WO2011027402A1 - 核医学用データ処理方法および核医学診断装置 - Google Patents
核医学用データ処理方法および核医学診断装置 Download PDFInfo
- Publication number
- WO2011027402A1 WO2011027402A1 PCT/JP2009/004381 JP2009004381W WO2011027402A1 WO 2011027402 A1 WO2011027402 A1 WO 2011027402A1 JP 2009004381 W JP2009004381 W JP 2009004381W WO 2011027402 A1 WO2011027402 A1 WO 2011027402A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- nuclear medicine
- parallel
- cross
- lor
- section
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01T—MEASUREMENT OF NUCLEAR OR X-RADIATION
- G01T1/00—Measuring X-radiation, gamma radiation, corpuscular radiation, or cosmic radiation
- G01T1/29—Measurement performed on radiation beams, e.g. position or section of the beam; Measurement of spatial distribution of radiation
- G01T1/2914—Measurement of spatial distribution of radiation
- G01T1/2985—In depth localisation, e.g. using positron emitters; Tomographic imaging (longitudinal and transverse section imaging; apparatus for radiation diagnosis sequentially in different planes, steroscopic radiation diagnosis)
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B6/00—Apparatus for radiation diagnosis, e.g. combined with radiation therapy equipment
- A61B6/02—Devices for diagnosis sequentially in different planes; Stereoscopic radiation diagnosis
- A61B6/03—Computerised tomographs
- A61B6/037—Emission tomography
Definitions
- the present invention relates to a nuclear medicine data processing method and a nuclear medicine diagnostic apparatus for performing forward projection and backprojection operations on list data generated from event data obtained by detecting radiation.
- the above-described nuclear medicine diagnosis apparatus that is, an ECT (Emission Computed Tomography) apparatus will be described by taking a PET (Positron Emission Tomography) apparatus as an example.
- the PET device detects a plurality of photons generated by the annihilation of positrons (Positrons), and only detects ⁇ rays simultaneously with a plurality of detectors (that is, only when they are counted simultaneously). It is configured to reconstruct an image.
- the process of drug accumulation in the target tissue is measured over time, whereby quantitative measurement of various biological functions is possible. Therefore, the image obtained by the PET apparatus has function information.
- a positron (positron) radioactive isotope for example, 15 O, 18 F, 11 C, etc.
- a detector array comprising a number of ⁇ -ray detectors arranged in a ring shape so as to surround the body axis that is the longitudinal axis of the subject.
- calculation is performed by a computer in the same manner as in ordinary X-ray CT (Computed Tomography), and the image is specified in the plane, and an image of the subject is created.
- Non-Patent Document 1 is an indispensable technique when an image is reconstructed in a nuclear medicine diagnostic apparatus such as a PET apparatus.
- an event for detecting ⁇ rays is referred to as an “event”
- data that is simultaneously counted is referred to as “coincidence data”
- single event data data that is not simultaneously counted
- Data for each event transferred from the gantry of the PET apparatus is called “list data”.
- This list data includes position information such as X coordinate, Y coordinate, and Z coordinate for each event, time information, and the like. These pieces of information are arranged in time series.
- the list data is converted into a histogram to obtain histogram data.
- This histogram data is data integrated at a predetermined time.
- this histogram data there is a sinogram with the vertical axis representing the projection direction and the horizontal axis representing the pixel.
- Non-Patent Document 2 a mode using list data
- list mode the data increases by the number of events (events), and the large amount of calculation becomes a problem.
- the amount of calculation is further enormous.
- MIMD Multiple Instruction-Multiple Data
- MPI Message Passing Interface
- SIMD type operations can simultaneously process a plurality of data with a single instruction, they are suitable for performing the same operation processing on a large number of operations.
- parallel operations such as forward projection processing and back projection processing using a GPU
- writing to the same memory may occur (so-called “memory contention”).
- memory contention Ingenuity is necessary.
- Non-Patent Document 3 uses a sinogram. In the sinogram, data of the same pixel is duplicated, resulting in memory contention. Then, the parallel study of the successive approximation method using list data is also performed as a parallel operation using the same GPU (for example, refer nonpatent literature 4).
- Non-Patent Document 4 uses a mechanism peculiar to GPU, and has a problem that it is difficult to apply to other SIMD type mechanisms other than GPU. Since the GPU is originally for image processing, the use of a mechanism unique to the GPU means that data is treated as a texture image and a pixel (pixel) shader or a vertex shader is used.
- the present invention has been made in view of such circumstances, and an object of the present invention is to provide a nuclear medicine data processing method and a nuclear medicine diagnostic apparatus that are highly versatile and capable of speeding up operations. To do.
- the nuclear medicine data processing method performs forward projection to obtain post-projection-processed data by performing forward projection operation in parallel for each LOR on list data generated from event data obtained by detecting radiation.
- the forward projection processing step performs forward projection operation in parallel for each LOR (Line Of ⁇ ⁇ ⁇ Response) on list data generated from event data obtained by detecting radiation.
- LOR Line Of ⁇ ⁇ ⁇ Response
- backprojection calculation is performed in parallel for each area obtained by dividing the cross section of the image space. Since list processing is performed in parallel for each LOR for each LOR, for example, forward projection and backprojection calculations can be realized without using a GPU-specific mechanism, and can be applied to other calculation mechanisms. Is. Further, parallel processing is performed on the list data, and the parallel processing is performed by forward projection calculation and back projection calculation, respectively, so that memory contention can be prevented and high speed can be realized.
