Medical image processing apparatus, medical image processing method, and program

Abstract

The goal is to enable the generation of time-series PET (Positron Emission Tomography) images of a subject with a width wider than that of the detector. [Solution] The medical image processing apparatus of the embodiment comprises an acquisition unit and a reconstructed image generation unit. The acquisition unit acquires a plurality of scan data of radiation based on a radiopharmaceutical administered to a subject, which are detected in time series by a detector. The reconstructed image generation unit generates a time-series PET image of the subject with a width wider than the width of the detector, based on the acquired plurality of scan data.

Description

Technical Field

[0001] The embodiments disclosed in this specification and the drawings relate to a medical image processing apparatus, a medical image processing method, and a program.

Background Art

[0002] In recent years, an apparatus for performing Total body PET (Positron Emission Tomography) that can scan the entire body of a subject at once (hereinafter referred to as a "Total body PET apparatus") is known. The Total body PET apparatus includes a PET detector having a length sufficient to cover the entire body in the body axis direction of the subject. Thereby, in the Total body PET apparatus, different from a conventional PET apparatus, it is possible to observe a time-series image (whole body dynamic image) of the process from drug administration to the subject to its spread throughout the body.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] [[ID=~35]]The length of the detector provided in the Total body PET apparatus as described above needs to be about 2 m in the body axis direction, which is more than seven times the length of the detector provided in a conventional PET apparatus that is currently widespread. A rare metal is used for the detector, and its cost is extremely high. Also, compared with a conventional PET apparatus, there are concerns that the Total body PET apparatus is large-sized and has limitations in its installation location, and that the labor and time required for maintenance increase.

[0005] The problem that the embodiments disclosed herein and in the drawings aim to solve is to enable the generation of time-series PET images of a subject with a width wider than the width of the detector. However, the problems that the embodiments disclosed herein and in the drawings aim to solve are not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described later can also be positioned as other problems. [Means for solving the problem]

[0006] The medical image processing apparatus of the embodiment comprises an acquisition unit and a reconstructed image generation unit. The acquisition unit acquires a plurality of scan data, which are scan data of radiation based on a radiopharmaceutical administered to a subject and are detected in time series by a detector. The reconstructed image generation unit generates a time-series PET image of the subject with a width wider than the width of the detector, based on the acquired plurality of scan data. [Brief explanation of the drawing]

[0007] [Figure 1] A diagram showing an example of a PET apparatus according to the embodiment. [Figure 2A] A diagram showing an image obtained using a total body PET scanner. [Figure 2B] A diagram illustrating the relationship between whole-body images acquired by a total-body PET scanner. [Figure 3A] A diagram showing the time-series transition of the scan range by the PET apparatus according to this embodiment. [Figure 3B] A diagram showing whole-body image data according to the embodiment (relationship between actual scan range and estimated range). [Figure 4] A flowchart showing an example of the whole-body image generation process using a PET apparatus according to the embodiment. [Figure 5A] A diagram illustrating the flow of the scanning process performed by the PET apparatus according to this embodiment. [Figure 5B] A diagram showing the input and output data of the learning model according to the embodiment. [Figure 6A]This diagram illustrates the training data prepared based on whole-body scan data acquired by a total-body PET scanner. [Figure 6B] A diagram showing the input and output data of the learning model during the learning phase according to the embodiment. [Modes for carrying out the invention]

[0008] The medical image processing apparatus, medical image processing method, and program of the embodiment will be described below with reference to the drawings. The medical image processing apparatus is, for example, a PET apparatus or a PET-CT (Computed Tomography) apparatus. In the following description, the case in which the medical image processing apparatus is a PET apparatus will be used as an example.

[0009] <First Embodiment> Figure 1 is a diagram showing the configuration of a PET apparatus 1 according to an embodiment. The PET apparatus 1 is a medical diagnostic device that detects radiation emitted by radioactive substances contained in a radiopharmaceutical administered to a subject P, and determines the tendency for the radiopharmaceutical to accumulate based on the detected radiation dose. The PET apparatus 1 generates and displays an image (hereinafter also referred to as "PET image") corresponding to the detected radiation dose. This allows the person performing the PET examination (doctor, technician, etc.) to visually confirm whether or not there is a lesion in the subject P.

[0010] The PET scanner 1 uses a detector that cannot cover the entire body of the subject, and repeatedly collects partial images of the subject in a short period of time. Based on the multiple collected images, it generates information equivalent to Total Body PET (for example, a time-series dynamic image of the whole body, hereinafter simply referred to as "whole-body image"). This makes it possible to obtain whole-body images equivalent to Total Body PET, for example, from drug administration to its distribution throughout the body, without using a detector that can cover the entire body of the subject.

