JP7362322B2 - X-ray CT system and medical processing equipment - Google Patents

Description

Embodiments of the present invention relate to an X-ray CT system and a medical processing device.

CT image data collected by an X-ray CT (Computed Tomography) scanner may contain noise and image artifacts due to various factors. Specifically, projection data may be missing or contain noise due to factors such as body movements of the subject, fluctuations in the output from the X-ray tube, and attenuation of X-rays depending on the body thickness of the subject. The reconstructed CT image data may contain noise and image artifacts.

In order to improve the quality of CT image data, it has been proposed to correct projection data sets used for reconstruction processing and reconstructed CT image data using AI (Artificial Intelligence). Here, in order to appropriately perform the correction by AI, it is preferable that learning is performed based on more information.

Japanese Patent Application Publication No. 2012-130667 International Publication No. 2017/141958 JP 2017-86903 Publication

The problem to be solved by the present invention is to generate high quality CT image data.

The X-ray CT system of the embodiment includes a scanning section and a processing section. The scanning unit executes a first scan to collect a first projection data set by irradiating a first range along the body axis direction of the subject, and after the first scan, A second scan is performed in which a second projection data set is acquired by irradiating a second range along the axial direction that is narrower than the first range. The processing unit includes at least a portion of the first projection data set or first image data based on at least a portion of the first projection data set, and a second projection data set or second image data based on the second projection data set. the first projection data for a trained model that generates a third projection data set or third image data having a higher quality than the second projection data set or the second image data based on the image data; The third projection data set or the third image data is generated by inputting the set or the first image data and the second projection data set or the second image data.

FIG. 1 is a block diagram showing an example of the configuration of an X-ray CT system according to the first embodiment. FIG. 2 is a diagram for explaining the trained model generation process according to the first embodiment. FIG. 3 is a diagram illustrating an example of use of the learned model according to the first embodiment. FIG. 4 is a flowchart for explaining a series of processing steps of the X-ray CT system according to the first embodiment. FIG. 5 is a block diagram showing an example of the configuration of a medical information processing system according to the third embodiment.

Hereinafter, embodiments of an X-ray CT system and a medical processing device will be described in detail with reference to the accompanying drawings. Note that the X-ray CT system and medical processing apparatus according to the present application are not limited to the embodiments described below.

(First embodiment)
First, the configuration of an X-ray CT system 10 according to a first embodiment will be described with reference to FIG. FIG. 1 is a block diagram showing an example of the configuration of an X-ray CT system 10 according to the first embodiment. As shown in FIG. 1, the X-ray CT system 10 includes a gantry device 110, a bed device 130, and a console device 140. Note that the X-ray CT system 10 is also called an X-ray CT device or an X-ray CT scanner.

In FIG. 1, the rotation axis of the rotation frame 113 in a non-tilted state or the longitudinal direction of the top plate 133 of the bed device 130 is defined as the Z-axis direction. Further, the axial direction that is orthogonal to the Z-axis direction and horizontal to the floor surface is defined as the X-axis direction. Further, the axial direction that is orthogonal to the Z-axis direction and perpendicular to the floor surface is defined as the Y-axis direction. Note that FIG. 1 depicts the gantry device 110 from multiple directions for explanation, and shows a case where the X-ray CT system 10 has one gantry device 110.

The gantry device 110 includes an X-ray tube 111, an X-ray detector 112, a rotating frame 113, an X-ray high voltage device 114, a control device 115, a wedge 116, a collimator 117, and a DAS 118.

The X-ray tube 111 is a vacuum tube that has a cathode (filament) that generates thermoelectrons and an anode (target) that generates X-rays upon collision with the thermoelectrons. The X-ray tube 111 generates X-rays to irradiate the subject P1 by irradiating thermoelectrons from the cathode to the anode by applying a high voltage from the X-ray high voltage device 114.

The X-ray detector 112 detects the X-rays emitted from the X-ray tube 111 and passed through the subject P1, and outputs a signal corresponding to the detected X-ray dose to the DAS 118. The X-ray detector 112 has, for example, a plurality of detection element rows in which a plurality of detection elements are arranged in the channel direction along one circular arc centered on the focal point of the X-ray tube 111. The X-ray detector 112 has, for example, a structure in which a plurality of detection element rows in which a plurality of detection elements are arranged in a channel direction are arranged in a column direction (slice direction, row direction).

For example, the X-ray detector 112 is an indirect conversion type detector that includes a grid, a scintillator array, and a photosensor array. A scintillator array has multiple scintillators. The scintillator has a scintillator crystal that outputs light with an amount of photons corresponding to the amount of incident X-rays. The grid is disposed on the X-ray incident side of the scintillator array and has an X-ray shielding plate that absorbs scattered X-rays. Note that the grid is sometimes called a collimator (one-dimensional collimator or two-dimensional collimator). The optical sensor array has a function of converting into an electrical signal according to the amount of light from the scintillator, and includes optical sensors such as photodiodes, for example. Note that the X-ray detector 112 may be a direct conversion type detector that includes a semiconductor element that converts incident X-rays into electrical signals.

The rotating frame 113 is an annular frame that supports the X-ray tube 111 and the X-ray detector 112 facing each other, and allows the X-ray tube 111 and the X-ray detector 112 to be rotated by the control device 115. For example, the rotating frame 113 is cast from aluminum. Note that, in addition to the X-ray tube 111 and the X-ray detector 112, the rotating frame 113 can also support an X-ray high voltage device 114, a wedge 116, a collimator 117, a DAS 118, and the like. Furthermore, the rotating frame 113 may further support various configurations not shown in FIG. Below, in the gantry device 110, the rotating frame 113 and a portion that rotates and moves together with the rotating frame 113 will also be referred to as a rotating section.

The X-ray high-voltage device 114 includes an electric circuit such as a transformer and a rectifier, and includes a high-voltage generator that generates a high voltage to be applied to the X-ray tube 111 and a high-voltage generator that generates the and an X-ray control device that controls the output voltage according to the output voltage. The high voltage generator may be of a transformer type or an inverter type. Note that the X-ray high voltage device 114 may be provided on the rotating frame 113 or may be provided on a fixed frame (not shown).