- the back projection processing step is performed for each region obtained by dividing the cross section of the image space. Back projection operations are performed in parallel. As a result, the versatility is high and the calculation speed can be increased.
- the main direction determination step determines the main direction of the LOR based on the LOR vector and the arrangement direction of the detector for detecting radiation.
- the cross section determining step determines a cross section of the image space through which the LOR passes based on the main direction determined in the main direction determining step.
- the calculation area determination step determines the calculation area where the LOR in the cross section traverses. More specifically, the forward projection processing step and the back projection processing step perform the following arithmetic processing.
- the forward projection processing step calculates in parallel for each calculation region
- the back projection processing step calculates the post-projection processing data or list data in parallel for each region obtained by dividing the calculation region into a plurality of regions. . In this way, determining the main direction makes it easy to classify the main direction and perform each calculation process.
- the back projection process step divides the cross section into a plurality of regions, and the divided and dispersed regions belong to one region. It is preferable to perform operations in parallel for each area to which the user belongs. By classifying the regions in this way, it is possible to prevent localization in a specific thread and improve the calculation efficiency.
- “Thread” indicates a unit of parallel data processing.
- a more preferable example is to associate a plurality of row regions separated by a predetermined interval as one region described above. In this way, by classifying the regions evenly distributed, it is possible to further prevent localization in a specific thread.
- An example of the nuclear medicine data processing method of these inventions described above is to use a sequential approximation method in which the forward projection process step and the backprojection process step are executed a plurality of times to sequentially approximate and update the image.
- image reconstruction is performed using the successive approximation method as described above. Therefore, the present invention may also be applied to the successive approximation method.
- filtered back projection FBP: “Filtered” Back Projection
- FBP Filtered” Back Projection
- the nuclear medicine diagnosis apparatus provides forward projection processing means for obtaining forward-projected data by performing forward projection operation on list data generated from event data obtained by detecting radiation in parallel for each LOR. And backprojection processing means for performing a backprojection operation in parallel for each area obtained by dividing the cross section of the image space.
- the forward projection processing means performs forward projection operation on the list data generated from the event data obtained by detecting radiation in parallel for each LOR, and performs forward projection processed data. Since the backprojection processing means performs the backprojection operation on the data or list data after the forward projection processing in parallel for each region obtained by dividing the cross section of the image space, the versatility is similar to the nuclear medicine data processing method. It is high and the calculation speed can be increased.
- the forward projection processing is performed by performing the forward projection operation on the list data generated from the event data obtained by detecting the radiation in parallel for each LOR. Since post-projection data is obtained and backprojection calculation is performed in parallel for each area obtained by dividing the cross section of the image space, the post-projection processing data or list data is highly versatile and the calculation speed can be increased.
- PET Positron Emission Tomography
- FIG. 6 is a schematic diagram showing a flow of list data in a series of steps S1 to S5.
- FIG. 1 is a side view and a block diagram of a PET (Positron Emission Tomography) apparatus according to an embodiment
- FIG. 2 is a schematic perspective view of a ⁇ -ray detector.
- the PET apparatus includes a top plate 1 on which a subject M is placed as shown in FIG.
- the top plate 1 is configured to move up and down and translate along the body axis Z of the subject M.
- the subject M placed on the top 1 is scanned from the head to the abdomen and foot sequentially through the opening 2a of the gantry 2, which will be described later. Get the image. Note that there is no particular limitation on the scanned part and the scanning order of each part.
- the PET apparatus includes a gantry 2 having an opening 2a and a ⁇ -ray detector 3.
- the ⁇ -ray detector 3 is arranged in a ring shape so as to surround the body axis Z of the subject M, and is embedded in the gantry 2.
- the ⁇ -ray detector 3 corresponds to the detector in the present invention.
- the PET apparatus includes a top plate driving unit 4, a controller 5, an input unit 6, an output unit 7, a memory unit 8, a coincidence circuit 9, and an arithmetic processing unit 10.
- the top plate driving unit 6 is a mechanism for driving the top plate 1 so as to perform the above-described movement, and is configured by a motor or the like not shown.
- the arithmetic processing unit 10 corresponds to the forward projection processing means and back projection processing means in the present invention.
- the controller 5 comprehensively controls each part constituting the PET apparatus according to the present embodiment.
- the controller 5 and the arithmetic processing unit 10 are configured by a central processing unit (CPU) or the like.
- the arithmetic processing unit 10 is configured by a SIMD type mechanism represented by a GPU or the like.
- the arithmetic processing unit 10 may be a SIMD type mechanism other than the GPU, and is not particularly limited as long as parallel arithmetic is possible.
- the input unit 6 sends data and commands input by the operator to the controller 5.
- the input unit 6 includes a pointing device represented by a mouse, a keyboard, a joystick, a trackball, a touch panel, and the like.
- the output unit 7 includes a display unit represented by a monitor, a printer, and the like.
- the memory unit 8 includes a storage medium represented by ROM (Read-only Memory), RAM (Random-Access Memory), and the like.
- the count value (count) simultaneously counted by the coincidence circuit 9 the data relating to the coincidence counting such as the detector pair consisting of the two ⁇ -ray detectors 3 and the LOR, and the arithmetic processing unit 10 perform the calculation.
- Various processed data are written and stored in the RAM, and are read from the RAM as necessary.
