JP2020067289A - Medical apparatus and learning data acquisition method - Google Patents

Abstract

To acquire information on linearity correction on the basis of a nuclear medicine image acquired without using a phantom collimator.SOLUTION: A medical apparatus includes an acquisition unit and a processing unit. The acquisition unit acquires a first image acquired without using a collimator in a nuclear medicine diagnostic apparatus and information regarding an arrangement of optical sensors. The processing unit outputs information regarding linearity correction for a learnt model that outputs information about the linearity correction on the basis of the first image and the information about the arrangement of optical sensors by entering the first image and the information about the arrangement of optical sensors.SELECTED DRAWING: Figure 7

Description

  Embodiments of the present invention relate to a medical device and a learning data acquisition method.

  In nuclear medicine diagnostic equipment such as SPECT (Single Photon Emission computed Tomography) equipment, drugs (blood flow markers, tracers) containing radioisotopes (hereinafter referred to as RI) are selectively applied to specific tissues or organs in the living body. Utilizing the property of being taken in, gamma rays emitted from RI distributed in the living body are detected by a gamma camera having a gamma ray detector arranged outside the living body. The nuclear medicine diagnostic apparatus can provide a functional image of a body organ or the like by generating a nuclear medicine image in which a dose distribution of gamma rays detected by a gamma ray detector is imaged.

  As the gamma ray detector, for example, an Anger type detector can be used. The incident position information of the gamma rays output from the Anger detector is obtained by weighted addition of the outputs of the photomultiplier tubes (Photo Multiplier Tubes, hereinafter referred to as PMTs) that are finely arranged. However, there is a difference in the sensitivity of the PMT between directly below and around the PMT, and because the sensitivity of the PMT has manufacturing variations and changes over time, the position information output by the Anger detector is distorted. However, the linearity cannot be guaranteed. Therefore, in general, for example, a linearity correction vector is created as information about linearity correction of the position information output by the Anger type detector during manufacturing, installation, maintenance, etc. of the Anger type detector. The linearity correction of the position information is performed using the property correction vector.

  An image used to create this type of linearity correction vector is a gamma ray from a distance when a lead plate with slit-shaped or dot-shaped holes (hereinafter referred to as phantom collimator) is fixed to the detection surface of the detector. An image in which only the counts corresponding to the positions of the gamma ray transmission holes of the phantom collimator are accumulated by irradiating (hereinafter, referred to as phantom image) is used. The linearity correction vector can be created by comparing the phantom image and the ideal position of the gamma ray transmission hole of the phantom collimator.

  However, the phantom image acquisition using the phantom collimator requires a long acquisition time in order to accumulate enough counts to accurately create the linearity correction vector. Further, when a high-dose radiation source is used in order to shorten the collection time, the risk of exposure of workers increases. In addition, since the phantom collimator is a heavy object of several kg to several tens of kg, the phantom image acquisition using the phantom collimator requires the work of attaching and detaching the phantom collimator, which is a heavy item, and is complicated. There is a risk of damaging other parts such as damage to the scintillator.

JP, 2011-137814, A

  The problem to be solved by the present invention is to acquire information regarding linearity correction based on a nuclear medicine image acquired without using a phantom collimator.

  The medical device according to the embodiment includes an acquisition unit and a processing unit. The acquisition unit acquires the first image acquired without using the collimator in the nuclear medicine diagnostic apparatus and the information regarding the arrangement of the optical sensors. The processing unit inputs the first image and the information regarding the arrangement of the photosensors to the learned model that outputs the information regarding the linearity correction based on the first image and the information regarding the arrangement of the photosensors. To output information about linearity correction. As the optical sensor, for example, PMT can be used.