[0011] To generate a whole-body image based on data acquired by a detector that cannot cover the entire body, it is crucial to obtain the missing data. Figure 2A shows an image of an image taken by a Total Body PET scanner equipped with a detector wide enough to cover the entire body. A Total Body PET scanner can image the entire body of subject P at once. In the example shown in Figure 2A, the first whole-body image is taken at time t1 when the drug is administered through a vein in subject P's arm A1, the second whole-body image is taken at time t2 when the drug reaches the heart A2, the third whole-body image is taken at time t3 when the drug has spread further throughout the body and reaches the legs A3, and the fourth whole-body image is taken at time t4 when the drug reaches the head A4. In this way, imaging with a Total Body PET scanner allows for the acquisition of a time-series whole-body image of the process from drug administration to the distribution of the drug throughout the body.

[0012] Figure 2B illustrates the relationships between whole-body images acquired by a total-body PET scanner. In Figure 2B, when focusing on a specific region within a whole-body image (for example, a region including the heart, hereinafter referred to as the "region of interest ROI"), there are time-series data connections between the whole-body images along the horizontal axis (time axis), and also physical data connections along the vertical axis (body axis direction) due to blood flow relationships within the body. In other words, there are data correlations between whole-body images in both the horizontal axis (time axis) and the vertical axis (body axis direction). In the PET scanner 1 of this embodiment, processing is performed to estimate the missing data in each scan by focusing on these data correlations.

[0013] FIG. 3A is a diagram showing a time-series transition of a scan range by the PET device 1 according to the embodiment. As shown in FIG. 3A, in the PET device 1 including a detector that cannot cover the whole body, for example, scanning (data collection) is repeatedly performed from the head to the feet in the body axis direction of the subject P. For example, if the actual scan range of the PET device 1 is the actual scan range SR, at scan 1 (time T1), the head of the subject P is scanned, at scan 2 (time T2), the chest of the subject P is scanned, at scan 3 (time T3), the lower abdomen of the subject P is scanned, at scan 4 (time T4), the legs of the subject P are scanned, and thereafter, scanning for each part of the body of the same subject P is repeated. From the scan data for each part of the body of the subject P thus obtained, the missing data in each scan is estimated, and a whole body image for each scan is generated.

[0014] FIG. 3B is a diagram showing the whole body image data (relationship between the actual scan range and the extrapolation range) according to the embodiment. As shown in FIG. 3B, at scan 1 (time T1), the head of the subject P is the actual scan range SR. In this case, the data of the extrapolation range ER, which is the missing data (data other than the head) in scan 1, is extrapolated, and a whole body image including the actual scan range SR and the extrapolation range ER in scan 1 is generated. Here, since the distribution of the drug changes from moment to moment, the time (scan interval) in each scan is set shorter than before.

[0015] <Configuration of PET device> Returning to FIG. 1, the PET device 1 includes, for example, a gantry device 10, a couch device 30, and a console device 40. In FIG. 1, for convenience of explanation, both a view of the gantry device 10 as seen from the Z-axis direction and a view as seen from the X-axis direction are shown. However, in reality, there is only one gantry device 10 included in the PET device 1. In the present embodiment, the central axis of the frame 13 in the non-tilted state or the longitudinal direction of the top plate 33 of the couch device 30 is defined as the Z-axis direction, an axis that is orthogonal to the Z-axis direction and horizontal with respect to the floor surface is defined as the X-axis direction, and a direction that is orthogonal to the Z-axis direction and perpendicular to the floor surface is defined as the Y-axis direction. The PET device 1 and / or the console device 40 is an example of the "medical image processing device" in the claims.

[0016] The gantry device 10 includes, for example, a radiation detector 11, a data acquisition system (hereinafter referred to as "DAS: Data Acquisition System") 12, a frame 13, and a control device 14.

[0017] The radiation detector 11 detects radiation such as gamma rays emitted from the subject P (more specifically, the radioactive drug administered to the subject P). The radiation detector 11 outputs an electrical signal (which may be an optical signal or the like) corresponding to the amount of detected radiation (radiation dose) to the DAS 12. The radiation detector 11 is formed in a cylindrical shape and is arranged to surround the imaging port formed in the gantry device 10. In the radiation detector 11, for example, a plurality of PET detector elements 110 are arranged in the circumferential direction and the central axis direction. Each PET detector element 110 detects radiation emitted from the subject P located in the imaging port and radiated to the surroundings. The radiation detector 11 outputs an electrical signal corresponding to the radiation dose detected by each PET detector element 110 to the DAS 12. The radiation detector 11 is an example of the "detector" in the claims.

