# US 9,437,018 B2 · Method and apparatus for reducing artifacts in computed tomography image reconstruction

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 9,437,018 B2
- Document ID: US9437018B2
- Status: Granted
- Kind: B2
- Application number: 14/551,601
- Filed: 2014-11-24
- Published: 2015-05-28
- Granted: 2016-09-06
- Earliest claimed: 2013-11-25
- Assignees: GE Medical Systems Global Technology Company, LLC
- Applicants: Ge Medical Systems Global Technology Co. LLC
- Inventors: Shuqin Dong; Jiaqin Dong; Shou Li
- Web page: https://patentgrep.com/patent/US9437018B2

## CPC

- G06T 12/30: Image post-processing, e.g. metal artefact correction
- G06T 12/20: Inverse problem, i.e. transformations from projection space into object space
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

The present invention provides a method and apparatus for reducing artifacts in CT image reconstruction. The method comprises obtaining an original reconstructed image and an original sinogram and determining a proportion of metal pixels in the original reconstructed image. If the proportion is greater than a first threshold value, then generating an expanded metal template based on the original reconstructed image, generating a metal-free, metal artifact reduced (MAR) image based on the expanded metal template and the original sinogram, and generating a final image based on the expanded metal template and the metal-free, MAR image. However, if the proportion of metal pixels is less than a second threshold value, then generating an expanded metal template based on the original reconstructed image, generating a metal-free, metal artifact reduced image based on a treatment, and generating a final image based on the expanded metal template and the metal-free, MAR image.

## Claims

### Claim 1

A method for reducing artifacts in computed tomography (CT) image reconstruction, comprising:
obtaining an original reconstructed image and an original sinogram; and
determining a proportion of metal pixels in the original reconstructed image;
wherein, if the proportion of metal pixels is greater than a first threshold value, then the method further comprises:
generating an expanded metal template based on the original reconstructed image, wherein the expanded metal template includes a two-dimensional or three-dimensional expansion for different pixels,
generating a metal-free, metal artifact reduced (MAR) image based on the expanded metal template and the original sinogram, and
generating a final image based on the expanded metal template and the metal-free, MAR image; and
wherein, if the proportion of metal pixels is less than a second threshold value, then the method further comprises:
generating an expanded metal template based on the original reconstructed image,
generating a metal-free, MAR image based on a treatment including weighting of the expanded metal template and the original sinogram, and
generating a final image based on the expanded metal template and the metal-free, MAR image.

### Claim 2 (depends on 1)

The method for reducing artifacts in CT image reconstruction according to claim 1, wherein generating an expanded metal template based on the original reconstructed image, when the proportion of metal pixels is greater than a first threshold value, further comprises:
generating an original metal template;
performing a first expansion of the original metal template to generate a two-dimensional expanded metal template; and
performing a second expansion of the original metal template based on said two-dimensional expanded metal template to generate said expanded metal template, including a two-dimensional or three-dimensional expansion for different pixels.

### Claim 3 (depends on 2)

The method for reducing artifacts in CT image reconstruction according to claim 2, wherein performing said second expansion comprises:
in the two-dimensional expanded metal template, for each metal pixel, given a 20×20 pixel rectangle centering on said metal pixel, if no metal exists on the boundary of the 20×20 pixel rectangle and the number of metal pixels within the 20×20 pixel rectangle is not greater than 30, then performing a 2D expansion on said metal pixel of the original metal template; otherwise, performing a 3D expansion on said metal pixel of the original metal template.

### Claim 4 (depends on 1)

The method for reducing artifacts in CT image reconstruction according to claim 1, wherein the step of generating a metal-free, MAR image, when the proportion of metal pixels is greater than a first threshold value comprises:
performing an orthographic projection on the expanded metal template to generate a metal sinogram;
interpolating the metal sinogram and the original sinogram to generate an interpolated sinogram; and
performing a back projection on the interpolated sinogram to generate the metal-free, MAR image.