- the ROM stores in advance a program for performing imaging including various types of nuclear medicine diagnosis, and the controller 5 and the arithmetic processing unit 10 execute the program to perform nuclear medicine diagnosis according to the program. Do each.
- a program related to CUDA is stored in advance in a ROM so that a parallel computing architecture called “CUDA (provided by NVida)” can execute a parallel operation using a GPU.
- the arithmetic processing unit 10 executes steps S1 to S5 described later.
- the ⁇ -rays generated from the subject M to which the radiopharmaceutical is administered are converted into light by the scintillator block 31 (see FIG. 2) of the ⁇ -ray detector 3, and the converted light is photoelectron of the ⁇ -ray detector 3.
- a multiplier tube (PMT: Photo Multiplier Tube) 33 (see FIG. 2) multiplies and converts it into an electrical signal. The electric signal is sent to the coincidence circuit 9 as an event.
- the coincidence circuit 9 checks the position of the scintillator block 31 (see FIG. 2) and the incident timing of the ⁇ rays, and only when the ⁇ rays are simultaneously incident on the two scintillator blocks 31 on both sides of the subject M. The sent event is determined to be appropriate data.
- the coincidence counting circuit 9 rejects. That is, the coincidence counting circuit 9 detects that ⁇ rays are simultaneously observed in the two ⁇ ray detectors 3 based on the above-described electrical signal.
- the event sent to the coincidence circuit 9 is sent to the arithmetic processing unit 10.
- the arithmetic processing unit 10 performs image reconstruction by forward projection processing or back projection processing to obtain an image of the subject M.
- the image is sent to the output unit 7 via the controller 5. In this manner, nuclear medicine diagnosis is performed based on the image obtained by the arithmetic processing unit 10. Specific functions of the arithmetic processing unit 10 will be described later.
- the ⁇ -ray detector 3 includes a scintillator block 31, a light guide 32 optically coupled to the scintillator block 31, and photoelectrons optically coupled to the light guide 32.
- a multiplier (hereinafter simply abbreviated as “PMT”) 33 is provided.
- Each scintillator element constituting the scintillator block 31 converts ⁇ rays into light by emitting light with the incidence of ⁇ rays. By this conversion, the scintillator element detects ⁇ rays.
- Light emitted from the scintillator element is sufficiently diffused by the scintillator block 31 and input to the PMT 33 via the light guide 32.
- the PMT 33 multiplies the light converted by the scintillator block 31 and converts it into an electric signal. The electric signal is sent to the coincidence circuit 9 (see FIG. 1) as an event as described above.
- FIG. 3 is a flowchart showing a flow of a series of list data processing methods
- FIG. 4 is a schematic diagram showing coincidence with a ⁇ -ray detector for explaining detection probabilities
- FIG. 6 is a schematic diagram that defines the relationship between the arrangement of ⁇ -ray detectors arranged in a three-dimensional manner and three-dimensional coordinates.
- FIG. 6 is a schematic diagram in which the LOR direction vector is shown together with the conceptual diagram of projection
- FIG. FIG. 8 is a schematic diagram of a calculation region in a cross section of the image space through which the LOR passes, FIG.
- FIG. 8 is a schematic diagram for explaining the parallel calculation of the LOR and projection values for one subset
- FIG. 9 is a diagram through which the LOR passes.
- FIG. 10 is a schematic diagram showing division of the calculation area in the cross section of the image space
- FIG. 10 is a schematic diagram showing the flow of list data in a series of steps S1 to S5. 4 and 5, only the scintillator block 31 is illustrated as the ⁇ -ray detector 3, and the light guide 32 and the PMT 33 are not illustrated.
- a description will be given by taking as an example a pixel composed of three-dimensional voxels.
- the list-mode 3D-DRAMA method (Dynamic-Row-Action-Maximum-Likelihood-Algorithm) using list data is adopted as the successive approximation algorithm, and corrections such as absorption correction, sensitivity correction, and random correction are incorporated.
- parallel operation using a GPU is implemented by the above-mentioned CUDA.
- the update formula of the list-mode 3D-DRAMA method is expressed by the following formula (1).
- i (t) in the above equation (1) indicates the t-th coincidence event.
- l (lowercase L) is the number of subsets
- a i (t) j is a probability that ⁇ -ray photons emitted from voxel j are detected by LOR i (t) , and is also referred to as a “system matrix”.
- the detector response function is also taken into consideration.
- C j represents a normalization matrix
- p lj represents a blocking factor.
- Non-Patent Document 2 Two types of blocking factors have been introduced in Non-Patent Document 2 described above, but in this study, convergence is high and a method of calculating using all list data is employed.
- ⁇ (k, l) is a relaxation coefficient and is related to the convergence speed.
- ⁇ and ⁇ 0 are parameters for controlling ⁇ (k, l) .
- a ′ i (t) j is a i (t) j after absorption correction, and the correction-related definition is shown in the following equation (2).
- a i (t) j represents an absorption correction coefficient
- ⁇ j represents a distribution of absorption coefficients.
- r i is a value obtained by predicting the random count rate from the count rates s i0 and s i1 of the single count and the time window 2 ⁇ of the coincidence count.
- FP in the above formula (1) represents Forward Projection, that is, forward projection, and calculation in the forward projection processing is performed in the portion in the above formula (1).
- BP in the above equation (1) represents Back Projection, that is, back projection, and the calculation in the back projection process is performed in the portion in the above equation (1).
- the initial image x j (0, 0) is set appropriately.