[0018] DAS12 has, for example, an amplifier, an integrator, and an A / D converter. The amplifier performs amplification processing on the electrical signals output by each PET detection element 110 of the radiation detector 11. The integrator integrates the amplified electrical signals at predetermined time intervals. The A / D converter converts the electrical signal indicating the integration result by the integrator into a digital signal. DAS12 outputs the collected data based on the digital signal to the console device 40. The collected data is, for example, a digital value representing the radiation dose collected for each position within the radiation detector 11 where the PET detection elements 110 are arranged.

[0019] The frame 13 is an annular member that supports the radiation detector 11 and DAS12. The frame 13 is not limited to an annular member as long as it can support the radiation detector 11 and DAS12, and may be a member such as an arm.

[0020] In addition, for example, when the medical information processing device of the present embodiment is a PET-CT device, the frame 13 may be a rotating frame that fixedly holds an X-ray tube that generates X-rays to irradiate the subject P and an X-ray detector that detects the intensity of the X-rays that have passed through the subject P and entered, at opposing positions, and supports them rotatably about the subject P introduced therein.

[0021] The control device 14 includes a processing circuit having a processor such as a CPU (Central Processing Unit). The control device 14 receives input signals from an input interface 43 attached to the console device 40 or the frame device 10 and controls the operation of the frame device 10 and the bed device 30. For example, the control device 14 tilts the frame device 10 or moves the top plate 33 of the bed device 30. When tilting the frame device 10, the control device 14 tilts the frame 13 about an axis parallel to the Z-axis direction based on the tilt angle input to the input interface 43. The control device 14 knows the tilt angle of the frame 13 by the output of a sensor (not shown), etc. The control device 14 also provides the tilt angle of the frame 13 to the processing circuit 50 as needed. The control device 14 may be installed on the frame device 10 or on the console device 40. Furthermore, if the medical information processing device is a PET-CT scanner, the control device 14 may include a drive mechanism, such as a motor or actuator, to rotate the rotating frame of the frame 13.

[0022] The patient bed device 30 is a device that places and moves the subject P to be scanned and introduces it into the frame 13 of the stand device 10. The patient bed device 30 comprises, for example, a base 31, a patient bed drive device 32, a top plate 33, and a support frame 34. The base 31 includes a housing that supports the support frame 34 so as to be movable in the vertical direction (Y-axis direction). The patient bed drive device 32 includes a motor and an actuator. The patient bed drive device 32 moves the top plate 33 on which the subject P is placed along the support frame 34 in the longitudinal direction (Z-axis direction) of the top plate 33. The top plate 33 is a plate-shaped member on which the subject P is placed.

[0023] The bed drive device 32 may move not only the top plate 33 but also the support frame 34 in the longitudinal direction of the top plate 33. Conversely, the pallet device 10 may be movable in the Z-axis direction, and the movement of the pallet device 10 may be controlled so that the frame 13 is positioned around the subject P. Alternatively, both the pallet device 10 and the top plate 33 may be movable. Furthermore, the PET scanner 1 may be a device in which the subject P is scanned in a standing or sitting position. In this case, the PET scanner 1 has a subject support mechanism instead of the bed device 30, and the pallet device 10 moves the frame 13 in a direction perpendicular to the floor surface.

[0024] The console device 40 includes, for example, a memory 41, a display 42, an input interface 43, a network connection circuit 44, and a processing circuit 50. In this embodiment, the console device 40 is described as being separate from the mounting device 10, but the mounting device 10 may include some or all of the components of the console device 40.

[0025] Memory 41 can be implemented using semiconductor memory elements such as ROM (Read Only Memory), RAM (Random Access Memory), or flash memory, or a hard disk drive (HDD), or an optical disc. Memory 41 stores data such as collected data output by DAS 12, sinograms generated based on the collected data, and reconstructed images (PET images) generated based on the sinograms. Memory 41 also stores the learning model M, which will be described later.

[0026] The sinogram is data that shows the radiation dose represented by each collected data output by DAS12, for each position and angle within the radiation detector 11 where the PET detection element 110 is located. For example, if there is a lesion in the body of subject P, the radioactive drug will accumulate at the location of this lesion, and more radiation will be detected. Therefore, the location of the lesion in the body of subject P can be identified from the radiation dose shown in the sinogram. The PET image is a visualized image that allows the person performing the PET scan to visually confirm the radiation dose at each location shown in the sinogram.