The block diagram which shows one structural example of the medical device which concerns on 1st Embodiment. (A) is explanatory drawing which shows an example of a horizontal slit phantom collimator, (b) is explanatory drawing which shows an example of a horizontal slit phantom image. (A) is explanatory drawing which shows an example of a vertical slit phantom collimator, (b) is explanatory drawing which shows an example of a vertical slit phantom image. (A) is explanatory drawing which shows an example of the arrangement | positioning position information of PMT, (b) is explanatory drawing which shows an example of a whole surface irradiation image. Explanatory drawing which shows an example of a data flow when a learning linearity correction vector is produced | generated by the learning data production | generation function. Explanatory drawing which shows an example of the data flow at the time of learning of a linearity correction vector generation function. Explanatory drawing which shows an example of the data flow at the time of operation of a linearity correction vector generation function. (A) is an explanatory view showing an example of a data flow at the time of learning the linearity correction vector generation function in the case of further using information on the sensitivity and response function of the PMT as input data, and (b) is an example of a data flow at the time of operation FIG. Explanatory drawing which shows an example of a data flow when the 2nd linearity correction vector for learning which considers at least one influence of the radiation incident position dependence of the sensitivity of a scintillator, and the influence of an edge effect is produced | generated by the learning data generation function. . The block diagram which shows one structural example of the SPECT apparatus containing the medical device which concerns on 2nd Embodiment. FIG. 9 is a schematic block diagram showing an example of functions implemented by a processor of the processing circuit according to the second embodiment.

  Hereinafter, embodiments of a medical device and a learning data acquisition method will be described in detail with reference to the drawings.

(First embodiment)
FIG. 1 is a block diagram showing a configuration example of a medical device 10 according to the first embodiment. The medical device 10 has at least a storage circuit 13 and a processing circuit 15. FIG. 1 shows an example in which the medical device 10 further includes an input interface 11, a display 12, and a network connection circuit 14. The medical device 10 can be configured by, for example, a portable personal computer or the like. Further, the medical device 10 may be incorporated in the console device of the SPECT device.

  The input interface 11 is composed of a general input device such as a trackball, a switch button, a mouse, a keyboard, and a numeric keypad, and outputs an operation input signal corresponding to a user's operation to the processing circuit 15. The display 12 is composed of a general display output device such as a liquid crystal display or an OLED (Organic Light Emitting Diode) display.

  The storage circuit 13 has a configuration including a recording medium readable by a processor, such as a RAM (Random Access Memory), a semiconductor memory device such as a flash memory, a hard disk, an optical disc, and the like, and is used by the processing circuit 15. And parameter data and other data. Part or all of the programs and data in the recording medium of the storage circuit 13 may be downloaded by communication via the network 100, or may be provided to the storage circuit 13 via a portable storage medium such as an optical disk. May be.

  The network connection circuit 14 implements various information communication protocols according to the form of the network 100. The network connection circuit 14 connects to other electric devices via the network 100 according to the various protocols. The network 100 refers to all information communication networks that use telecommunications technology, and includes wireless / wired LANs such as hospital backbone LANs (Local Area Networks) and Internet networks, as well as telephone communication line networks, optical fiber communication networks, and cable communications. Includes networks and satellite communication networks.

  The medical apparatus 10 is connected to the nuclear medicine diagnostic apparatus 101 and the image server 102 via the network 100 so that data can be transmitted and received between them. The nuclear medicine diagnostic apparatus 101 is, for example, a SPECT apparatus including an Anger type detector.

  The processing circuit 15 realizes a function of integrally controlling the medical device 10. Further, the processing circuit 15 reads the program stored in the storage circuit 13 and executes the program to perform processing for outputting information regarding linearity correction based on the nuclear medicine image acquired without using the phantom collimator. The processor to execute.

  As shown in FIG. 1, the processor of the processing circuit 15 realizes a learning data generation function 21, an acquisition function 22, and a linearity correction vector generation function 23. Each of these functions is stored in the storage circuit 13 in the form of a program.