The control device 115 includes a processing circuit including a CPU (Central Processing Unit), and a drive mechanism such as a motor and an actuator. The control device 115 receives input signals from the input interface 143 and controls the operations of the gantry device 110 and the bed device 130. For example, the control device 115 controls the rotation of the rotating frame 113, the tilt of the gantry device 110, the operation of the bed device 130, and the like. For example, as a control for tilting the gantry device 110, the control device 115 rotates the rotation frame 113 about an axis parallel to the X-axis direction based on input inclination angle (tilt angle) information. Note that the control device 115 may be provided on the gantry device 110 or may be provided on the console device 140.

The wedge 116 is an X-ray filter for adjusting the amount of X-rays irradiated from the X-ray tube 111. Specifically, the wedge 116 is an X-ray filter that attenuates the X-rays emitted from the X-ray tube 111 so that the X-rays emitted from the X-ray tube 111 to the subject P1 have a predetermined distribution. It is. For example, the wedge 116 is a wedge filter or a bow-tie filter, and is manufactured by processing aluminum or the like to have a predetermined target angle and a predetermined thickness.

The collimator 117 is a lead plate or the like for narrowing down the irradiation range of the X-rays transmitted through the wedge 116, and forms a slit by combining a plurality of lead plates or the like. Note that the collimator 117 is sometimes called an X-ray diaphragm. Furthermore, although FIG. 1 shows a case where the wedge 116 is arranged between the X-ray tube 111 and the collimator 117, it is also the case where the collimator 117 is arranged between the X-ray tube 111 and the wedge 116. Good too. In this case, the wedge 116 transmits and attenuates the X-rays irradiated from the X-ray tube 111 and whose irradiation range is limited by the collimator 117.

DAS 118 collects X-ray signals detected by each detection element included in X-ray detector 112. For example, the DAS 118 includes an amplifier that performs amplification processing on the electrical signal output from each detection element and an A/D converter that converts the electrical signal into a digital signal, and generates detection data. DAS 118 is realized by a processor, for example.

The data generated by the DAS 118 is transmitted from a transmitter having a light emitting diode (LED) provided in the rotating frame 113 to a non-rotating portion of the gantry device 110 (for example, a fixed frame, etc. in FIG. 1) by optical communication. (not shown), which has a photodiode, and is transferred to the console device 140. Here, the non-rotating portion is, for example, a fixed frame that rotatably supports the rotating frame 113. Note that the method of transmitting data from the rotating frame 113 to the non-rotating part of the gantry device 110 is not limited to optical communication, but any non-contact data transmission method may be used, or a contact data transmission method may be used. I don't mind if you hire me.

The bed device 130 is a device on which the subject P1 to be scanned is placed and moved, and includes a base 131, a bed driving device 132, a top plate 133, and a support frame 134. The base 131 is a casing that supports the support frame 134 so as to be movable in the vertical direction. The bed driving device 132 is a drive mechanism that moves the top plate 133 on which the subject P1 is placed in the longitudinal direction of the top plate 133, and includes a motor, an actuator, and the like. The top plate 133 provided on the upper surface of the support frame 134 is a plate on which the subject P1 is placed. In addition to the top plate 133, the bed driving device 132 may move the support frame 134 in the longitudinal direction of the top plate 133.

Console device 140 has memory 141 , display 142 , input interface 143 , and processing circuit 144 . Note that although the console device 140 will be described as being separate from the gantry device 110, the gantry device 110 may include the console device 140 or a part of each component of the console device 140.

The memory 141 is realized by, for example, a RAM (Random Access Memory), a semiconductor memory element such as a flash memory, a hard disk, an optical disk, or the like. For example, the memory 141 stores various data collected by performing a scan on the subject P1. Further, for example, the memory 141 stores programs for the circuits included in the X-ray CT system 10 to realize their functions. Note that the memory 141 may be realized by a server group (cloud) connected to the X-ray CT system 10 via a network.

Display 142 displays various information. For example, the display 142 displays a CT image for display generated by the processing circuit 144, a GUI (Graphical User Interface), etc. for accepting various operations from the user. For example, the display 142 is a liquid crystal display or a CRT (Cathode Ray Tube) display. The display 142 may be of a desktop type, or may be composed of a tablet terminal or the like that can communicate wirelessly with the main body of the console device 140.

The input interface 143 accepts various input operations from the user, converts the received input operations into electrical signals, and outputs the electrical signals to the processing circuit 144 . For example, the input interface 143 receives from the user reconstruction conditions when reconstructing CT image data, image processing conditions when generating a CT image for display from CT image data, and the like. For example, the input interface 143 may include a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touchpad that performs input operations by touching the operation surface, a touchscreen that integrates a display screen and a touchpad, and an optical sensor. This is realized by using a non-contact input circuit, a voice input circuit, etc. Note that the input interface 143 may be provided in the gantry device 110. Furthermore, the input interface 143 may be configured with a tablet terminal or the like that can communicate wirelessly with the main body of the console device 140. Further, the input interface 143 is not limited to one that includes physical operation parts such as a mouse and a keyboard. For example, an electric signal processing circuit that receives an electric signal corresponding to an input operation from an external input device provided separately from the console device 140 and outputs this electric signal to the processing circuit 144 is also an example of the input interface 143. included.

The processing circuit 144 controls the overall operation of the X-ray CT system 10 by executing a scanning function 144a, a processing function 144b, and a control function 144c. Note that the scan function 144a is an example of a scan unit. Further, the processing function 144b is an example of a processing unit.

For example, the processing circuit 144 reads a program corresponding to the scan function 144a from the memory 141 and executes it, thereby executing a scan on the subject P1. For example, the scan function 144a executes various scans such as positioning imaging and main scan on the subject P1.

Here, positioning imaging may be performed in two dimensions or three dimensions. In this embodiment, as an example, a case will be described in which two-dimensional positioning imaging is performed. In this case, the scan function 144a does not rotate the position of the X-ray tube 111 around the subject P1, and moves the position of the X-ray tube 111 and/or the subject P1 in the body axis direction of the subject P1 (Fig. Positioning photographing is performed while moving along the Z-axis direction shown in FIG.

For example, the scan function 144a fixes the position of the X-ray tube 111 at a predetermined rotation angle, moves the top plate 133 in the Z-axis direction, and causes the X-ray tube 111 to irradiate the subject P1 with X-rays. . Further, while positioning imaging is executed by the scan function 144a, the DAS 118 collects X-ray signals from each detection element in the X-ray detector 112, and generates detection data. Further, the scan function 144a performs preprocessing on the detection data output from the DAS 118. For example, the scan function 144a performs preprocessing such as logarithmic conversion processing, offset correction processing, inter-channel sensitivity correction processing, and beam hardening correction on the detection data output from the DAS 118. Note that data after preprocessing is also referred to as raw data. In addition, detection data before preprocessing and raw data after preprocessing are collectively referred to as projection data.

