Back to results
US 9,317,942 B2Granted
Fast computation of the sensitivity matrix in iterative algorithms
AssigneeKoninklijke Philips N.V.
InventorPatrick Olivier
Life of this patent
Estimating the term…
Abstract
An image processing apparatus includes a geometric correction memory which stores a pre-calculated geometric matrix specific to an imaging apparatus. An attenuation map memory which stores an attenuation map of a subject to be imaged in the imaging apparatus. A buffer stores a plurality of lines generated by the imaging apparatus to be reconstructed. A processor reconstructs the lines into a attenuation corrected image representation of the subject using the lines from the buffer, the attenuation map, and the geometric matrix.
Claims16
A image processing apparatus comprising:
- a geometric correction memory which stores a pre-calculated geometric matrix specific to an imaging apparatus;
- an attenuation map memory which stores an attenuation map of a subject to be imaged in the imaging apparatus;
- a buffer which stores a plurality of lines generated by the imaging apparatus to be reconstructed;
- a processor which reconstructs the lines into an attenuation corrected image representation of the subject using the lines from the buffer, the attenuation map, and the geometric matrix;
- and
- wherein the geometric correction memory includes a look-up table storing pre-calculated geometric matrix values which are calculated for the imaging apparatus during manufacture and are not calculated for each patient.
3 more elements
An image processing apparatus comprising:
- a geometric correction memory configured to store a pre-calculated geometric matrix specific to an imaging apparatus;
- an attenuation map memory configured to store an attenuation map of a subject to be imaged in the imaging apparatus;
- a buffer configured to store a plurality of lines generated by the imaging apparatus to be reconstructed in a list mode format;
- and
- a processor configured to reconstruct the lines into an attenuation corrected image representation of the subject using the lines from the buffer, the attenuation map, and the geometric matrix, wherein the processor is configured to reconstruct the image using a reconstruction algorithm represented by
- λ
- i
- m
- +
- 1
- =
- λ
- i
- m
-
-
- ∑
- j
- =
- 1
- J
-
- (
- h
- ji
- /
- ∑
- i
- ′
- =
- 1
- I
-
- h
- ji
- ′
-
- λ
- i
- ′
- m
- )
- ∑
- i
- ′
- =
- 1
- I
-
- b
- i
- ′
-
- g
- ii
- ′
- wherein λim is a image value at voxel i on a mth iteration, λim+1 is the image value at voxel i on a mth+1 iteration,
- ∑
- j
- =
- 1
- J
-
- (
- h
- ji
- /
- ∑
- i
- ′
- =
- 1
- I
-
- h
- ji
- ′
-
- λ
- i
- ′
- m
- )
- is a iterative reconstruction algorithm, and
- ∑
- i
- ′
- =
- 1
- I
-
- b
- i
- ′
-
- g
- ii
- ′
- is a sensitivity matrix, wherein gii′ is the geometric matrix, bi′ is the attenuation map, j is one the lines, and hji is a geometric factor.
96 more elements
The method of image processing comprising:
- in one or more memories, storing a pre-calculated geometric matrix specific to an imaging apparatus;
- in the one or more memories, storing an attenuation map of a subject imaged in the imaging apparatus;
- in the one or more memories, storing a plurality of lines generated by the imaging apparatus to be reconstructed;
- and
- with a processor, reconstructing the lines into an attenuation corrected image representation of the subject using the lines, the attenuation map, and the geometric matrix,
- wherein the stored pre-calculated geometric matrix values are calculated for the imaging apparatus during manufacture and are not calculated for each patient.
3 more elements
A method of image processing comprising:
- with a processor, receiving a plurality of lines generated by a imaging apparatus to be reconstructed;
- with the processor, reconstructing the lines into an attenuation corrected image representation using the lines, an attenuation map, and a geometric matrix, wherein the attenuation map is decoupled from the geometric matrix;
- wherein the image representation is constructed using a reconstruction algorithm represented by
- λ
- i
- m
- +
- 1
- =
- λ
- i
- m
-
- ∑
- j
- =
- 1
- J
-
- (
- h
- ji
- /
- ∑
- i
- ′
- =
- 1
- I
-
- h
- ji
- ′
-
- λ
- i
- ′
- m
- )
- ∑
- i
- ′
- =
- 1
- I
-
- b
- i
- ′
-
- g
- ii
- ′
- wherein λim is a image value at voxel i on a mth iteration, λim+1 is the image value at voxel i on a mth+1 iteration,
- ∑
- j
- =
- 1
- J
-
- (
- h
- ji
- /
- ∑
- i
- ′
- =
- 1
- I
-
- h
- ji
- ′
-
- λ
- i
- ′
- m
- )
- is a iterative reconstruction algorithm, and
- ∑
- i
- ′
- =
- 1
- I
-
- b
- i
- ′
-
- g
- ii
- ′
- is a sensitivity matrix, wherein gii′ is the geometric matrix, bi′ is the attenuation map, j is one the lines, and hji is a geometric factor.
93 more elements
Dependencies are inferred from the claim text.