  The learning data generation function 21 generates information about linearity correction based on the phantom image acquired using the phantom collimator and the ideal position information of the phantom collimator. In the following description, an example in which the information regarding linearity correction is a linearity correction vector is shown.

  The linearity correction vector generated by the learning data generation function 21 is used as teacher data when learning the learned model according to this embodiment.

  The acquisition function 22 is an Anger type detector that wants to generate information regarding linearity correction, and acquires an image (hereinafter referred to as a total irradiation image) acquired by irradiating gamma rays from a distance on the entire surface of the scintillator without using a collimator. . The overall irradiation image is acquired from the nuclear medicine diagnostic device 101 or the image server 102. The acquisition function 22 is an example of an acquisition unit.

The linearity correction vector generation function 23 is an Anger-type detector that wants to generate information on linearity correction for a learned model that outputs information on linearity correction based on the overall irradiation image and information on the arrangement of photosensors. By inputting the overall irradiation image acquired in step 3 and the information about the placement of the optical sensor of the Anger type detector for which you want to generate information about linearity correction, you can generate the information about linearity correction for linearity correction of the Anger type detector. Output information about. The linearity correction vector generation function 23 is an example of a processing unit. In the following description, an example is shown in which the Anger detector has an array of PMTs as a photosensor and a scintillator as a phosphor provided in front of the photosensor.
Next, the linearity correction will be described.

  2A is an explanatory diagram showing an example of the horizontal slit phantom collimator 31, and FIG. 2B is an explanatory diagram showing an example of the horizontal slit phantom image 32. The horizontal slit phantom image 32 is acquired using the horizontal slit phantom collimator 31. 3A is an explanatory diagram showing an example of the vertical slit phantom collimator 33, and FIG. 3B is an explanatory diagram showing an example of the vertical slit phantom image 34. The vertical slit phantom image 34 is acquired using the vertical slit phantom collimator 33. 4A is an explanatory diagram showing an example of the arrangement position information 35 of the PMT 35a, and FIG. 4B is an explanatory diagram showing an example of the entire surface irradiation image 36.

  The Anger detector has a single crystal scintillator formed of sodium iodide (NaI) or the like. When a gamma ray enters the scintillator, the array of PMTs detects the light emitted from the scintillator and outputs a signal. The Anger type detector detects the incident position of the gamma ray based on this signal. However, the output signal of the PMT has a linear distortion due to the influence of the arrangement position of the PMT (see FIG. 4A). For this reason, the detected incident position is different from the actual position.

  For example, as shown in FIGS. 2B and 3B, the slit images depicted in the horizontal slit phantom image 32 and the vertical slit phantom image 34 include distortion instead of straight lines. In a place where the distortion is large and the gap between the slits is narrow, the density of the count amount is higher than in a place where the gap between the slits is wide.

  Conventionally, by comparing the horizontal slit phantom image 32 and the vertical slit phantom image 34 with the ideal position information (actually, the slit position information) of the corresponding horizontal slit phantom collimator 31 and vertical slit phantom collimator 33, or by dot shape There is known a method of creating a linearity correction vector by comparing an image using the phantom collimator of No. 1 with the ideal position information of the dot phantom collimator. However, the method of using the linearity correction vector using the phantom collimator has the image acquisition time, the risk of exposure of the operator, the complicated attachment / detachment work of heavy objects, and the risk of damaging members as described above.

  On the other hand, as shown in FIG. 4B, the entire surface irradiation image (an image obtained by irradiating the entire surface of the scintillator with gamma rays from a distance without using a collimator) 35 is projected from the center of each PMT 35a to the outside. This is an image in which irregularities reflecting the sensitivity difference and the arrangement position information 35 of the PMT 35a are drawn, and of course, an image including information related to linear distortion. Therefore, the processing circuit 15 according to the present embodiment arranges the overall irradiation image acquired by the Anger type detector for which information on linearity correction is to be generated and the optical sensor of the Anger type detector for which information on linearity correction is to be generated. Based on the information regarding the linearity correction, the information regarding the linearity correction of the Anger type detector for which the information regarding linearity correction is desired to be generated is generated.