That is, the scan function 144a fixes the position of the X-ray tube 111 at a predetermined rotation angle, moves the top plate 133 in the Z-axis direction, and causes the X-ray tube 111 to irradiate the subject P1 with X-rays. In this way, projection data is collected for each of a plurality of positions in the body axis direction of the subject P1. Below, a plurality of projection data are collectively referred to as a projection data set. That is, the scan function 144a collects a projection data set by performing positioning imaging.

Furthermore, the processing circuit 144 generates image data based on the scan result by reading out and executing a program corresponding to the processing function 144b from the memory 141. For example, the processing function 144b generates positioning image data based on a projection data set collected by positioning photography. Note that the positioning image data is sometimes called scano image data or scout image data.

Further, the main scan may be executed using, for example, a conventional scan method, a helical scan method, or a step-and-shoot method.

When performing the main scan using the conventional scan method, the scan function 144a changes the position of the X-ray tube 111 around the subject P1 without moving the position of the X-ray tube 111 and the subject P1 along the body axis direction. Execute the main scan while rotating with . For example, the scan function 144a causes the X-ray tube 111 to irradiate the subject P1 with X-rays while rotating the X-ray tube 111 around the subject P1 with the top plate 133 stopped.

Furthermore, when performing the main scan using a helical scan method, the scan function 144a moves the position of the X-ray tube 111 and the subject P1 along the body axis direction, and also moves the position of the X-ray tube 111 to the subject P1. Execute the main scan while rotating around the . For example, the scan function 144a moves the top plate 133 in the Z-axis direction, rotates the X-ray tube 111 around the subject P1, and irradiates the subject P1 with X-rays from the X-ray tube 111. .

When performing the main scan using the step-and-shoot method, the scan function 144a first changes the position of the X-ray tube 111 to the subject P1 without moving the position of the X-ray tube 111 and the subject P1 along the body axis direction. Execute the main scan while rotating around the . For example, the scan function 144a causes the X-ray tube 111 to irradiate the subject P1 with X-rays while rotating the X-ray tube 111 around the subject P1 with the top plate 133 stopped. Next, the scan function 144a moves the top plate 133 in the Z-axis direction while stopping the irradiation of X-rays from the X-ray tube 111. Then, the scan function 144a causes the X-ray tube 111 to irradiate the subject P1 with X-rays again while rotating the X-ray tube 111 around the subject P1 with the top plate 133 stopped.

While the main scan is executed by the scan function 144a, the DAS 118 collects X-ray signals from each detection element in the X-ray detector 112 and generates detection data. Further, the scan function 144a performs preprocessing such as logarithmic conversion processing, offset correction processing, inter-channel sensitivity correction processing, and beam hardening correction on the detection data output from the DAS 118.

That is, the scan function 144a collects projection data for each of a plurality of irradiation directions (views) by irradiating the subject P1 with X-rays while rotating the position of the X-ray tube 111 around the subject P1. . That is, the scan function 144a collects a projection data set by executing the main scan.

Furthermore, the processing function 144b generates CT image data (volume data) based on the projection data set collected by the main scan. For example, the processing function 144b generates CT image data by performing reconstruction processing using a filtered back projection method, a successive approximation reconstruction method, a successive approximation applied reconstruction method, etc. based on the projection data set. . Furthermore, the processing function 144b can also generate CT image data by performing reconstruction processing using AI. For example, the processing function 144b generates CT image data using a DLR (Deep Learning Reconstruction) method.

Here, the processing function 144b performs correction on the CT image data. Specifically, the processing function 144b corrects the CT image data by inputting the positioning image data collected through positioning imaging and the CT image data collected through the main scan to the trained model M1. . Note that correction using the learned model M1 will be described later.

Furthermore, the processing circuit 144 controls the display on the display 142 by reading a program corresponding to the control function 144c from the memory 141 and executing it. For example, the control function 144c converts the positioning image data and CT image data generated by the processing function 144b into a display image using a known method. For example, the control function 144c converts CT image data into tomographic image data of an arbitrary cross section, three-dimensional image data, etc. based on input operations received from the user via the input interface 143. The control function 144c then causes the display 142 to display the converted display image. The control function 144c also transmits various data via the network. For example, the control function 144c transmits the positioning image data and CT image data generated by the processing function 144b to an image storage device (not shown) and stores them.

In the X-ray CT system 10 shown in FIG. 1, each processing function is stored in the memory 141 in the form of a computer-executable program. The processing circuit 144 is a processor that reads programs from the memory 141 and executes them to implement functions corresponding to each program. In other words, the processing circuit 144 in a state where the program has been read has a function corresponding to the read program.

Although the scanning function 144a, the processing function 144b, and the control function 144c are realized in the single processing circuit 144 in FIG. 1, the processing circuit 144 may be configured by combining a plurality of independent processors. , functions may be realized by each processor executing a program. Further, each processing function of the processing circuit 144 may be appropriately distributed or integrated into a single processing circuit or a plurality of processing circuits.

Furthermore, the processing circuit 144 may implement its functions using a processor of an external device connected via a network. For example, the processing circuit 144 reads and executes a program corresponding to each function from the memory 141, and uses a server group (cloud) connected to the X-ray CT system 10 via a network as a computational resource. Each function shown in FIG. 1 is realized.

The configuration example of the X-ray CT system 10 has been described above. Under such a configuration, the processing circuit 144 in the X-ray CT system 10 improves the quality of CT image data through processing described in detail below.

First, the processing function 144b generates a learned model M1 and stores it in the memory 141. That is, the processing function 144b performs a process of generating the learned model M1 prior to positioning and photographing the subject P1 and performing the main scan. Hereinafter, the generation process of the learned model M1 by the processing function 144b will be explained using FIG. 2. FIG. 2 is a diagram for explaining the generation process of the trained model M1 according to the first embodiment.

Image data C11, image data C12, and image data C13 shown in FIG. 2 are image data collected by scanning the subject P1, a subject P2 different from the subject P1, a phantom imitating a human body, and the like. The image data C11, the image data C12, and the image data C13 may be image data collected in the X-ray CT system 10, or may be image data collected in another system. Note that, as an example, the following description assumes that the image data C11, the image data C12, and the image data C13 are collected by scanning the subject P2 in the X-ray CT system 10.

