;
b) a scanning point source of radiation capable of being scanned over the entirety of said plurality of pinhole apertures; and
c) a processor for collecting individual point source response data from each of said plurality of pinhole apertures, modeling said point source response data as a Gaussian function, adding together Gaussian function data for each of said plurality of pinhole apertures to obtain a simulated uniform flood image, and using said simulated uniform flood image to correct input image data for spatial distortions.
2. The apparatus of claim 1 , wherein said array of photodetectors comprises an array of photomultiplier tubes.
3. The apparatus of claim 2 , wherein said array of pinhole apertures includes apertures at triple points of said photomultiplier tubes.
4. The apparatus of claim 2 , wherein said array of pinhole apertures includes apertures at double points of said photomultiplier tubes.
5. The apparatus of claim 2 , wherein said array of pinhole apertures includes apertures at double points and at triple points of said photomultiplier tubes.
6. The apparatus of claim 1 , wherein said scanning point source of radiation includes a source of Co-57.
7. The apparatus of claim 6 , wherein said Co-57 source is mounted to an X-Y scanner.
8. A method for spatial calibration of a gamma camera including a scintillation crystal and an array of photodetectors optically coupled to the scintillation crystal, comprising the steps of:
a) mounting a mask having a plurality of pinhole apertures arrayed in a non-rectangular distribution adjacent to said scintillation crystal;
b) scanning a scanning point source of radiation over the entirety of said plurality of pinhole apertures;
c) collecting individual point source response data from each of said plurality of pinhole apertures;
d) modeling said point source response data as a Gaussian function; and
e) adding together Gaussian function data for each of said plurality of pinhole apertures to obtain a simulated uniform flood image that is used to correct input image data for spatial distortions.
9. The method of claim 8 , wherein the step of modeling comprises the step of:
modeling point source response data from each pinhole aperture as a 2D Gaussian surface, defined as G(x, y)=c 0 +λe −1/2U , wherein U is an elliptical function defined as
U
=
(
x
σ
x
)
2
+
(
y
σ
y
)
2
,
where x and y are spatial coordinates of a pinhole aperture, 2σ x and 2σ y are lengths of axes of the ellipse U, and c 0 is a photopeak constant.
10. The method of claim 9 , further comprising the step of rotating ellipse U by a predetermined amount τ in the clockwise direction from the X axis, with a rotated coordinate system being defined as:
[
x
y
]
=
[
cos
τ
-
sin
τ
sin
τ
cos
τ
]
[
x
′
-
x
0
y
′
-
y
0
]
11. A method of correcting input image data from a gamma camera for spatial distortions, comprising the steps of:
selecting a Region of Interest (ROI) of said input image data to be corrected, and obtaining image T from input image data within said ROI;
obtaining a Gaussian model image M of said ROI from said simulated uniform flood image obtained by the method of claim 8 ;
subtracting a Gaussian model image G for each pinhole aperture within said ROI from said model image M, to obtain a background image Q;
subtracting background image Q from said image T to obtain Gaussian peak image F; and
performing a 2D Gaussian fit for peak image F to obtain corrected image P.
12. The method of claim 11 , further comprising the step of repeating said 2D Gaussian fit on corrected image P, until a preselected error convergence parameter is reached.
13. The method of claim 12 , wherein said convergence parameter is defined as:
E
i
-
E
i
-
1
E
i
〉
ɛ
,
wherein E i is the error in the least squares sense of the i th iteration of said Gaussian fit process.