  Next, the operation of the learning data generation function 21 will be described.

  FIG. 5 is an explanatory diagram showing an example of a data flow when the learning linearity correction vector 43 is generated by the learning data generation function 21.

  The learning data generating function 21 generates a learning linearity correction vector 43 based on the phantom image 41 and the phantom collimator ideal position information 42. Various methods are conventionally known as a method of generating a linearity correction vector based on a phantom image, and the learning data generation function 21 can use any of these methods.

  Next, the operation of the linearity correction vector generation function 23 will be described.

  The processing of the linearity correction vector generation function 23 may be performed using machine learning. In the following description, an example in which the linearity correction vector generation function 23 includes the neural network 231 and outputs the linearity correction vector based on the overall irradiation image using deep learning is shown.

  FIG. 6 is an explanatory diagram showing an example of a data flow during learning of the linearity correction vector generation function 23.

  The linearity correction vector generation function 23 receives a large amount of training data and performs learning to sequentially update the parameter data 232.

  The training data includes sets 511, 512, 513, ... Of the entire surface irradiation image (see FIG. 4B) and the PMT placement position information (see FIG. 4A) that form the training input data group 51. , And a teacher data group 61 composed of learning linearity correction vectors 611, 612, 613, ... Corresponding to this group.

  The learning linearity correction vectors 611, 612, 613, ... Are generated by the learning data generation function 21. For example, when the learning linearity correction vector 43 shown in FIG. 5 is used as the learning linearity correction vector 611, the entire surface irradiation image in the set 511 of the entire surface irradiation image and the PMT arrangement position information is obtained when the phantom image 41 is acquired. It is an image acquired by irradiating the entire surface of the scintillator with gamma rays from a distance without using a collimator for the Anger type detector used in the above, and the PMT placement position information is information of the Anger type detector.

  The linearity correction vector generation function 23 processes the set of the entire irradiation image and PMT placement position information 511, 512, 513, ... With the neural network 231 every time the training data is given, and the result is the learning linearity. The parameter data 232 is updated so as to approach the correction vectors 611, 612, 613 ,. Generally, when the rate of change of the parameter data 232 converges within the threshold value, the learning is determined to be completed. Hereinafter, the parameter data 232 after learning is particularly referred to as learned parameter data 232t.

  FIG. 7 is an explanatory diagram showing an example of a data flow during operation of the linearity correction vector generation function 23. In operation, the linearity correction vector generation function 23 uses the overall irradiation image 71 acquired by the Anger type detector for which information about linearity correction is to be generated and the PMT placement position information of the Anger type detector for which information about linearity correction is desired to be generated. 72 is input, and the linearity correction vector 81 of the Anger type detector for which the information about linearity correction is to be generated is output using the learned model 230.

  The neural network 231 and the learned parameter data 232t constitute the learned model 230. The neural network 231 is stored in the storage circuit 13 in the form of a program. The learned parameter data 232t may be stored in the storage circuit 13 of the medical device 10 or may be stored in a storage medium connected to the medical device 10 via the network 100. When the learned model 230 (neural network 231 and learned parameter data 232t) is stored in the storage circuit 13 of the medical device 10, linearity correction vector generation as an example of a processing unit realized by the processor of the processing circuit 15 is generated. The function 23 reads the learned model 230 from the storage circuit 13 of the medical device 10 and executes it to output the linearity correction vector 81 based on the overall irradiation image 71 and the PMT placement position information 72.

  Further, the learned model 230 may be constructed by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) and an FPGA (Field Programmable Gate Array).