For example, the X-ray CT system 10 collects image data C11 by performing a scan A11 on the subject P2. For example, first, the scan function 144a fixes the position of the X-ray tube 111 at a predetermined rotation angle, moves the top plate 133 in the Z-axis direction, and scans the subject P2 from the X-ray tube 111. By irradiating X-rays, projection data is collected for each of a plurality of positions in the body axis direction of the subject P2. Hereinafter, the plurality of projection data collected by scan A11 will be referred to as projection data set B11. The processing function 144b then generates image data C11 shown in FIG. 2 based on the collected projection data set B11. That is, the image data C11 is two-dimensional image data collected by two-dimensional scanning. To give an example, the image data C11 is positioning image data collected by positioning imaging during the examination of the subject P2.

Further, for example, the X-ray CT system 10 collects image data C12 by performing a scan A12 on the subject P2. For example, first, the scan function 144a rotates the X-ray tube 111 around the subject P2 and irradiates the subject P2 with X-rays at a dose of X1, thereby generating projection data for each of a plurality of views. Collect. Hereinafter, the plurality of projection data collected by scan A12 will be referred to as projection data set B12. Then, the processing function 144b generates image data C12 shown in FIG. 2 by executing a reconstruction process based on the collected projection data set B12. That is, the image data C12 is three-dimensional image data collected by three-dimensional scanning. For example, the image data C12 is CT image data collected through a main scan in the examination of the subject P2.

Further, for example, the X-ray CT system 10 collects image data C13 by performing a scan A13 on the subject P2. For example, first, the scan function 144a rotates the X-ray tube 111 around the subject P2 and irradiates the subject P2 with X-rays at a dose of X2, thereby generating projection data for each of a plurality of views. Collect. Here, the dose X2 is higher than the dose X1. Hereinafter, the plurality of projection data collected by scan A13 will be referred to as projection data set B13. Then, the processing function 144b generates image data C13 shown in FIG. 2 by executing a reconstruction process based on the collected projection data set B13. That is, the image data C13 is three-dimensional image data collected by three-dimensional scanning. For example, the image data C13 is CT image data collected by the main scan in the examination of the subject P2.

Furthermore, the image data C13 is image data collected using a higher dose of X-rays than the image data C12. That is, the image data C13 is high-quality image data collected using high-dose X-rays. For example, in image data C13, artifacts and noise are reduced compared to image data C12.

The processing function 144b generates the learned model M1 by executing machine learning using the image data C11 and image data C12 shown in FIG. 2 as input data and the image data C13 as output data. Here, the learned model M1 can be configured by, for example, a neural network.

A neural network is a network that has a structure in which adjacent layers arranged in a layered manner are connected, and information is propagated from the input layer side to the output layer side. For example, the processing function 144b generates the trained model M1 by performing deep learning on a multilayer neural network using the image data C11, image data C12, and image data C13. Note that a multilayer neural network includes, for example, an input layer, a plurality of intermediate layers (hidden layers), and an output layer.

For example, the processing function 144b inputs the image data C11 and the image data C12 to the neural network as input data. Here, in the neural network, information is propagated in one direction from the input layer side to the output layer side while connecting only between adjacent layers, and the output layer outputs image data obtained by estimating the image data C13. . That is, in the neural network, processing is performed to improve the quality of the image data C12, and the output layer outputs image data in which image data C13, which is higher in quality than the image data C12, is estimated. Note that a neural network in which information propagates in one direction from the input layer side to the output layer side is also called a convolutional neural network (CNN).

The processing function 144b generates the learned model M1 by adjusting the parameters of the neural network so that the neural network can output a preferable result when input data is input. For example, the processing function 144b adjusts the parameters of the neural network using a function (error function) representing the closeness between image data.

For example, the processing function 144b calculates an error function E1 indicating the closeness between the image data C13 and the image data estimated by the neural network. Furthermore, the processing function 144b calculates an error function E2 that indicates the consistency between the image data C11 and the image data estimated by the neural network. For example, the processing function 144b generates two-dimensional image data C14 by forward projecting the image data estimated by the neural network according to the X-ray irradiation direction when the image data C11 was collected. Furthermore, the processing function 144b calculates an error function E2 indicating the closeness between the generated image data C14 and image data C11.

Furthermore, the processing function 144b adjusts the parameters of the neural network so that the calculated error function becomes minimum. For example, the processing function 144b adjusts the parameters of the neural network so that the sum of the error function E1 and the error function E2 becomes minimum. Thereby, the processing function 144b receives input of two-dimensional image data and three-dimensional image data, and generates a trained model M1 equipped with a function to improve the quality of the input three-dimensional image data. do. Furthermore, the processing function 144b stores the generated trained model M1 in the memory 141.

Then, when the scan for the subject P1 is executed, the processing function 144b reads out the learned model M1 from the memory 141, and uses the read out learned model M1 to improve the quality of the CT image data collected from the subject P1. improve. The scan performed on the subject P1 and the correction process of CT image data using the learned model M1 will be described below with reference to FIG. FIG. 3 is a diagram showing an example of how the trained model M1 according to the first embodiment is used.

First, the scan function 144a executes scan A21. Specifically, as shown in FIG. 3, the scan function 144a changes the position of the X-ray tube 111 relative to the subject P1 without rotating the position of the X-ray tube 111 around the subject P1. While moving, the subject P1 is irradiated with X-rays. Thereby, the scan function 144a irradiates the range R1 along the body axis direction of the subject P1 with X-rays and collects the projection data set B21. For example, the scan function 144a fixes the position of the X-ray tube 111 at a predetermined rotation angle, moves the top plate 133 in the Z-axis direction, and collects the projection data set B21. Furthermore, the processing function 144b generates two-dimensional image data C21 based on the projection data set B21 collected by the scan A21.

Note that scan A21 is an example of the first scan. Furthermore, range R1 is an example of a first range. Furthermore, the projection data set B21 is an example of the first projection data set. Further, the image data C21 is an example of first image data. The processing function 144b may generate the image data C21 using the entire projection data set B21, or may generate the image data C21 using a part of the projection data set B21.

