Random estimation in positron emission tomography with tangential time-of-flight mask
View Patent ↗A method of estimating random events in positron emission tomography list mode data, including obtaining time-of-flight (TOF) list mode count data that includes TOF information; converting the obtained TOF list mode count data into four-dimensional (4D) raw sinogram count data, without using the TOF information, wherein the 4D raw sinogram count data includes random count values; interpolating the 4D raw sinogram count data to generate 4D interpolated sinogram count data; low-pass filtering the 4D interpolated sinogram count data to remove noise; converting the low-pass filtered 4D interpolated sinogram count data into filtered 4D raw sinogram count data; and generating, by a processor, five-dimensional (5D) TOF raw sinogram count data from the filtered 4D raw sinogram count data by effectively applying a TOF mask filter to the filtered 4D raw sinogram count data.
1. A method of estimating random events in positron emission tomography (PET) list mode data for improved reconstruction of a PET image, the method comprising:
obtaining time-of-flight (TOF) list mode count data that includes TOF information;
converting the obtained TOF list mode count data into four-dimensional (4D) non-TOF raw sinogram count data, without using the TOF information, wherein the 4D non-TOF raw sinogram count data includes random count values;
interpolating the 4D non-TOF raw sinogram count data to generate 4D non-TOF interpolated sinogram count data;
low-pass filtering the 4D non-TOF interpolated sinogram count data to remove noise;
converting the low-pass filtered 4D non-TOF interpolated sinogram count data into filtered 4D non-TOF raw sinogram count data;
generating, by a processing circuit, five-dimensional (5D) TOF raw sinogram count data from the filtered 4D non-TOF raw sinogram count data; and
reconstructing, by the processing circuit, the PET image using the 5D TOF raw sinogram count data.
2. The method of claim 1 , wherein the generating step comprises:
generating the 5D TOF raw sinogram count data {circumflex over (r)}(rad, phi, slice, t) at each TOF bin t according to the following equation:
r
^
(
ra
d
,
phi
,
slice
,
t
)
=
t
binsz
t
coin
cos
θ
r
(
ra
d
,
phi
,
slice
)
wherein t coin is a size of a coincidence window in mm, t binsz a TOF bin size in mm, {circumflex over (r)}(rad, phi, slice) is filtered 4D non-TOF raw sinogram random count data, (rad,phi,slice) are coordinates of a line-of-response (LOR) in the raw sinogram space, and θ is an oblique angle of the LOR.
3. The method of claim 1 , wherein the obtaining step comprises:
obtaining the TOF list mode data, which for each coincident event, has a format {x a , z a , x b , z b , e a , e b , tof}, wherein x a , x b are incident transverse crystal numbers of events a and b, respectively; z a , z b are incident axial crystal numbers of events a and b, respectively; and e a , e b are energy levels of events a and b, respectively, and tof is an arrival time difference of events a and b.
4. The method of claim 1 , further comprising:
estimating the random count values within the 4D non-TOF raw sinogram count data using a delayed coincidence window method.
5. An apparatus for estimating random events in positron emission tomography (PET) list mode data for improved reconstruction of a PET image, the apparatus comprising:
a memory storing time-of-flight (TOF) list mode count data that includes TOF information; and
a processor configured to
convert the stored TOF list mode count data into four-dimensional (4D) non-TOF raw sinogram count data, without using the TOF information, wherein the 4D non-TOF raw sinogram count data includes random count values;
interpolate the 4D non-TOF raw sinogram count data to generate 4D non-TOF interpolated sinogram count data;
low-pass filter the 4D non-TOF interpolated sinogram count data to remove noise;
convert the low-pass filtered 4D non-TOF interpolated sinogram count data into filtered 4D non-TOF raw sinogram count data;
generate five-dimensional (5D) TOF raw sinogram count data from the filtered 4D non-TOF raw sinogram count data; and
reconstruct the PET image using the 5D TOF raw sinogram count data.
6. The apparatus of claim 5 , wherein the processor is further configured to:
generate the 5D TOF raw sinogram count data {circumflex over (r)}(rad, phi, slice, t) at each TOF bin t according to the following equation:
r
^
(
ra
d
,
phi
,
slice
,
t
)
=
t
binsz
t
coin
cos
θ
r
(
ra
d
,
phi
,
slice
)
wherein t coin is a size of a coincidence window in mm, t binsz is a TOF bin size in mm, {circumflex over (r)}(rad, phi, slice) is filtered 4D non-TOF raw sinogram random count data, (rad,phi,slice) are coordinates of a line-of-response (LOR) in the raw sinogram space, and θ is an oblique angle of the LOR.
7. The method of claim 5 , wherein the memory stores the TOF list mode data, which for each coincident event, has a format {x a , z a , x b , z b , e a , e b , tof}, wherein x a , x b are incident transverse crystal numbers of events a and b, respectively; z a , z b are incident axial crystal numbers of events a and b, respectively; and e a , e b are energy levels of events a and b, respectively, and tof is an arrival time difference of events a and b.
8. The method of claim 5 , wherein the processor is further configured to:
estimate the random count values within the 4D non-TOF raw sinogram count data using a delayed coincidence window method.