[0027] These data may be stored not in memory 41 (or in addition to memory 41) but in external memory that the PET device 1 can communicate with. The external memory is controlled by a cloud server, for example, by a cloud server that manages the external memory and accepts read / write requests. The external memory is implemented by a system called PACS (Picture Archiving and Communication Systems). PACS is a medical image management system that systematically stores images taken by various imaging diagnostic devices.

[0028] The display 42 displays various types of information. For example, the display 42 displays images generated by the processing circuit 50 (e.g., whole-body images) and GUI (Graphical User Interface) images that accept various operations from the operator of the PET device 1 (e.g., a doctor or technician). The display 42 can be, for example, a liquid crystal display (LCD), a CRT (Cathode Ray Tube) display, or an organic EL (Electroluminescence) display. The display 42 may be provided on the stand device 10. The display 42 may be a desktop type or a display device (e.g., a tablet terminal) that can communicate wirelessly with the main body of the console device 40.

[0029] The input interface 43 receives various input operations from the operator of the PET device 1 and outputs an electrical signal indicating the content of the received input operation to the processing circuit 50. For example, the input interface 43 receives input operations such as collection conditions when collecting scan data, generation conditions when generating a sinogram, reconstruction conditions when reconstructing a PET image, and image processing conditions when generating a post-processed image from a PET image. The input interface 43 includes, for example, a mouse, keyboard, trackball, switch, button, joystick, touch panel, etc. The input interface 43 may also be a user interface that accepts audio input, such as a microphone. If the input interface 43 is a touch panel, the input interface 43 may also have the display function of the display 42.

[0030] In this specification, the term "input interface" is not limited to those equipped with physical operating components such as a mouse or keyboard. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device located separately from the device and outputs this electrical signal to a control circuit is also included as an example of an input interface.

[0031] The network connection circuit 44 includes, for example, a network card having a printed circuit board, or a wireless communication module. The network connection circuit 44 implements an information communication protocol according to the type of network to be connected. The network includes, for example, a LAN (Local Area Network), a WAN (Wide Area Network), the Internet, a cellular network, a dedicated line, etc.

[0032] The processing circuit 50 controls the overall operation of the PET apparatus 1. The processing circuit 50 performs, for example, system control functions 51, acquisition functions 52, inference functions 53, reconstructed image generation functions 54, display control functions 55, learning functions 56, etc. The processing circuit 50 realizes these functions, for example, by having a hardware processor execute a program (software) stored in memory 41.

[0033] A hardware processor refers to circuits such as CPUs, GPUs (Graphics Processing Units), Application Specific Integrated Circuits (ASICs), and programmable logic devices (e.g., Simple Programmable Logic Devices (SPLDs), Complex Programmable Logic Devices (CPLDs), and Field Programmable Gate Arrays (FPGAs)). Instead of storing the program in memory 41, the hardware processor may be configured to directly embed the program within its circuitry. In this case, the hardware processor performs its functions by reading and executing the program embedded within the circuitry. A hardware processor is not limited to being configured as a single circuit; it may also be configured as a single hardware processor by combining multiple independent circuits to implement each function. Furthermore, multiple components may be integrated into a single hardware processor to implement each function. Alternatively, multiple components may be incorporated into a single dedicated LSI to implement each function. Here, the program (software) may be stored in advance in a storage device that constitutes memory 41, such as ROM, RAM, HDD, or flash memory (a storage device equipped with a non-transient storage medium), or it may be stored in a removable storage medium (a non-transient storage medium) such as a DVD or CD-ROM, and installed in the storage device of the console device 40 when the storage medium is inserted into a drive device provided in the console device 40. Alternatively, the program (software) may be downloaded in advance from another computer device via a network connected by the network connection circuit 44 and installed in the storage device of the console device 40.

[0034] Each component of the console device 40 or the processing circuit 50 may be distributed and implemented by multiple hardware components. The processing circuit 50 may not be implemented in the configuration of the console device 40, but rather by a processing unit that can communicate with the console device 40. The processing unit may be, for example, a workstation connected to one PET device, or a device (e.g., a cloud server) connected to multiple PET devices that performs processing equivalent to that of the processing circuit 50 described below in a batch. In other words, the configuration of this embodiment can also be implemented as a PET examination system in which a PET device and other processing units are connected via a network. In this case, other processing units such as workstations are examples of "medical information processing units" in the claims.