Next, the control function 144c causes the display 142 to display a reference image based on the image data C21. Further, the control function 144c sets a range R2, which is the scan range of the scan A22, by accepting an input operation from a user who has referred to the reference image. Note that scan A22 is an example of the second scan. Furthermore, range R2 is an example of a second range.

As described above, the range R2 is generated based on at least a portion of the projection data set B21 and is set by operating the positioning image data displayed on the display 142. That is, scan A21 is positioning photographing for setting range R2. Therefore, it is preferable that the range R1 of the scan A21 is set to be relatively wide so as to include the organ to be diagnosed. Since range R2 is set within range R1, it is usually narrower than range R1, as shown in FIG.

Next, the scan function 144a performs a scan A22 on the range R2 set as the reference image. Specifically, as shown in FIG. 3, the scan function 144a irradiates the subject P1 with X-rays while rotating the position of the X-ray tube 111 around the subject P1. Thereby, the scan function 144a irradiates the range R2 along the body axis direction of the subject P1 with X-rays and collects the projection data set B22. For example, the scan function 144a collects the projection data set B22 by performing a conventional scan, a helical scan, or a step-and-shoot scan. Note that the projection data set B22 is an example of the second projection data set.

Furthermore, the processing function 144b generates three-dimensional image data C22 based on the projection data set B22 collected by scan A22. Note that the image data C22 is an example of second image data. For example, the processing function 144b performs reconstruction processing such as a filtered back projection method, a successive approximation reconstruction method, a successive approximation applied reconstruction method, and a DLR method based on the projection data set B22, thereby creating a three-dimensional image. Data C22 is reconstructed. That is, scan A22 is a main scan for collecting CT image data (volume data).

Note that the scan function 144a may execute the scan A21 and the scan A22 in synchronization according to the heartbeat or breathing cycle of the subject P1. That is, the scan function 144a may execute scan A21 and scan A22 in synchronization so that positional deviation between scan A21 and scan A22 does not occur due to periodic movements due to heartbeat or respiration.

For example, the scan function 144a acquires the electrocardiographic waveform of the subject P1 in parallel with scan A21 and scan A22. For example, the scan function 144a acquires an electrocardiographic waveform of the subject P1 using an electrocardiograph attached to the subject P1. Then, the scan function 144a executes scan A22 such that the phase of the electrocardiographic waveform in scan A22 matches the phase of the electrocardiographic waveform in scan A21.

Further, for example, the scan function 144a acquires the respiratory waveform of the subject P1 in parallel with the scan A21 and the scan A22. For example, the scan function 144a acquires the respiratory waveform of the subject P1 using a respiratory sensor. For example, the scan function 144a acquires a respiratory waveform using a laser generator and a light receiver as a respiratory sensor. Specifically, the scan function 144a processes the reflected light signal from the abdominal surface of the subject P1, and uses the laser generator and the A respiratory waveform is obtained by repeatedly calculating the distance to the subject's abdominal surface in real time. Note that the example of the respiratory sensor is not limited to this, and the scan function 144a may detect the respiratory waveform using, for example, a pressure sensor attached to the abdomen of the subject P1 or an optical camera that photographs the subject P1. It is okay to obtain it. Then, the scan function 144a executes scan A22 such that the phase of the respiratory waveform in scan A22 matches the phase of the respiratory waveform in scan A21.

Furthermore, the processing function 144b may align the projection data set B21 collected by scan A21 and the projection data set B22 collected by scan A22. Alternatively, the processing function 144b may align the image data C21 collected by scan A21 and the image data C22 collected by scan A22. Thereby, the processing function 144b can reduce the influence of the body movement of the subject P1 that occurs between the scan A21 and the scan A22.

Next, the processing function 144b evaluates the quality of the image data C22. For example, the processing function 144b evaluates the quality of the image data C22 by calculating a signal-to-noise ratio for the image data C22 and comparing the signal-to-noise ratio with a threshold value. Further, for example, the control function 144c generates a display image based on the image data C22, and causes the display 142 to display the generated image. Then, the processing function 144b receives an input of evaluation regarding the quality of the image data C22 from the user who has referred to the display image.

Next, the processing function 144b determines whether or not to perform correction processing on the image data C22, depending on the evaluation result of the quality of the image data C22. That is, if the quality of the image data C22 is sufficiently high, the processing function 144b may omit the correction processing described later and terminate the processing. Note that below, a case will be described in which correction processing is performed on the image data C22.

When performing correction processing on the image data C22, the processing function 144b inputs the image data C21 and the image data C22 to the learned model M1 read from the memory 141. As a result, the trained model M1 outputs image data C23 having higher quality than the image data C22, as shown in FIG. That is, the learned model M1 generates the image data C23 shown in FIG. 3 by correcting the image data C22 while maintaining consistency with the image data C21. Note that the image data C23 is an example of third image data.

Note that the processing function 144b may input all or only a part of the image data C21 to the learned model M1. For example, the processing function 144b may input only the portion included in the range R2 of the image data C21 to the trained model M1. In other words, the processing function 144b inputs the image data C21 based on at least a portion of the projection data set B21 collected from the range R1 into the trained model M1.

The control function 144c then causes the display 142 to display a display image based on the image data C23. For example, the control function 144c first performs rendering processing on the image data C23. Here, an example of the rendering process is a process of generating a two-dimensional image of an arbitrary cross section from the image data C23 using a cross-sectional reconstruction method (MPR: Multi Planar Reconstruction). Further, as another example of rendering processing, a two-dimensional image reflecting three-dimensional information is generated from the image data C23 by volume rendering processing or maximum intensity projection (MIP). An example of this is the process of The control function 144c then causes the display 142 to display the display image generated by the rendering process.

Next, an example of a processing procedure by the X-ray CT system 10 will be described using FIG. 4. FIG. 4 is a flowchart for explaining a series of processing steps of the X-ray CT system 10 according to the first embodiment.

Step S101 and step S106 correspond to the scan function 144a. Step S102, step S107, step S108, step S109, step S110, and step S111 correspond to the processing function 144b. Step S103, step S104, step S105, and step S112 correspond to the control function 144c.

First, the processing circuit 144 executes a scan A21 on the subject P1 and collects a projection data set B21 (step S101). Next, the processing circuit 144 generates image data C21 based on the projection data set B21 (step S102). Next, the processing circuit 144 generates a reference image by performing rendering processing on the image data C21 (step S103), and displays the generated reference image on the display 142 (step S104).