  When the medical device 10 generates the information regarding the linearity correction of the Anger type detector, the medical device 10 only needs to acquire the entire irradiation image 71 of the Anger type detector, and does not need to attach or detach the phantom collimator. Since the entire surface irradiation image 71 is an image acquired by irradiating the entire surface of the scintillator with gamma rays from a distance, the collection efficiency of gamma rays is extremely high as compared with the phantom image, and the time required for image acquisition can be significantly shortened. it can.

  Further, at the time of collecting the image of the entire surface irradiation image 71, the collimator is not used, so that the gamma ray collection efficiency is high and a low-dose radiation source can be used, so that the risk of exposure of the operator can be greatly reduced. . The radiation source is installed at a predetermined position facing the Anger detector. As the radiation source, for example, a glass or plastic case having a diameter and length of several cm and containing RI solution such as Tc-99m of several cc can be used.

  Further, since the work of attaching and detaching the phantom collimator, which is a heavy object, is not necessary at all, it is possible to prevent the risk of damaging other members during the attaching and detaching work.

  FIG. 8A is an explanatory diagram showing an example of a data flow at the time of learning of the linearity correction vector generation function 23 when the information regarding the sensitivity and response function of the PMT is further used as input data, and FIG. 4 is an explanatory diagram showing an example of a data flow in FIG.

  When information about the sensitivity and response function of the PMT is used as input data, at the time of learning, a set 521, 522, 523 of a full-face irradiation image as training input data, PMT placement position information, and information about the sensitivity and response function of the PMT. , Are input to the neural network 231 as the training input data group 52 (see FIG. 8A). As the teacher data group 61, the same one as the example shown in FIG. 6 can be used.

  Further, during operation, the linearity correction vector generation function 23 desires to generate information regarding linearity correction for the learned model 230 read from the storage circuit 13, together with the overall irradiation image 71 and the PMT placement position information 72. By inputting the information 73 regarding the sensitivity and response function of the PMT of the die detector, the linearity correction vector 81 of the Anger detector is output (see FIG. 8B).

  The linearity correction vector generation function 23 further uses the sensitivity and response function of the PMT as input data, so that the estimation accuracy is higher than in the case where only the overall irradiation image 71 and the PMT placement position information 72 are input data. The linearity correction vector 81 can be output.

  By the way, since the scintillator has a predetermined area, it is difficult to make the radiation sensitivity uniform at all radiation incidence positions. The overall irradiation image contains the influence of the radiation incident position dependency of the sensitivity of the scintillator. In addition, the effect of the edge effect is also included in the overall irradiation image. The edge effect is a phenomenon in which linearity is blocked due to reflection of scintillation light by the end surface of the scintillator in the peripheral portion of the detector and loss of periodicity of the PMT structurally around the detector. (Refer to FIG. 2 (b) and FIG. 3 (b)).

  Therefore, the linearity correction vector used as the teacher data of the learning data is a linearity correction vector that considers at least one of the radiation incident position dependence of the sensitivity of the scintillator and the influence of the edge effect (hereinafter, the second learning straight line). (Referred to as the sex correction vector 43a).

  FIG. 9 is a data flow when the learning data generation function 21 generates the learning second linearity correction vector 43a in consideration of at least one of the radiation incident position dependency of the sensitivity of the scintillator and the influence of the edge effect. It is explanatory drawing which shows an example.

  In this case, the learning data generation function 21 is acquired by irradiating the entire surface of the scintillator with gamma rays from a distance, in addition to the phantom image 41 acquired using the phantom collimator and the ideal position information 42 of the phantom collimator. The linearity second correction vector is generated based on the entire irradiation image 511a. The overall irradiation image contains at least one of the effect of the radiation incident position dependency of the sensitivity of the scintillator and the edge effect. Therefore, the learning second linearity correction vector 43a generated using the entire irradiation image may take into consideration at least one of the radiation incident position dependency of the sensitivity of the scintillator and the edge effect. it can.