[0035] The system control function 51 controls various functions of the processing circuit 50 based on input operations received by the input interface 43, for example. The system control function 51 also controls various functions of the support structure 10 by issuing instructions to the DAS 12, the control device 14, and the bed drive device 32 based on input operations received by the input interface 43, for example. The system control function 51 is an example of a "system control unit" in the claims.

[0036] The acquisition function 52 acquires scan data of radiation based on a radioactive drug administered to subject P, and acquires multiple scan data detected in time series by the radiation detector 11. The acquisition function 52 also acquires multiple scan data detected by the radiation detector 11 by repeatedly scanning each body part in the axial direction of subject P in time series. The acquisition function 52 is an example of the "acquisition unit" in the claims.

[0037] The inference function 53 infers missing scan data (hereinafter referred to as "missing scan data") corresponding to parts of the subject P other than the scan target area in each scan, based on the relationship between multiple scan data acquired by the acquisition function 52, either in the body axis direction or in the time axis direction of the time series. The inference function 53 infers the missing scan data using a learning model M. The inference function 53 infers whole-body scan data based on output data obtained by inputting multiple scan data detected by the radiation detector 11, which are obtained by repeatedly scanning each part of the subject P in the body axis direction in a time series, into the learning model M. The inference function 53 is an example of the "inference unit" in the claims.

[0038] Furthermore, instead of using a learning model, the inference function 53 may infer missing scan data based on multiple scan data acquired by the acquisition function 52. The inference function 53 may also infer missing scan data by performing numerical calculations, such as arranging the acquired scan data in chronological order and blending the scan data.

[0039] The reconstructed image generation function 54 performs predetermined processing on the collected data output by the DAS 12 to generate a sinogram, performs predetermined reconstruction processing on the generated sinogram using methods such as filtered back projection and iterative reconstruction to generate a PET image, and stores the generated PET image in the memory 41. Based on multiple scan data acquired by the acquisition function 52, the reconstructed image generation function 54 generates a time-series PET image of the subject P with a width wider than the width of the radiation detector 11. Based on multiple scan data detected by the radiation detector 11, which has a width that does not cover the entire body of the subject P, the reconstructed image generation function 54 generates a time-series PET image of the entire body of the subject P. The reconstructed image generation function 54 is an example of the "reconstructed image generation unit" in the claims.

[0040] The display control function 55 controls the display mode of the display 42. For example, the display control function 55 controls the display 42 to display PET images generated by the processing circuit 50, GUI images that accept various operations from the operator of the PET device 1, etc. The display control function 55 is an example of a "display control unit" in the claims.

[0041] The learning function 56 generates a learning model M that, when multiple scan data are input, scanned for each body part in the axial direction of the subject P, outputs whole-body scan data corresponding to the entire body of subject P, including missing scan data other than the scanned body part in each scan. The learning model M is generated by learning using techniques such as a Convolutional Neural Network (CNN) or a Deep Neural Network (DNN). A CNN is a neural network in which several layers such as convolutional layers and pooling layers are connected. A DNN is a neural network in which layers of any form are connected in multiple layers. The learning model M may also be generated using any machine learning technique such as gradient methods such as SGD (Stochastic Gradient Descent), Momentum SGD, AdaGrad, RMSprop, AdaDelta, Adam (Adaptive moment estimation), logistic regression analysis, or techniques based on support vector machines. The learning function 56 is an example of the "learning unit" in the claims.

[0042] Furthermore, the learning model M may be trained to output a whole-body image corresponding to the entire body of subject P, including any missing images other than the body parts scanned in each scan, when an image generated based on multiple scan data obtained from scanning each body part of subject P along the body axis is input. Alternatively, the learning model M may be trained to output a whole-body sinogram corresponding to the entire body of subject P, including any missing sinograms other than the body parts scanned in each scan, when a sinogram generated based on multiple scan data obtained from scanning each body part of subject P along the body axis is input.

[0043] <Whole-body image generation process> Next, the flow of whole-body image generation processing in the PET device 1 will be described. Figure 4 is a flowchart showing an example of whole-body image generation processing by the PET device 1 according to the embodiment. The processing shown in Figure 4 starts when the operator has completed setting the imaging conditions (imaging protocol) via the input interface 43 and the subject P to be examined is placed on the examination table 30. The data acquisition method in the processing shown in Figure 4 may be the Step & shoot method or the Continuous method in which data is collected while the examination table 30 moves continuously. In the following, the Step & shoot method will be used as an example.