Here, the processing circuit 144 determines whether or not a scan range has been set by the user who referred to the reference image (step S105), and if the scan range has not been set, the processing circuit 144 enters a standby state (step S105 negative). . On the other hand, if the scan range has been set (Yes in step S105), the processing circuit 144 executes scan A22 on the scan range set in the reference image and collects projection data set B22 (step S106). Next, the processing circuit 144 generates image data C22 based on the projection data set B22 (step S107).

Next, the processing circuit 144 evaluates the quality of the image data C22 (step S108), and determines whether or not to perform correction processing on the image data C22 according to the evaluation result (step S109). When performing the correction process (Yes at step S109), the processing circuit 144 inputs the image data C21 and the image data C22 to the learned model M1 (step S110), and generates the image data C23 (step S111).

Then, the processing circuit 144 generates a display image and displays it on the display 142 (step S112), and ends the process. Note that if the image data C23 has been generated, the processing circuit 144 causes the display 142 to display a display image based on the image data C23 in step S112. On the other hand, if the correction process is not performed (No in step S109), the processing circuit 144 causes the display 142 to display a display image based on the image data C22 in step S112.

Note that Step S108 and Step S109 described above may be omitted. That is, the processing circuit 144 may input the image data C21 and the image data C22 to the learned model M1 and generate the image data C23, regardless of the quality of the image data C22.

As described above, according to the first embodiment, the scan function 144a executes the scan A21 to collect the projection data set B21 by irradiating the range R1 along the body axis direction of the subject P1 with X-rays. do. Further, after the scan A21, the scan function 144a executes a scan A22 that collects a projection data set B22 by irradiating a range R2 along the body axis direction of the subject P1 with X-rays. Furthermore, the processing function 144b inputs image data C21 based on at least a part of the projection data set B21 and image data C22 based on the projection data set B22 to the trained model M1, thereby improving the quality of the image data C22. image data C23 with a high value is generated. That is, the X-ray CT system 10 according to the first embodiment can generate image data C23, which is CT image data of higher quality than image data C22.

In particular, the X-ray CT system 10 generates image data C23 based not only on the image data C22 but also on the image data C21. That is, in the X-ray CT system 10, since additional anatomical information can be obtained from the image data C21, more appropriate correction processing is performed compared to the case where only the image data C22 is used, and the image data The quality of C23 can be further improved.

Further, the X-ray CT system 10 sets the range R2 in the scan A22 based on the result of the scan A21, and improves the quality of the CT image data. That is, the X-ray CT system 10 can make the exposure of the subject P1 due to the scan A21 more meaningful. Note that scan A21 is positioning imaging for setting range R2 in scan A22. That is, scan A21 is executed before scan A22 in the normal processing flow, and does not complicate the processing flow.

(Second embodiment)
In the first embodiment described above, a case has been described in which high-quality CT image data is generated by inputting the image data C21 and the image data C22 to the trained model M1. In contrast, in the second embodiment, a case will be described in which high-quality CT image data is generated by inputting a projection data set B21 and a projection data set B22 to a learned model M2, which will be described later.

The X-ray CT system 10 according to the second embodiment has the same configuration as the X-ray CT system 10 according to the first embodiment shown in FIG. Hereinafter, the same reference numerals as in FIG. 1 will be given to the points explained in the first embodiment, and the explanation will be omitted.

First, the processing function 144b generates a learned model M2 and stores it in the memory 141. For example, the processing function 144b acquires the projection data set B11, the projection data set B12, and the projection data set B13 as learning data used to generate the learned model M2.

For example, the scan function 144a fixes the position of the X-ray tube 111 at a predetermined rotation angle and irradiates the subject P2 with X-rays from the X-ray tube 111 while moving the top plate 133 in the Z-axis direction. , collects a projection data set B11. Furthermore, the scan function 144a collects a projection data set B12 by irradiating the subject P2 with X-rays having a dose of X1 while rotating the X-ray tube 111 around the subject P2. Furthermore, the scan function 144a collects the projection data set B13 by irradiating the subject P2 with X-rays at a dose X2 higher than the dose X1 while rotating the X-ray tube 111 around the subject P2. That is, the projection data set B13 is a projection data set of higher quality than the projection data set B12.

Then, the processing function 144b generates the learned model M2 by executing machine learning using the projection data set B11 and the projection data set B12 as input data and the projection data set B13 as output data. Here, the learned model M2 can be configured by, for example, a neural network. For example, the processing function 144b generates the trained model M2 by performing deep learning on a multilayer neural network using the projection data set B11, the projection data set B12, and the projection data set B13.

For example, the processing function 144b inputs the projection data set B11 and the projection data set B12 as input side data to the neural network. Here, in the neural network, information is propagated in one direction from the input layer side to the output layer side while connecting only between adjacent layers, and the output layer outputs a projection dataset that estimates the projection dataset B13. be done. That is, in the neural network, processing is performed to improve the quality of the projection data set B12, and the output layer outputs a projection data set in which the projection data set B13, which is higher in quality than the projection data set B12, is estimated.

The processing function 144b generates the trained model M2 by adjusting the parameters of the neural network so that the neural network can output a preferable result when input side data is input. For example, processing function 144b adjusts the parameters of the neural network using an error function that represents the closeness between the projection data sets.

As an example, the processing function 144b calculates an error function E3 indicating the closeness between the projection data set B13 and the projection data set estimated by the neural network. The processing function 144b also calculates an error function E4 that indicates the consistency between the projection data set B11 and the image data estimated by the neural network. Then, the processing function 144b adjusts the parameters of the neural network so that the calculated error function becomes minimum. For example, the processing function 144b adjusts the parameters of the neural network so that the sum of the error function E3 and the error function E4 becomes minimum. Thereby, the processing function 144b accepts input of a two-dimensional projection data set and a three-dimensional projection data set, and a trained model equipped with a function to improve the quality of the input three-dimensional projection data set. Generate M2. Furthermore, the processing function 144b stores the generated learned model M2 in the memory 141.

Then, when the scan for the subject P1 is executed, the processing function 144b reads out the learned model M2 from the memory 141, and uses the read out learned model M2 to improve the quality of the projection data set collected from the subject P1. and thus improve the quality of CT image data based on projection data sets.