  When the learning second linearity correction vector 43a thus generated is used as teacher data during the learning shown in FIG. 6 or FIG. 8A, the linearity correction vector generation function 23 is used during operation of the learned model 230. Can output the linearity correction vector 81 considering the influence of at least one of the radiation incident position dependency of the sensitivity of the scintillator and the edge effect.

(Second embodiment)
FIG. 10 is a block diagram showing a configuration example of a SPECT device 301 including the medical device according to the second embodiment. Further, FIG. 11 is a schematic block diagram showing an example of a function implemented by the processor of the processing circuit 335 according to the second embodiment.

  The SPECT device 301 includes a scanner device 302 and a console 303 as an example of the medical device 10. The SPECT device 301 shown in the second embodiment is different from the medical device 10 shown in the first embodiment in that it is possible to use the overall irradiation image 71 collected by itself. Other configurations and operations are similar to those of the medical device 10 shown in FIG.

  FIG. 10 shows an example in which a two-detector gamma ray detector rotating type SPECT apparatus is used as the SPECT apparatus 301. The gamma ray detector rotating type SPECT apparatus may have one or three or more gamma ray detectors.

  The scanner device 302 includes gamma ray detectors 311 and 312, two collimators 313 and 314 detachably attached to the gamma ray detectors 311 and 312, a rotary mount 315, a fixed mount 316, and a rotary drive device 321. And a data collection circuit 322 and a bed device 323.

  The gamma ray detectors 311 and 312 are held on the rotary mount 315. The rotary gantry 315 rotates about a predetermined rotation axis r (around the z axis (body axis)) via the rotation drive device 321, so that the two gamma ray detectors 311 and 312 integrally rotate around the rotation axis r. Rotate. The rotary base 315 has a rotary plate rotatably supported by the fixed base 316. The gamma ray detectors 311 and 312 are held on the rotary plate of the rotary mount 315. The fixed base 316 is composed of a housing fixed to the base. The rotary mount 315 and the fixed mount 316 are covered with, for example, a mount cover. An opening that encloses an imaging region is provided in the central portion of the rotary mount 315, the fixed mount 316, and the mount cover that covers these.

  The gamma ray detector 311 is an Anger type detector, and detects gamma rays emitted from RI (radioactive isotope) such as Tl-201 or Tc-99m administered to the subject (eg, patient) P. Since the gamma ray detector 312 has the same configuration and operation as the gamma ray detector 311, the description thereof will be omitted.

  The gamma ray detector 311 detects a collimator 313 for defining an incident angle of the gamma ray, a scintillator that emits a momentary flash when the gamma ray collimated by the collimator 313 is incident, a light guide, and light emitted from the scintillator. It has a plurality of PMTs (photomultiplier tubes) arranged in two dimensions, a scintillator electronic circuit, and the like.

  The collimator 313 is removed when the whole-face irradiation image 71 is collected.

  The scintillator is composed of, for example, thallium-activated sodium iodide NaI (Tl). The electronic circuit for a scintillator, when an event (event) that a gamma ray is incident occurs, based on the outputs of the plurality of PMTs, the incident position information, the incident intensity information, and the incident intensity information of the gamma rays in the detection plane formed by the plurality of PMTs The time information is generated and output to the processing circuit 335 of the console 303. This position information may be information of two-dimensional coordinates in the detection surface, or the detection surface may be virtually divided into a plurality of divided areas (primary cells) in advance (for example, divided into 128 × 128 pieces). Alternatively, the information may be information indicating which primary cell has been incident.

  The rotation driving device 321 includes a rotating means such as a motor for rotating the rotating base 315 at a high speed around a predetermined rotation axis r, an electronic component for controlling the rotation of the rotating means, and a rotating base 315 for rotating the rotating means. It has a transmission means such as a roller for transmitting to. The rotation driving device 321 is controlled by the processing circuit 335 via the data collection circuit 322 to rotate the rotary mount 315 around a predetermined rotation axis r. For example, the processing circuit 335 collects SPECT projection data of the subject from a plurality of directions by rotating the gamma ray detectors 311 and 312 around the subject P continuously or stepwise via the rotary mount 315. You can

  The data acquisition circuit 322 is composed of, for example, a printed circuit board, and is controlled by the processing circuit 335 to control the gamma ray detectors 311 and 312, the rotation driving device 321, and the top driving device 325, thereby imaging the subject P. To execute.