[0044] First, the system control function 51 controls the rigging device 10 based on an input operation (scan start operation) received by the operator via the input interface 43, for example, to perform the first bed scan of the subject P (scan 1, for example, a scan of the head) (step S101). As a result, the acquisition function 52 acquires the scan data corresponding to scan 1 and stores the acquired scan data in the memory 41.

[0045] Next, the system control function 51 controls the rigging device 10 to perform a scan of the next bed of the subject P (scan 2, for example, a chest scan) (step S103). As a result, the acquisition function 52 acquires the scan data corresponding to scan 2 and stores the acquired scan data in the memory 41.

[0046] Next, the system control function 51 determines whether the scan of the last bed has been completed based on the preset shooting conditions (shooting protocol) (step S105). If the system control function 51 determines that the scan of the last bed has not been completed, it returns to step S103 and repeats the same process.

[0047] Figure 5A is a diagram illustrating the flow of the scanning process performed by the PET apparatus 1 according to the embodiment. In scan 1 (1st bed) performed at time T1, the first scan data SD_1 of the subject P's head is acquired. Next, in scan 2 (2nd bed) performed at time T2, the second scan data SD_2 of the subject P's chest is acquired. Next, in scan 3 (3rd bed) performed at time T3, the third scan data SD_3 of the subject P's lower abdomen is acquired. Next, in scan 4 (4th bed) performed at time T4, the fourth scan data SD_4 of the subject P's legs is acquired. Subsequently, a set of four scans, consisting of the 5th scan data SD_5 (head), 6th scan data SD_6 (chest), 7th scan data SD_7 (lower abdomen), and 8th scan data SD_8 (legs), is repeatedly performed. Note that the series of scans may be performed in one direction from head to legs as described above, or they may be performed in a round trip, for example, from head to legs and from legs to head.

[0048] Returning to Figure 4, in step S105, if the system control function 51 determines that the scan of the last bed is complete, the estimation function 53 estimates the missing scan data corresponding to parts of the subject P other than the scan target area in each scan, based on the multiple scan data acquired by the acquisition function 52 (step S107). The estimation function 53 estimates whole-body scan data by inputting the scan data detected by the radiation detector 11 into the learning model M, by repeatedly scanning each part of the subject P's body in the axial direction in a time series.

[0049] Figure 5B shows the input and output data of the learning model M according to the embodiment. The input data is scan data (1st scan data SD_1 to 8th scan data SD_8, ...) obtained by scanning each part of the subject P's body acquired in steps S101 and S103 above. The output data is whole-body scan data corresponding to each scan performed in steps S101 and S103 above. As shown in Figure 5A, this output data is, for example, the 1st whole-body scan data SD_11 for scan 1. This 1st whole-body scan data SD_11 includes the 1st scan data SD_1, which is the actual scan data of the subject P's head that was actually scanned, and the estimated scan data of parts of the subject P other than the head.

[0050] Returning to Figure 4, the reconstruction image generation function 54 then generates a whole-body image of subject P using the whole-body scan data generated by the analogy function 53 (step S109).

[0051] Next, the display control function 55 displays the whole-body image of subject P generated by the reconstructed image generation function 54 on the display 42 (step S111). This allows the operator to confirm the whole-body image of each scan displayed on the display 42. This completes the processing of this flowchart.

[0052] Furthermore, the images displayed on the display 42 are not limited to whole-body images of subject P. For example, data from the subject's lower limbs (legs, etc.) may not be important depending on the pathology and may not need to be output. In such cases, the display control function 55 may display on the display 42 an image of the subject P that excludes the unimportant parts of subject P from the whole-body image of subject P generated by the reconstructed image generation function 54. Alternatively, the analogy function 53 may exclude unimportant data of the subject (for example, scan data of the lower limbs) from the analogy target. For example, unimportant data of the subject may be excluded from the output data of the learning model M. Alternatively, the reconstructed image generation function 54 may generate an image that excludes unimportant parts of subject P using the whole-body scan data generated by the analogy function 53.

[0053] <Learning Process> Next, the learning process by the learning function 56 will be explained. The learning function 56 uses whole-body scan data collected using a conventional Total Body PET device with a detector capable of covering the entire body as training data. From the whole-body scan data, scan data for each part of the subject P's body (hereinafter referred to as "extracted scan data") is extracted, and the pair of extracted scan data and the original whole-body scan data is used as training data.