For example, the scan function 144a first executes the scan A21 shown in FIG. 3 and collects the projection data set B21. Furthermore, the processing function 144b generates image data C21 based on the projection data set B21 collected by scan A21. Next, the control function 144c sets a range R2, which is the scan range of the scan A22, by accepting an input operation from the user who referred to the reference image based on the image data C21. Next, the scan function 144a executes a scan A22 on the range R2 set in the reference image, and collects a projection data set B22.

Processing function 144b then evaluates the quality of projection data set B22. For example, processing function 144b evaluates the quality of projection dataset B22 by calculating a signal-to-noise ratio for projection dataset B22 and comparing the signal-to-noise ratio with a threshold. Further, for example, the control function 144c causes the display 142 to display an image (for example, a sinogram, etc.) based on the projection data set B22. The processing function 144b then receives an input of evaluation regarding the quality of the projection data set B22 from the user who has viewed the image based on the projection data set B22.

Next, the processing function 144b determines whether or not to perform correction processing on the projection data set B22, depending on the evaluation result of the quality of the projection data set B22. That is, if the quality of the projection data set B22 is sufficiently high, the processing function 144b may omit the correction processing described later. In this case, the processing function 144b can reconstruct high quality image data C22 based on the projection data set B22. Note that the process of determining whether or not to perform the correction process on the projection data set B22 may be omitted as appropriate.

When performing correction processing on the projection data set B22, the processing function 144b inputs the projection data set B21 and the projection data set B22 to the learned model M2 read from the memory 141. As a result, the trained model M2 outputs a projection data set B23 having higher quality than the projection data set B22. That is, the learned model M2 generates the projection data set B23 by correcting the projection data set B22 while maintaining consistency with the projection data set B21. Note that the projection data set B23 is an example of the third projection data set.

Note that the processing function 144b may input the entire projection data set B21 to the learned model M2, or may input only a part of it. For example, the processing function 144b may input only the portion included in the range R2 of the projection data set B21 to the learned model M2. In other words, the processing function 144b inputs at least a portion of the projection data set B21 collected from the range R1 to the learned model M2.

Furthermore, the processing function 144b generates three-dimensional image data C24 based on the projection data set B23 generated using the learned model M2. For example, the processing function 144b performs reconstruction processing such as a filtered back projection method, a successive approximation reconstruction method, a successive approximation applied reconstruction method, and a DLR method based on the projection data set B23, thereby creating a three-dimensional image. Data C24 is reconstructed. Here, the image data C24 is generated based on the high quality projection data set B23, and is higher quality CT image data than the image data C22 etc. generated based on the projection data set B22. be. Furthermore, the control function 144c generates a display image based on the image data C24, and causes the display 142 to display the generated image.

As described above, according to the second embodiment, the scan function 144a executes the scan A21 to collect the projection data set B21 by irradiating the range R1 along the body axis direction of the subject P1 with X-rays. do. Further, after the scan A21, the scan function 144a executes a scan A22 that collects a projection data set B22 by irradiating a range R2 along the body axis direction of the subject P1 with X-rays. Furthermore, the processing function 144b generates a projection data set B23 having higher quality than the projection data set B22 by inputting at least a part of the projection data set B21 and the projection data set B22 to the learned model M1. . The X-ray CT system 10 according to the second embodiment can generate image data C24, which is high-quality CT image data, based on the projection data set B23.

In particular, the X-ray CT system 10 generates the projection data set B23 based not only on the projection data set B22 but also on the projection data set B21. That is, in the X-ray CT system 10, since additional anatomical information can be obtained from the projection data set B21, more appropriate correction processing can be performed compared to the case where only the projection data set B22 is used. The quality of the projection data set B23 can be further improved. As a result, the X-ray CT system 10 can further improve the quality of the image data C24 based on the projection data set B23.

(Third embodiment)
Although the first and second embodiments have been described so far, the present invention may be implemented in various different forms in addition to the above-described embodiments.

For example, although the description has been made so far that the range R2, which is the scan range of the scan A22, is set by the user, the embodiment is not limited to this. For example, the control function 144c may automatically set the range R2 by analyzing the image data C21 or a reference image generated based on the image data C21 and extracting an organ to be diagnosed.

Furthermore, in the embodiments described above, a case has been described in which positioning imaging is performed in two dimensions. However, the embodiment is not limited to this, and the X-ray CT system 10 may perform positioning imaging in three dimensions.

When performing positioning photography in three dimensions, the processing function 144b first performs machine learning using three-dimensional positioning image data instead of two-dimensional positioning image data to generate a learned model. For example, the processing function 144b performs machine learning using the image data C15 collected by performing three-dimensional positioning imaging for the subject P2 and the image data C12 and image data C13 shown in FIG. 2 as learning data. By executing this, a learned model M3 is generated.

For example, the X-ray CT system 10 collects image data C15 by performing a scan A15 on the subject P2. To give an example, first, the scan function 144a rotates the X-ray tube 111 around the subject P2 and irradiates X-rays from the X-ray tube 111 to the subject P2, thereby generating multiple views. Collect projection data for each. Hereinafter, the plurality of projection data collected by scan A15 will be referred to as projection data set B15. Then, the processing function 144b generates image data C15 by executing a reconstruction process based on the collected projection data set B15. That is, the image data C15 is three-dimensional image data collected by three-dimensional scanning. For example, the image data C15 is positioning image data for setting the scan range of scan A12 and scan A13.

Furthermore, the scan function 144a executes a scan A12 by irradiating the subject P2 with X-rays at a dose of X1 while rotating the X-ray tube 111 around the subject P2, and collects a projection data set B12. Furthermore, the processing function 144b generates image data C12 by executing a reconstruction process based on the projection data set B12. In addition, the scan function 144a executes a scan A13 by irradiating the subject P2 with X-rays at a dose X2 higher than the dose X1 while rotating the X-ray tube 111 around the subject P2, and sets the projection data. Collect B13. Furthermore, the processing function 144b generates image data C13 by executing a reconstruction process based on the projection data set B13. That is, the image data C13 is image data of higher quality than the image data C12.

Then, the processing function 144b generates a learned model M3 by executing machine learning using the image data C15 and the image data C12 as input side data and the image data C13 as output side data. Here, the trained model M3 can be configured by, for example, a neural network. For example, the processing function 144b generates the trained model M3 by performing deep learning on a multilayer neural network using the image data C15, image data C12, and image data C13.