### Claim 5 (depends on 1)

The method for reducing artifacts in CT image reconstruction according to claim 1, wherein the step of generating a metal-free, MAR image, when the proportion of metal pixels is less than a second threshold value comprises:
performing an orthographic projection on the expanded metal template to generate a metal sinogram;
interpolating the metal sinogram and the original sinogram to generate an interpolated sinogram;
weighting the metal sinogram, the interpolated sinogram and the original sinogram to generate a mixed sinogram; and
performing a back projection on the mixed sinogram to generate the metal-free, MAR image.

### Claim 6 (depends on 5)

The method for reducing artifacts in CT image reconstruction according to claim 5, wherein the step of weighting comprises:
subtracting the metal sinogram fmetal from the original sinogram forig to obtain a difference forig-metal, and comparing the difference with the interpolated sinogram f interpolation for calculation of a coefficient coef as follows:
{
coef
=
1
/
(
f
orig
⁢
-
⁢
metal
-
f
interpolation
)
if
⁢
⁢
f
orig
⁢
-
⁢
metal
<
f
interpolation
coef
=
0
if
⁢
⁢
f
orig
⁢
-
⁢
metal
>
f
interpolation
;
then let T be:
{
T
=
coef
if
⁢
⁢
f
orig
⁢
-
⁢
metal
<
f
interpolation
T
=
1
-
coef
if
⁢
⁢
f
orig
⁢
-
⁢
metal
>
f
interpolation
;
calculating two parameters p1 and p2 using a maximum value f max and a minimum value f min of the metal sinogram,
p1=α*fmin+(1−α)*fmax
p2=(1−β)*fmin+β*fmax
wherein α and β are obtained by experiment; let α be 0.35 and β be 0.95 to calculate t as follows:
t
=
f
orig
-
p
⁢
⁢
1
p
⁢
⁢
2
-
p
⁢
⁢
1
;
then calculating a weight w as follows:
{
w
=
w
max
*
T
if
⁢
⁢
f
⁡
(
v
,
c
,
r
)
<
p
⁢
⁢
1
w
=
w
max
⁡
(
1
-
(
6
*
t
5
-
15
*
t
4
+
10
*
t
3
)
)
*
T
if
⁢
⁢
p
⁢
⁢
2
>
f
⁡
(
v
,
c
,
r
)
≥
p
⁢
⁢
1
w
=
0
if
⁢
⁢
f
⁡
(
v
,
c
,
r
)
≥
p
⁢
⁢
2
wherein Wmax is a fixed value 0.5, and f (v,c,r) represents the projection value for view v, channel c, and row r;
performing a weighting using the above weight to obtain a mixed sinogram fblend:
fblend=w*forig-Metal+(1−w)*finterpolation.

### Claim 7 (depends on 1)

The method for reducing artifacts in CT image reconstruction according to claim 1, wherein the step of generating a final image comprises:
adding the expanded metal template and the metal-free, MAR image to generate the final image.

### Claim 8 (depends on 1)

The method for reducing artifacts in CT image reconstruction according to claim 1, wherein the first threshold value is greater than the second threshold value.

### Claim 9

A computed tomography apparatus, comprising:
a scanner for scanning an object using X-rays to obtain original data for generating an original reconstructed image; and
a processor operably coupled to said scanner and programmable to obtain an original reconstructed image and an original sinogram, and programmable to determine a proportion of metal pixels in the original reconstructed image,
wherein, if the proportion of metal pixels is greater than a first threshold value, the processor is configured to:
generate an expanded metal template based on the original reconstructed image, wherein the expanded metal template includes a two-dimensional or three-dimensional expansion for different pixels,
generate a metal-free, MAR image based on the expanded metal template and the original sinogram, and
generate a final image based on the expanded metal template and the metal-free, MAR image; and
wherein, if the proportion of metal pixels is less than a second threshold value, the processor is configured to:
generate an expanded metal template based on the original reconstructed image,
generate a metal-free, MAR image based on a treatment including weighting of the expanded metal template and the original sinogram, and
generate a final image based on the expanded metal template and the metal-free, MAR image.

### Claim 10 (depends on 9)

The computed tomography apparatus according to claim 9, wherein if the proportion of metal pixels is greater than the first threshold value, the processor, in generating the expanded metal template based on the original reconstructed image is further configured to:
generate an original metal template;
perform a first expansion of the original metal template to generate a two-dimensional expanded metal template; and
perform a second expansion of the original metal template based on said two-dimensional expanded metal template to generate said expanded metal template, including a two-dimensional or three-dimensional expansion for different pixels.