[0054] Figure 6A illustrates the training data prepared based on whole-body scan data acquired by a Total Body PET scanner. As shown in Figure 6A, for example, first whole-body scan data TSD_1 to eighth whole-body scan data TSD_8 of subject P acquired by the Total Body PET scanner at each of the times T1 to T8 are prepared. Then, from each of these first whole-body scan data TSD_1 to eighth whole-body scan data TSD_8, first extracted scan data ESD_1 to eighth extracted scan data ESD_8 for each body part of subject P are extracted, and pairs of extracted scan data and the original whole-body scan data are used as training data. The above preprocessing is performed on whole-body scan data obtained from scans of multiple subjects, and a large number of training data are prepared.

[0055] Figure 6B shows the input and output data of the learning model M during the learning phase according to the embodiment. The input data consists of extracted scan data for each subject (for example, the first extracted scan data ESD_1 to the eighth extracted scan data ESD_8). Each of these extracted scan data is associated with scan time information. Alternatively, if the first whole-body scan data TSD_1 to the eighth whole-body scan data TSD_8 were acquired in accordance with the actual scan conditions (scan interval, etc.) during the examination of the subject by the PET device 1 of this embodiment, then the scan time information does not need to be associated with this extracted scan data. The output data consists of whole-body scan data for each subject (for example, the first whole-body scan data TSD_1 to the eighth whole-body scan data TSD_8). The learning function 56 uses the training data (pairs of input data and output data) prepared in this way to perform the learning process of the learning model M.

[0056] In other words, the learning function 56 generates a learning model M by learning training data that includes whole-body scan data detected by a detector capable of covering the entire body of the subject, and extracted scan data extracted from the whole-body scan data for each part of the subject P's body.

[0057] As described above, according to the first embodiment, it is possible to generate time-series PET images of a subject with a width wider than the width of the detector. In particular, it is possible to generate time-series whole-body images even without a detector that can cover the entire body. This reduces the cost of the detector. In addition, it eliminates the need for large devices such as total-body PET scanners, and reduces the effort and time required for maintenance.

[0058] <Second Embodiment> Next, a second embodiment will be described. The second embodiment differs from the first embodiment in that it makes the scan time for each part of the subject's body variable, taking into consideration biologically important body parts. In the following description, components and functions identical to those of the first embodiment will be denoted by the same reference numerals as in the first embodiment, and detailed explanations will be omitted.

[0059] The system control function 51 controls the scan to extend the scan time for biologically important body parts on the subject. For example, for areas important from a blood flow perspective, such as the heart, or areas where a target pathology is likely to occur, extending the acquisition time when scanning these areas can generate a more accurate whole-body image. Similarly, assuming that the drug enters the heart at a specific time after the start of drug injection, the data at that time is likely to be important. Based on research results from Total Body PET with whole-body detectors, areas with large time-series changes can be identified, and the acquisition time for specific areas at such specific times can be considered important, allowing for an extended scan time.

[0060] Important areas may be set based on input operations by the operator received through the input interface 43. Alternatively, important areas may be set automatically according to the imaging conditions (imaging protocol). An example of control to lengthen the scan time is to increase the number of scans corresponding to important areas.

[0061] Furthermore, during the training phase of the learning model M using the learning function 56, the accuracy of inferring important body parts can be improved by including more data on biometrically important body parts on the subject in the training data. This makes it possible to generate whole-body images with greater accuracy.

[0062] Furthermore, for data from areas other than the critical parts, whole-body images may be output as reference data. In other words, since the data acquisition time for less important parts is relatively shorter, treating this as reference data and explicitly indicating that it is reference data (for example, displaying a note such as "This part is reference data") allows the operator to focus on reviewing data from important areas and treat other less important parts only as reference, thereby reducing the burden of image diagnosis.

[0063] In other words, the system control function 51 performs the scan under conditions where the data collection time for each part of the subject's body in the axial direction is variable, based on the subject's biological information. The display control function 55 displays a time-series PET image of the subject's entire body, generated based on the estimated missing scan data, and displays non-important parts of the subject's body as reference data.

[0064] <Third Embodiment> Next, a third embodiment will be described. The third embodiment differs from the second embodiment in that the importance of extending the scan time is determined not from a bio-informatics perspective, but by considering the parts of the body (such as the heart) that have a large influence on the results (output data) during the training of the learning model M. In the following description, components and functions identical to those of the first and second embodiments will be denoted by the same reference numerals as those of the first and second embodiments, and detailed explanations will be omitted.