- the initial image x j (0,0) may be an image having a uniform pixel value, for example, and x j (0,0) > 0.
- x j ( 0,0) and a ′ i (t) j corrected by the above equation (2) are used, and repeatedly substituting it into the above equation (1), x j ( 0,0), ..., x j ( 0, L-1) is determined sequentially, x finally the obtained x j of (0, L-1) by the x j (1, 0) Raise to j (1,0) .
- x j is sequentially incremented (x j (0,0) , x j (1,0) ..., X j (k, 0) ).
- the number of times k representing repetition is not particularly limited, and may be set as appropriate.
- Step S1 Determination of Main Direction
- the horizontal axis is x
- the vertical axis is y
- the body axis of the subject M is z.
- the body axis z is an axis in the depth direction.
- the direction vector parallel to the LOR is defined as d (d x , d y , d z ) (where
- 1, unit vector), and the LOR vector d and ⁇ -ray detection Based on the arrangement direction of the vessel 3, the main direction of the LOR is determined.
- FIG. 6 is a schematic diagram when the x direction is the main direction. This step S1 corresponds to the main direction determining step in the present invention.
- Step S2 Cross Section Determination
- the cross section of the image space through which the LOR passes is determined based on the main direction determined in step S1.
- the plane in which the main directions are orthogonal is determined as the cross section of the image space through which the LOR passes.
- the cross section is the yz plane as shown in FIGS.
- This step S2 corresponds to the cross section determining step in the present invention.
- Step S3 Calculation Area Determination Based on the cross section determined in step S2 (the yz plane in FIG. 6), the calculation area crossed by the LOR in the cross section is determined.
- the cross section that is the start point of the LOR (denoted by “slice_start” in FIG. 6) and the cross section that is the end point of the LOR (denoted by “slice_end” in FIG. 6) are determined.
- the coordinates on the cross section In FIGS. 6 and 7, since the cross section is the yz plane, the coordinates on the cross section that is the start point are y and z coordinates, and the coordinates on the cross section that is also the end point are y and z coordinates.
- one of the y-coordinates of the start and end points is the minimum coordinate (indicated as “Min_0” in FIG. 7), and the other is the maximum coordinate (in FIG. 7, “Max_0”). ).
- one of the z coordinates of the start point and the end point is the minimum coordinate (indicated as “Min_1” in FIG. 7), and the other is the maximum coordinate (indicated as “Max_1” in FIG. 7). That is, on the cross section, the LOR crosses the maximum coordinate position from the minimum coordinate position on each cross section. Therefore, the calculation area traversed by the LOR is an area delimited by Min_0, Max_0, Min_1, and Max_1 as shown in FIG.
- This step S3 corresponds to the calculation area determination step in the present invention.
- Step S4 Forward Projection Processing
- the list data is subjected to calculation processing in parallel in the calculation area determined in step S3.
- the system matrix is obtained for each voxel j, and the projection value is obtained by multiplying and adding the pixel value in the portion indicated by FP in the above expression (1) within the range of the determined calculation area.
- the parallel operation in the above equation (1) is performed for each LOR.
- the memory into which the projection value is substituted is configured to write one element per event per subset, for example, as shown in FIG. In FIG. 8, “Thread_0”, “Thread_1”, “Thread_2”, “Thread_3”,... Indicate processing units of parallel data to be written.
- Each LOR has a one-to-one correspondence with each thread per subset.
- This step S4 corresponds to the forward projection processing step in this invention.
- Step S5 Backprojection processing (region division) Back projection processing is performed on the post-forward projection processing data or list data on which forward projection processing has been performed in step S4.
- each cross-section of the image space through which the LOR passes is divided into a plurality of regions, and the divided and dispersed regions belong to one region, and are parallel to each of the belonging regions.
- the backprojection value is updated by adding in the part indicated by BP in the above-described equation (1) in each of the areas to which it belongs.
- row regions that are separated by a predetermined interval belong as one region.
- the calculation area is a cross section (slice) of 256 ⁇ 256 pixels
- FIG. 9B is an enlarged view of FIG.
- it is defined as one area of every 7th row, it is divided into 8 regions per slice. That is, it is divided by “Thread_0”, “Thread_1”,... “Thread_7” shown in FIG.
- This step S5 corresponds to the back projection process step in this invention.
- list data for each event (ie, for each LOR) (indicated by “Event_0”, “Event_1”, “Event_2”,... In FIG. 10) is classified in the main direction by determining the main direction in step S1.
- the main direction (denoted by in FIG. 10, "the main direction x list data") list data in the x-direction
- the main direction is classified into list data in the y direction (represented as “list data in the main direction y” in FIG. 10), and in each step S5, the back projection process is performed in parallel.
- the successive approximation method described above is not limited to the above-described DRAMA method, and may be a static (that is, static) RAMLA method (Row-Action Maximum Likelihood Algorithm) or an ML-EM method (Maximum Likelihood Expectation Maximization). ) Or OSEM method (Ordered Subset ML-EM).
- step S4 forward projection process
- step S5 back projection processing
- the data after the forward projection processing or the list data is back-projected in parallel for each region obtained by dividing the cross section of the image space. Since list processing is performed in parallel for each LOR for each LOR, for example, forward projection and backprojection calculations can be realized without using a GPU-specific mechanism, and can be applied to other calculation mechanisms. Is.
- step S5 back projection processing divides the cross section of the image space. Back projection operations are performed in parallel for each region. As a result, the versatility is high and the calculation speed can be increased.