### Claim 11 (depends on 10)

The computed tomography apparatus according to claim 10, wherein in the two-dimensional expanded metal template, for each metal pixel, given a 20×20 pixel rectangle centering on said metal pixel, if no metal exists on the boundary of the 20×20 pixel rectangle and the number of metal pixels within the 20×20 pixel rectangle is not greater than 30, the processor is configured to perform a 2D expansion on said metal pixel of the original metal template; otherwise, the processor is configured to perform a 3D expansion on said metal pixel of the original metal template.

### Claim 12 (depends on 9)

The computed tomography apparatus according to claim 9, wherein in generating a metal-free, MAR image when the proportion of metal pixels is greater than a first threshold value, the processor is further configured to:
perform an orthographic projection on the expanded metal template to generate a metal sinogram;
interpolate the metal sinogram and the original sinogram to generate an interpolated sinogram; and
perform a back projection on the interpolated sinogram to generate the metal-free, MAR image.

### Claim 13 (depends on 9)

The computed tomography apparatus according to claim 9, wherein in generating a metal-free, MAR image when the proportion of metal pixels is less than a second threshold value, the processor is further configured to:
perform an orthographic projection on the expanded metal template to generate a metal sinogram;
interpolate the metal sinogram and the original sinogram to generate an interpolated sinogram;
weight the metal sinogram, the interpolated sinogram and the original sinogram to generate a mixed sinogram; and
perform a back projection on the mixed sinogram to generate the metal-free, MAR image.

### Claim 14 (depends on 13)

The computed tomography apparatus according to claim 13, wherein in weighting the metal sonogram, the proceccor is further configured to:
subtract the metal sinogram fmetal from the original sinogram forig to obtain a difference forig-metal, and comparing the difference with the interpolated sinogram finterpolation for calculation of a coefficient coef as follows:
{
coef
=
1
/
(
f
orig
⁢
-
⁢
metal
-
f
interpolation
)
if
⁢
⁢
f
orig
⁢
-
⁢
metal
<
f
interpolation
coef
=
0
if
⁢
⁢
f
orig
⁢
-
⁢
metal
>
f
interpolation
;
then let T be:
{
T
=
coef
if
⁢
⁢
f
orig
⁢
-
⁢
metal
<
f
interpolation
T
=
1
-
coef
if
⁢
⁢
f
orig
⁢
-
⁢
metal
>
f
interpolation
;
calculate two parameters p1 and p2 using a maximum value f max and a minimum value f min of the metal sinogram,
p1=α*fmin+(1−α)*fmax
p2=(1−β)*fmin+β*fmax
wherein α and β are obtained by experiment; let α be 0.35 and β be 0.95 to calculate t as follows:
t
=
f
orig
-
p
⁢
⁢
1
p
⁢
⁢
2
-
p
⁢
⁢
1
;
then calculating a weight w as follows:
{
w
=
w
max
*
T
if
⁢
⁢
f
⁡
(
v
,
c
,
r
)
<
p
⁢
⁢
1
w
=
w
max
⁡
(
1
-
(
6
*
t
5
-
15
*
t
4
+
10
*
t
3
)
)
*
T
if
⁢
⁢
p
⁢
⁢
2
>
f
⁡
(
v
,
c
,
r
)
≥
p
⁢
⁢
1
w
=
0
if
⁢
⁢
f
⁡
(
v
,
c
,
r
)
≥
p
⁢
⁢
2
wherein Wmax is a fixed value 0.5, and f (v,c,r) represents the projection value for view v, channel c, and row r; and
performing a weighting using the above weight to obtain a mixed sinogram fblend:
fblend=w*forig-Metal+(1−w)*finterpolation.

### Claim 15 (depends on 9)

The computed tomography apparatus according to claim 9, wherein in generating a final image, the processor is further configured to add the expanded metal template and the metal-free, MAR image to generate the final image.

### Claim 16 (depends on 9)

The computed tomography apparatus according to claim 9, wherein the first threshold value is greater than the second threshold value.