[0065] During the training phase of the learning model M by the learning function 56, the accuracy of the learning is improved by including more scan data in the training data that corresponds to parts of the body (such as the heart) that have features that have a large influence on the analogy results (output data) of the learning model M (by extending the scan time for those parts). The analogy function 53 can improve its analogy accuracy by using the learning model M trained with the training data prepared in this way. The scan time for important parts may also be extended during the actual scan of the subject P. Alternatively, the learning may be performed without creating such data during training, and the scan time for important parts may be extended only during the actual scan.

[0066] Furthermore, for data from areas other than the critical parts, whole-body images may be output as reference data. In other words, since the data acquisition time for less important parts is relatively shorter, treating this as reference data and explicitly indicating that it is reference data (for example, displaying a note such as "This part is reference data") allows the operator to focus on reviewing data from important areas and treat other less important parts only as reference, thereby reducing the burden of image diagnosis.

[0067] In other words, the system control function 51 performs the scan under conditions where the data collection time for each body part in the axial direction of the subject is variable, based on the magnitude of the impact on the output of the learning model. The display control function 55 displays a time-series PET image of the entire body of the subject, generated based on the estimated missing scan data, and displays less important parts of the subject's body as reference data.

[0068] While several embodiments of the present invention have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of Symbols]

[0069] 1...PET scanner, 10...Stand system, 11...Radiation detector, 12...Data acquisition system (DAS), 13...Frame, 14...Control device, 30...Bed system, 31...Base, 32...Bed drive system, 33...Tabletop, 34...Support frame, 40...Console system, 41...Memory, 42...Display, 43...Input interface, 44...Network connection circuit, 50...Processing circuit, 51...System control function, 52...Acquisition function, 53...Inference function, 54...Reconstructed image generation function, 55...Display control function, 56...Learning function

Claims

1. A unit that acquires scan data of radiation based on a radioactive drug administered to a subject, wherein the unit acquires multiple scan data detected in a time series by a detector, A reconstructed image generation unit generates a time-series PET (Positron Emission Tomography) image of the subject with a width wider than the width of the detector, based on the acquired plurality of scan data. A medical image processing device equipped with [a specific feature].

2. The reconstructed image generation unit generates a time-series PET image of the entire body of the subject based on the plurality of scan data detected by the detector, which has a width that does not cover the entire body of the subject. The medical image processing apparatus according to claim 1.

3. The acquisition unit acquires the plurality of scan data detected by the detector by repeatedly scanning each body part in the axial direction of the subject in a time series. The system further includes an estimation unit that estimates missing scan data corresponding to areas of the subject other than the scanned area in each scan, based on the relationship between the acquired plurality of scan data in the body axis direction or the time axis direction of the time series. The medical image processing apparatus according to claim 1 or 2.

4. The aforementioned estimation unit uses a learning model to estimate the missing scan data. The medical image processing apparatus according to claim 3.

5. The system further includes a control unit that, based on the biological information of the subject, causes the system to perform a scan under conditions in which the data collection time for each body part in the axial direction of the subject is made variable. The medical image processing apparatus according to claim 3.

6. The system further includes a control unit that performs a scan under conditions in which the data collection time for each body part in the axial direction of the subject is made variable based on the magnitude of the influence on the output of the learning model. The medical image processing apparatus according to claim 4.

7. A display control unit that displays a time-series PET image of the entire body of the subject, generated based on the estimated missing scan data, further comprising a display control unit that displays non-essential parts of the subject's body as reference data. The medical image processing apparatus according to claim 3.

8. The aforementioned learning model is trained to output whole-body scan data corresponding to the entire body of the subject when it receives multiple scan data, each scanned for a different part of the subject's body along the axial direction of the subject's body. The estimation unit estimates the whole-body scan data based on the output data obtained by inputting the acquired plurality of scan data into the learning model. The medical image processing apparatus according to claim 4.

9. The system further comprises a learning unit that generates the learning model by learning training data which includes whole-body scan data detected by a detector capable of covering the entire body of the subject, and extracted scan data extracted from the whole-body scan data for each part of the subject's body. The medical image processing apparatus according to claim 4.

10. Computers This involves obtaining scan data of radiation based on a radioactive drug administered to a subject, where multiple scan data are acquired in a time series by a detector. Based on the acquired plurality of scan data, a time-series PET image of the subject with a width wider than the width of the detector is generated. Medical image processing methods.

11. On the computer, This involves obtaining scan data of radiation based on a radioactive drug administered to a subject, where multiple scan data are acquired in a time series by a detector. Based on the acquired plurality of scan data, a time-series PET image of the subject with a width wider than the width of the detector is generated. program.

Images