  The data acquisition circuit 322 acquires the respective outputs of the gamma ray detectors 311 and 312 in the list mode, for example, and outputs the acquired projection data to the console 303. In the list mode, gamma ray detection position information, intensity information, information indicating the relative position between the gamma ray detectors 311 and 312 and the subject P (positions and angles of the gamma ray detectors 311 and 312, etc.), and gamma ray detection time are displayed. Collected for each gamma ray incident event.

  The bed device 323 includes a top plate 324 and a top plate drive device 325. The subject P is placed on the top plate 324. The top driving device 325 moves the top 324 under the control of the processing circuit 335 via the data collecting circuit 322. Specifically, the top driving device 325 has a motor as a drive source for moving the top 324 along the z-axis direction and the y-axis direction, an electronic component for controlling the motor, and the like. The top driving device 325 has a motor as a drive source for moving the top 324 up and down, an electronic component for controlling the motor, and the like.

  On the other hand, the console 303 as an example of the medical device 10 is configured by, for example, a general personal computer or a workstation, and has an input interface 331, a display 332, a storage circuit 333, a network connection circuit 334, and a processing circuit 335. Note that the console 303 does not have to be provided independently, and for example, a part of the components 331 to 335 of the console 303 may be provided dispersedly on the fixed base 316.

  The input interface 331, the display 332, the storage circuit 333, and the network connection circuit 334 have the same configuration and operation as the input interface 11, the display 12, the storage circuit 13, and the network connection circuit 14 of the medical device 10 according to the first embodiment. Therefore, the description is omitted.

  The processor of the processing circuit 335 of the console 303 as an example of the medical device 10 realizes an acquisition function 351 and a linearity correction vector generation function 352 as shown in FIG. Each of these functions is stored in the storage circuit 333 in the form of a program.

  Hereinafter, a case where the linearity correction vector of the gamma ray detector 311 among the gamma ray detectors 311 and 312 is generated will be described as an example.

  In this case, the acquisition function 351 acquires the overall irradiation image 71 acquired by the gamma ray detector 311 with the collimator 313 of the gamma ray detector 311 removed. The acquisition function 351 is an example of an acquisition unit.

  The linearity correction vector generation function 352 receives the overall irradiation image 71 and the PMT placement position information 72 of the gamma ray detector 311 and outputs the linearity correction vector 81 of the gamma ray detector 311 using the learned model 230. The linearity correction vector generation function 352 is an example of a processing unit.

  With the SPECT apparatus 301 according to the third embodiment, as with the medical apparatus 10 according to the first embodiment, there is no need to collect phantom images again when generating the linearity correction vector, and the full-face irradiation image 71 and the PMT arrangement position are arranged. The linearity correction vector 81 can be output by inputting the information 72 and the learned model 230.

  According to at least one embodiment described above, it is possible to acquire information regarding linearity correction based on a nuclear medicine image acquired without using a phantom collimator.

  In the above-described embodiment, the word “processor” means, for example, a dedicated or general-purpose CPU (Central Processing Unit), GPU (Graphics Processing Unit), or an application-specific integrated circuit (ASIC), A circuit such as a programmable logic device (for example, a Simple Programmable Logic Device (SPLD), a Complex Programmable Logic Device (CPLD), and an FPGA) is meant. The processor realizes various functions by reading and executing a program stored in a storage medium.

  Further, in the above embodiment, an example in which a single processor of the processing circuit realizes each function has been described. However, a processing circuit is configured by combining a plurality of independent processors, and each processor realizes each function. Good. When a plurality of processors are provided, the storage medium for storing the program may be provided individually for each processor, or one storage medium may collectively store the programs corresponding to the functions of all the processors. Good.