- step S1 main direction determination
- step S2 determines the cross section of the image space through which the LOR passes based on the main direction determined in step S1 (determining the main direction).
- step S3 determines the calculation area where the LOR in the cross section traverses.
- step S4 forward projection process
- step S5 back projection process
- step S4 forward projection processing
- step S5 back projection processing
- step S4 forward projection processing
- step S5 back projection processing
- the above-described step S5 (back projection process) is performed by dividing the above-described cross section into a plurality of regions, and the divided and dispersed regions belong to one region.
- the calculation is preferably performed in parallel for each area. By classifying the regions in this way, it is possible to prevent localization in a specific thread and improve the calculation efficiency.
- each row region separated by a predetermined interval belongs as one region (see FIG. 9). In this way, by classifying the regions evenly distributed, it is possible to further prevent localization in a specific thread.
- step S4 forward projection process
- step S5 back projection process
- step S4 forward projection process
- step S5 back projection process
- the PET apparatus obtains post-projection-processed data by performing the forward projection operation in parallel for each LOR on the list data generated from the event data obtained by detecting ⁇ , and the forward projection.
- an arithmetic processing unit 10 that performs the arithmetic processing of the above-described steps S1 to S5 that performs backprojection calculation in parallel for each region obtained by dividing the cross section of the image space.
- the present invention is not limited to the above embodiment, and can be modified as follows.
- step S4 forward projection processing
- step S5 back projection processing
- the Feldkamp method using filtered back projection may be applied.
- the main direction is classified by determining the main direction, but it is not always necessary to determine the main direction. However, in order to facilitate each calculation process, it is more preferable to classify the main direction by determining the main direction as in the embodiment.
- the main direction is determined after setting the horizontal axis to x, the vertical axis to y, and the body axis of the subject M to z, but the main direction is not necessarily limited to these directions.
- the main direction is determined after arranging the ⁇ -ray detectors 3 arranged in a ring shape on the xy plane as shown in FIG. 5, but depending on the arrangement state of the ⁇ -ray detectors.
- the main direction may be determined as appropriate.
- the main direction may be determined based on the planes on the two detectors to be counted simultaneously.
- the larger direction is set as the main direction when determining the main direction, but the present invention is not necessarily limited to this. If the main direction is determined based on a certain rule, for example, the smaller direction may be set as the main direction.
- the plane in which the main direction is orthogonal is determined as the cross section of the image space through which the LOR passes.
- the plane in which the main direction is orthogonal is not necessarily determined as the cross section.
- a plane crossing at an angle other than 90 ° may be determined as the cross section.
- each row area separated by a predetermined interval is assigned as one area, but is divided into block areas other than the row area, and some of them are collectively assigned as one area. You may let them.
- the nuclear medicine diagnosis apparatus has been described.
- the present invention can be applied to a gamma camera having a collimator mechanism and SPECT.
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- High Energy & Nuclear Physics (AREA)
- Molecular Biology (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Spectroscopy & Molecular Physics (AREA)
- Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
- Biophysics (AREA)
- Optics & Photonics (AREA)
- Pathology (AREA)
- Radiology & Medical Imaging (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- General Health & Medical Sciences (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Nuclear Medicine (AREA)
Abstract
Description
田中栄一, 「PET画像の再構成法の現状と展望」,日本放射線技術学会雑誌, 浜松ホトニクス株式会社, Vol.62, No.6, 771-777,(2006). A. Fukano, F. Nakayama, H. Kubo, "Performance evaluation of relaxed block-iterative algorithms for 3-D PET reconstruction," IEEE Trans. Nucl. Sci., Vol.5, pp.2830-2834, 2004. H. Yang, M. Li, K. Koizumi, and H. Kubo, "Accelerating backprojections via CUDA architecture," in Proceedings of the 9th International Meeting on Fully Three-Dimensional Image Reconstruction in Radiology and Nuclear Medicine, pp.52-55, Lindau, Germany, July 2007. Guillem Pratx, Craig S. Levin et al, "Fast, Accurate and Shift-Varying Line Projections for Iterative Reconstruction Using the GPU", IEEE Trans. Med. Imaging., Vol.28, No 3, pp.435-445, 2009.