  Although some embodiments of the present invention have been described, these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, replacements, and changes can be made without departing from the spirit of the invention. These embodiments and their modifications are included in the invention described in the claims and the equivalents thereof as well as included in the scope and the gist of the invention.

10 Medical Device 15 Processing Circuit 21 Learning Data Generation Function 22 Acquisition Function 23 Linearity Correction Vector Generation Function 31 Horizontal Slit Phantom Collimator 32 Horizontal Slit Phantom Image 33 Vertical Slit Phantom Collimator 34 Vertical Slit Phantom Image 36 Full Illumination Image 41 Phantom Image 42 Phantom ideal position information 43 Learning linearity correction vector 43a Learning second linearity correction vector 61 Teacher data group 71 Whole irradiation image 72 PMT placement position information 73 PMT sensitivity and response function information 81 Linearity correction vector 101 Nuclear medicine Diagnostic device 230 Trained model 301 SPECT device 303 Console 311 and 312 Gamma ray detector 335 Processing circuit 351 Acquisition function 352 Linearity correction vector generation function 511a Whole surface irradiation image

Claims (7)

An acquisition unit that acquires a first image that is acquired without using a collimator in the nuclear medicine diagnostic apparatus and information about the arrangement of the optical sensor;
Inputting the first image and the information about the arrangement of the photosensors to a learned model that outputs information about the linearity correction based on the first image and the information about the arrangement of the photosensors. And a processing unit that outputs information related to the linearity correction,
Medical device equipped with. The trained model is
Outputting information about the linearity correction further based on information about the sensitivity and response function of the optical sensor,
The acquisition unit is
Further obtaining information about the sensitivity and response function of the light sensor,
The processing unit is
By inputting the first image, the information about the arrangement of the photosensors, and the information about the sensitivity and response function of the photosensors to the learned model, the information about the linearity correction is output.
The medical device according to claim 1. The trained model is
Phantom images acquired by using a phantom collimator in the nuclear medicine diagnosis device, ideal position information of the phantom collimator, and learning based on the linearity correction information generated based on the teacher data,
The medical device according to claim 1 or 2. The information regarding the linearity correction output by the learned model is
It is information considering at least one of the radiation incident position dependency of the sensitivity of the phosphor provided on the front surface of the optical sensor and the edge effect,
The medical device according to any one of claims 1 to 3. The trained model is
A phantom image acquired by using a phantom collimator in the nuclear medicine diagnostic device, ideal position information of the phantom collimator, and an image acquired without using the collimator in the nuclear medicine diagnostic device, and generated based on , Learning is performed by using, as teacher data, information relating to linearity correction in which at least one of the dependency of the sensitivity of the phosphor on the radiation incident position and the edge effect is taken into consideration.
The medical device according to claim 4. A method for acquiring learning data of a learned model that outputs information related to linearity correction based on an image acquired without using a collimator in a nuclear medicine diagnostic device and information related to the arrangement of optical sensors,
Phantom image acquired by using a phantom collimator in the nuclear medicine diagnostic device, ideal position information of the phantom collimator, and information regarding linearity correction generated based on the step of acquiring as training data of the learning data When,
An image acquired without using a collimator in the nuclear medicine diagnostic device, and a step of acquiring information regarding the arrangement of optical sensors as input data of the learning data,
A method for acquiring learning data having. The step of acquiring the information regarding the linearity correction as the teacher data,
The phantom image, ideal position information of the phantom collimator, an image obtained without using a collimator in the nuclear medicine diagnostic device, and generated based on, of the phosphor provided in front of the optical sensor A step of acquiring, as the teacher data, information relating to linearity correction in consideration of at least one of the radiation incident position dependency of sensitivity and the edge effect,
The method for acquiring learning data according to claim 6.