すなわち、この発明の核医学用データ処理方法は、放射線を検出することにより得られた事象データから生成されたリストデータをLOR毎に並列に順投影演算して順投影処理後データを得る順投影処理工程と、前記順投影処理後データあるいはリストデータを画像空間の断面を分割した領域毎に並列に逆投影演算する逆投影処理工程とを備えることを特徴とするものである。
10 … 演算処理部
d … (方向)ベクトル
M … 被検体
図5に示すように、水平軸をx、垂直軸をyとし、被検体Mの体軸をzとする。リング状に配置されたγ線検出器3をxy平面上に並べたときには、体軸zは奥行き方向の軸になる。図6に示すように、LORに平行な方向ベクトルをd(dx,dy,dz)と定義し(ただし||d||=1、単位ベクトル)、LORのベクトルdとγ線検出器3の配置方向とに基づいて、LORの主方向を決定する。γ線検出器3はxy平面上で配置されているので、ベクトルd(dx,dy,dz)のうちdxとdyとの大きさを比較して大きい方を主方向とする。図6では、x方向を主方向としているときの模式図である。このステップS1は、この発明における主方向決定工程に相当する。
ステップS1で決定された主方向に基づいてLORが通過する画像空間の断面を決定する。本実施例では、主方向が直交する面を、LORが通過する画像空間の断面として決定する。図6では、x方向を主方向としている図であるので、図6、図7に示すように断面はyz平面となる。このステップS2は、この発明における断面決定工程に相当する。
ステップS2で決定された断面(図6ではyz平面)に基づいてその断面におけるLORが横断する演算領域を決定する。まず、LORの始点となる断面(図6では「slice_start」で表記)およびLORの終点となる断面(図6では「slice_end」で表記)を決定して、始点となる断面上の座標と終点となる断面上の座標とを決定する。図6、図7では、断面をyz平面としているので、始点となる断面上の座標はy,z座標となり、同じく終点となる断面上の座標をy,z座標となる。ベクトルは一方向に延びているので、始点・終点のy座標のいずれか一方が最小値の座標(図7では「Min_0」で表記)となり、他方が最大値の座標(図7では「Max_0」で表記)となる。同じく、始点・終点のz座標のいずれか一方が最小値の座標(図7では「Min_1」で表記)となり、他方が最大値の座標(図7では「Max_1」で表記)となる。つまり、断面上では、LORは各断面上の最小値の座標位置から最大値の座標位置を横切ることになる。したがって、LORが横断する演算領域は、図7に示すようにMin_0,Max_0,Min_1およびMax_1で区切られた領域となる。このステップS3は、この発明における演算領域決定工程に相当する。
ステップS3で決定された演算領域でリストデータについて並列に演算処理を行う。具体的には、それぞれのボクセルjについてシステムマトリックスを求めて、決定された演算領域の範囲で上記(1)式中のFPで表記された部分において画素値と乗算して加算することにより投影値の更新を行う。上記(1)式での並列演算をLOR毎に行う。投影値を代入するメモリについては、1サブセット当たり、1イベント当たり1要素を書き込む構成となり、例えば図8のように書き込まれる。図8中の「Thread_0」、「Thread_1」、「Thread_2」、「Thread_3」、…は、書き込まれる並列データの処理単位を示す。各LORは、1サブセット当たり各スレッドに一対一に対応している。
ステップS4で順投影処理が行われた順投影処理後データあるいはリストデータについて逆投影処理を行う。この場合には、LORが通過する画像空間の各断面を複数の領域に分割して、分割されて分散された複数の領域を1つの領域に所属させて、所属されたそれぞれの領域毎に並列に演算を行う。具体的には、所属されたそれぞれの領域で上記(1)式中のBPで表記された部分において加算することにより逆投影値の更新を行う。
Claims (6)
- 放射線を検出することにより得られた事象データから生成されたリストデータをLOR毎に並列に順投影演算して順投影処理後データを得る順投影処理工程と、
前記順投影処理後データあるいはリストデータを画像空間の断面を分割した領域毎に並列に逆投影演算する逆投影処理工程と
を備えることを特徴とする核医学用データ処理方法。 - 請求項1に記載の核医学用データ処理方法において、
前記LORのベクトルと前記放射線を検出する検出器の配置方向とに基づいて前記LORの主方向を決定する主方向決定工程と、
その主方向決定工程で決定された前記主方向に基づいて前記LORが通過する画像空間の断面を決定する断面決定工程と、その断面決定工程で決定された前記断面に基づいてその断面における前記LORが横断する演算領域を決定する演算領域決定工程とを備え、
前記順投影処理工程は、前記演算領域毎に並列に演算し、
前記逆投影処理工程は、前記演算領域内を複数の領域に分割した領域毎に並列に演算することを特徴とする核医学用データ処理方法。 - 請求項1または請求項2に記載の核医学用データ処理方法において、
前記逆投影処理工程は、前記断面を複数の領域に分割して、分割されて分散された複数の領域を1つの領域に所属させて、所属されたそれぞれの領域毎に並列に演算を行うことを特徴とする核医学用データ処理方法。 - 請求項3に記載の核医学用データ処理方法において、
所定間隔離れた複数の行領域を前記1つの領域として所属させることを特徴とする核医学用データ処理方法。 - 請求項1から請求項4のいずれかに記載の核医学用データ処理方法において、前記順投影処理工程および前記逆投影処理工程を複数回実行して、画像を逐次に近似して更新する逐次近似法を用いることを特徴とする核医学用データ処理方法。
- 放射線を検出することにより得られた事象データから生成されたリストデータをLOR毎に並列に順投影演算して順投影処理後データを得る順投影処理手段と、
前記順投影処理後データを画像空間の断面を分割した領域毎に並列に逆投影演算する逆投影処理手段と
を備えることを特徴とする核医学診断装置。
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2011529700A JP5263402B2 (ja) | 2009-09-04 | 2009-09-04 | 核医学用データ処理方法および核医学診断装置 |
CN200980161257.2A CN102483459B (zh) | 2009-09-04 | 2009-09-04 | 核医学用数据处理方法以及核医学诊断装置 |
US13/388,792 US8987674B2 (en) | 2009-09-04 | 2009-09-04 | Data processing method for nuclear medicine, and a nuclear medicine diagnostic apparatus |
PCT/JP2009/004381 WO2011027402A1 (ja) | 2009-09-04 | 2009-09-04 | 核医学用データ処理方法および核医学診断装置 |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/JP2009/004381 WO2011027402A1 (ja) | 2009-09-04 | 2009-09-04 | 核医学用データ処理方法および核医学診断装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2011027402A1 true WO2011027402A1 (ja) | 2011-03-10 |
Family
ID=43648964
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2009/004381 WO2011027402A1 (ja) | 2009-09-04 | 2009-09-04 | 核医学用データ処理方法および核医学診断装置 |
Country Status (4)
Country | Link |
---|---|
US (1) | US8987674B2 (ja) |
JP (1) | JP5263402B2 (ja) |
CN (1) | CN102483459B (ja) |
WO (1) | WO2011027402A1 (ja) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101356881B1 (ko) | 2012-04-02 | 2014-02-12 | 한국과학기술원 | 고해상도 양전자 방출 단층 촬영에서 병렬 처리를 위해 영상을 재구성하는 방법 및 장치 |
WO2015056299A1 (ja) * | 2013-10-15 | 2015-04-23 | 株式会社島津製作所 | 断層画像処理方法およびそれを用いた放射型断層撮影装置 |
JP2019120530A (ja) * | 2017-12-28 | 2019-07-22 | 浜松ホトニクス株式会社 | 画像処理装置および画像処理方法 |
WO2021186504A1 (ja) * | 2020-03-16 | 2021-09-23 | 株式会社島津製作所 | リストモード画像再構成方法および核医学診断装置 |
Families Citing this family (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9111381B2 (en) * | 2010-01-27 | 2015-08-18 | Koninklijke Philips N.V. | Shift-varying line projection using graphics hardware |
NL2010267C2 (en) * | 2013-02-07 | 2014-08-11 | Milabs B V | High energy radiation detecting apparatus and method. |
US9839401B2 (en) * | 2016-01-19 | 2017-12-12 | Shimadzu Corporation | Radiation tomography apparatus |
EP3350776B1 (en) | 2016-04-20 | 2021-09-22 | Shanghai United Imaging Healthcare Co., Ltd. | System and method for image reconstruction |
WO2017214766A1 (zh) * | 2016-06-12 | 2017-12-21 | 上海联影医疗科技有限公司 | 正电子发射断层成像系统及其图像重建方法 |
JP6932253B2 (ja) * | 2017-09-21 | 2021-09-08 | コーニンクレッカ フィリップス エヌ ヴェKoninklijke Philips N.V. | 陽電子放出断層撮影の偶発同時推定のための緩和された反復的最大尤度期待値最大化 |
CN107908363B (zh) * | 2017-11-10 | 2021-12-07 | 湖北锐世数字医学影像科技有限公司 | 基于cuda的pet符合事件筛选的方法、系统及装置 |
CN107945203A (zh) * | 2017-11-24 | 2018-04-20 | 中国科学院高能物理研究所 | Pet图像处理方法及装置、电子设备、存储介质 |
US11232612B2 (en) * | 2019-03-15 | 2022-01-25 | University Of Florida Research Foundation, Incorporated | Highly accurate and efficient forward and back projection methods for computed tomography |
CN111881412A (zh) * | 2020-07-28 | 2020-11-03 | 南京航空航天大学 | 一种基于cuda的pet系统矩阵计算方法 |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007232548A (ja) * | 2006-02-28 | 2007-09-13 | Hitachi Ltd | 核医学診断装置 |
Family Cites Families (15)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US5224037A (en) * | 1991-03-15 | 1993-06-29 | Cti, Inc. | Design of super-fast three-dimensional projection system for Positron Emission Tomography |
JP4764050B2 (ja) * | 2005-03-31 | 2011-08-31 | 株式会社日立製作所 | 核医学診断装置および核医学診断装置の冷却方法 |
US7498581B2 (en) * | 2005-10-05 | 2009-03-03 | Koninklijke Philips Electronics N.V. | Distributed iterative image reconstruction |
JP4679348B2 (ja) * | 2005-11-22 | 2011-04-27 | ジーイー・メディカル・システムズ・グローバル・テクノロジー・カンパニー・エルエルシー | X線ct装置 |
WO2007100954A2 (en) * | 2006-02-22 | 2007-09-07 | Koninklijke Philips Electronics, N.V. | Image reconstruction using data ordering |
US8314796B2 (en) * | 2006-02-24 | 2012-11-20 | The Board Of Trustees Of The Leland Stanford Junior University | Method of reconstructing a tomographic image using a graphics processing unit |
JP4656233B2 (ja) * | 2006-03-10 | 2011-03-23 | 株式会社島津製作所 | 核医学診断装置およびそれに用いられる診断システム |
JP2007271509A (ja) * | 2006-03-31 | 2007-10-18 | Shimadzu Corp | 核医学診断装置およびそれに用いられる診断システム |
DE102007020879A1 (de) * | 2006-05-10 | 2009-04-02 | Gachon University Of Medicine & Science Industry-Academic Cooperation Foundation | Verfahren und Vorrichtung für die äußerst schnelle Symmetrie- und SIMD- gestützte Projektion/Rückprojektion für die 3D-PET-Bildrekonstruktion |
US7876941B2 (en) * | 2007-03-09 | 2011-01-25 | Siemens Medical Solutions Usa, Inc. | Incorporation of axial system response in iterative reconstruction from axially compressed data of cylindrical scanner using on-the-fly computing |
US7983465B2 (en) * | 2007-05-09 | 2011-07-19 | Société De Commercialisation Des Produits De La Recherche Appliquée - Socpra Sciences Santé Et Humaines, S.E.C. | Image reconstruction methods based on block circulant system matrices |
EP2156408B1 (en) * | 2007-05-30 | 2021-03-17 | Koninklijke Philips N.V. | Pet local tomography |
US8359345B2 (en) * | 2008-05-09 | 2013-01-22 | Siemens Medical Solutions Usa, Inc. | Iterative algorithms for variance reduction on compressed sinogram random coincidences in PET |
US8000513B2 (en) * | 2008-09-22 | 2011-08-16 | Siemens Medical Solutions Usa, Inc. | System and method for 3D time of flight PET forward projection based on an exact axial inverse rebinning relation in fourier space |
US8265365B2 (en) * | 2010-09-20 | 2012-09-11 | Siemens Medical Solutions Usa, Inc. | Time of flight scatter distribution estimation in positron emission tomography |
-
2009
- 2009-09-04 JP JP2011529700A patent/JP5263402B2/ja active Active
- 2009-09-04 WO PCT/JP2009/004381 patent/WO2011027402A1/ja active Application Filing
- 2009-09-04 CN CN200980161257.2A patent/CN102483459B/zh not_active Expired - Fee Related
- 2009-09-04 US US13/388,792 patent/US8987674B2/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2007232548A (ja) * | 2006-02-28 | 2007-09-13 | Hitachi Ltd | 核医学診断装置 |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
KR101356881B1 (ko) | 2012-04-02 | 2014-02-12 | 한국과학기술원 | 고해상도 양전자 방출 단층 촬영에서 병렬 처리를 위해 영상을 재구성하는 방법 및 장치 |
WO2015056299A1 (ja) * | 2013-10-15 | 2015-04-23 | 株式会社島津製作所 | 断層画像処理方法およびそれを用いた放射型断層撮影装置 |
JPWO2015056299A1 (ja) * | 2013-10-15 | 2017-03-09 | 株式会社島津製作所 | 断層画像処理方法およびそれを用いた放射型断層撮影装置 |
JP2019120530A (ja) * | 2017-12-28 | 2019-07-22 | 浜松ホトニクス株式会社 | 画像処理装置および画像処理方法 |
WO2021186504A1 (ja) * | 2020-03-16 | 2021-09-23 | 株式会社島津製作所 | リストモード画像再構成方法および核医学診断装置 |
JPWO2021186504A1 (ja) * | 2020-03-16 | 2021-09-23 | ||
JP7302737B2 (ja) | 2020-03-16 | 2023-07-04 | 株式会社島津製作所 | リストモード画像再構成方法および核医学診断装置 |
Also Published As
Publication number | Publication date |
---|---|
US20120126125A1 (en) | 2012-05-24 |
US8987674B2 (en) | 2015-03-24 |
CN102483459B (zh) | 2016-08-24 |
JPWO2011027402A1 (ja) | 2013-01-31 |
CN102483459A (zh) | 2012-05-30 |
JP5263402B2 (ja) | 2013-08-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5263402B2 (ja) | 核医学用データ処理方法および核医学診断装置 | |
JP5152202B2 (ja) | ポジトロンct装置 | |
US8208599B2 (en) | Iterative reconstruction with enhanced noise control filtering | |
US9155514B2 (en) | Reconstruction with partially known attenuation information in time of flight positron emission tomography | |
US20100074500A1 (en) | System and method for 3d time of flight pet forward projection based on an exact axial inverse rebinning relation in fourier space | |
Yamaya et al. | First human brain imaging by the jPET-D4 prototype with a pre-computed system matrix | |
US8975587B2 (en) | Positron CT apparatus and a reconstruction method | |
US9480450B2 (en) | Scatter component estimating method | |
WO1998047103A1 (en) | Direct tomographic reconstruction | |
US9916670B1 (en) | Fast, efficient, and list-mode compatible tomographic image reconstruction using a novel quadratic surrogate | |
JP6014738B2 (ja) | 三次元画像の投影方法 | |
Szlávecz et al. | GPU-based acceleration of the MLEM algorithm for SPECT parallel imaging with attenuation correction and compensation for detector response | |
KR101493683B1 (ko) | 콘-빔 기반 반응선 재구성을 이용한 초고해상도 pet 영상 재구성 장치 및 그 방법 | |
JP2010266235A (ja) | ポジトロンct装置および再構成方法 | |
US10354417B2 (en) | Medical image processing apparatus and medical image diagnosis apparatus and medical image processing method | |
JP2010203806A (ja) | 画像処理装置、画像再構成システム、画像処理方法およびプログラム | |
US20230206516A1 (en) | Scatter estimation for pet from image-based convolutional neural network | |
WO2023228910A1 (ja) | 画像処理装置および画像処理方法 | |
JP6206501B2 (ja) | 断層画像処理方法およびそれにおける各工程を実行するための手段を備えた放射線放出型断層撮影装置 | |
Al-Saffar | Decomposing of projected Footprints of Voxels in Separable Footprints 3D CT Reconstruction Method | |
Zou et al. | Image Reconstruction for Source Trajectories with Kinks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
WWE | Wipo information: entry into national phase |
Ref document number: 200980161257.2 Country of ref document: CN |
|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 09848933 Country of ref document: EP Kind code of ref document: A1 |
|
ENP | Entry into the national phase |
Ref document number: 2011529700 Country of ref document: JP Kind code of ref document: A |
|
WWE | Wipo information: entry into national phase |
Ref document number: 13388792 Country of ref document: US |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 09848933 Country of ref document: EP Kind code of ref document: A1 |