For example, the processing function 144b inputs the image data C15 and the image data C12 as input data to the neural network. Here, in the neural network, information is propagated in one direction from the input layer side to the output layer side while connecting only between adjacent layers, and the output layer outputs image data obtained by estimating the image data C13. . That is, in the neural network, processing is performed to improve the quality of the image data C12, and the output layer outputs image data in which image data C13, which is higher in quality than the image data C12, is estimated.

The processing function 144b generates a learned model M3 by adjusting the parameters of the neural network so that the neural network can output a preferable result when input data is input. For example, the processing function 144b adjusts the parameters of the neural network using an error function representing the closeness between image data.

For example, the processing function 144b calculates an error function E5 indicating the closeness between the image data C13 and the image data estimated by the neural network. Furthermore, the processing function 144b calculates an error function E6 indicating the closeness between the image data C15 and the image data estimated by the neural network. Then, the processing function 144b adjusts the parameters of the neural network so that the calculated error function becomes minimum. For example, the processing function 144b adjusts the parameters of the neural network so that the sum of the error function E5 and the error function E6 becomes minimum. Thereby, the processing function 144b accepts the input of the first image data collected by three-dimensional positioning photography and the second image data collected by three-dimensional main scanning, and improves the quality of the input second image data. A trained model M3 that is functionalized to improve the performance is generated. Furthermore, the processing function 144b stores the generated trained model M3 in the memory 141.

Then, when the scan for the subject P1 is executed, the processing function 144b reads out the learned model M3 from the memory 141, and uses the read out learned model M3 to improve the quality of the CT image data collected from the subject P1. improve.

For example, the scan function 144a first collects image data C25 by performing a scan A25 on the subject P1. To give an example, first, the scan function 144a rotates the X-ray tube 111 around the subject P1 and irradiates the subject P1 with X-rays from the X-ray tube 111, thereby generating multiple views. Collect projection data for each. Hereinafter, the plurality of projection data collected by scan A25 will be referred to as projection data set B25. For example, the scan function 144a collects the projection data set B25 by irradiating the range R1 shown in FIG. 3 with X-rays. Then, the processing function 144b generates image data C25 by executing a reconstruction process based on the collected projection data set B25. That is, the image data C25 is three-dimensional image data collected by three-dimensional scanning.

Note that scan A25 is an example of the first scan. Further, the projection data set B25 is an example of the first projection data set. Further, the image data C25 is an example of first image data.

Next, the control function 144c sets a range R2, which is the scan range of the scan A22, by accepting an input operation from the user who referred to the reference image based on the image data C25. That is, the image data C25 is positioning image data for setting the scan range of the scan A22. Next, the scan function 144a executes a scan A22 on the range R2 set in the reference image, and collects a projection data set B22. Furthermore, the processing function 144b generates image data C22 based on the projection data set B22.

Next, the processing function 144b evaluates the quality of the image data C22. Next, the processing function 144b determines whether or not to perform correction processing on the image data C22, depending on the evaluation result of the quality of the image data C22. That is, if the quality of the image data C22 is sufficiently high, the processing function 144b may omit the correction processing described later. Note that the process of determining whether or not to perform correction processing on the image data C22 may be omitted as appropriate.

When performing correction processing on the image data C22, the processing function 144b inputs the image data C25 and the image data C22 to the learned model M3 read out from the memory 141. As a result, the trained model M3 outputs image data C26 having higher quality than the image data C22. That is, the trained model M3 generates the image data C26 by correcting the image data C22 while maintaining consistency with the image data C25. Note that the image data C26 is an example of third image data.

Note that the processing function 144b may input all of the image data C25 or only a portion of the image data C25 to the trained model M3. For example, the processing function 144b may input only the portion included in the range R2 of the image data C25 to the trained model M3. In other words, the processing function 144b inputs the image data C25 based on at least a portion of the projection data set B25 collected from the range R1 into the trained model M3.

To give another example, the processing function 144b uses the projection data set B15 collected by performing three-dimensional positioning imaging for the subject P2, the projection data set B12, and the projection data set B13 as learning data. A learned model M4 is generated by executing machine learning.

For example, the scan function 144a executes a scan A15 by irradiating the subject P2 with X-rays while rotating the X-ray tube 111 around the subject P2, and collects a projection data set B15. Furthermore, the scan function 144a executes a scan A12 by irradiating the subject P2 with X-rays at a dose of X1 while rotating the X-ray tube 111 around the subject P2, and collects a projection data set B12. In addition, the scan function 144a executes a scan A13 by irradiating the subject P2 with X-rays at a dose X2 higher than the dose X1 while rotating the X-ray tube 111 around the subject P2, and sets the projection data. Collect B13. That is, the projection data set B13 is a projection data set of higher quality than the projection data set B12.

Then, the processing function 144b generates a learned model M4 by executing machine learning using the projection data set B15 and the projection data set B12 as input data and the projection data set B13 as output data. Here, the learned model M4 can be configured by, for example, a neural network. For example, the processing function 144b generates the trained model M4 by performing deep learning on a multilayer neural network using the projection data set B15, the projection data set B12, and the projection data set B13.

For example, the processing function 144b inputs the projection data set B15 and the projection data set B12 as input data to the neural network. Here, in the neural network, information is propagated in one direction from the input layer side to the output layer side while connecting only between adjacent layers, and the output layer outputs a projection dataset that estimates the projection dataset B13. be done. That is, in the neural network, a process is performed to improve the quality of the projection data set B12, and the output layer outputs a projection data set in which the projection data set B13, which is higher in quality than the projection data set B12, is estimated.

The processing function 144b generates a learned model M4 by adjusting the parameters of the neural network so that the neural network can output a preferable result when input data is input. For example, processing function 144b adjusts the parameters of the neural network using an error function that represents the closeness between the projection data sets.

As an example, the processing function 144b calculates an error function E7 indicating the closeness between the projection data set B13 and the projection data set estimated by the neural network. Furthermore, the processing function 144b calculates an error function E8 indicating the closeness between the projection data set B15 and the image data estimated by the neural network. Then, the processing function 144b adjusts the parameters of the neural network so that the calculated error function becomes minimum. For example, the processing function 144b adjusts the parameters of the neural network so that the sum of the error function E7 and the error function E8 becomes minimum. Thereby, the processing function 144b receives the input of the first projection data set collected in the three-dimensional positioning imaging and the second projection data set collected in the three-dimensional main scan, and the processing function 144b receives the input second projection data set. Generate a trained model M4 that is functionalized to improve the quality of the set. Furthermore, the processing function 144b stores the generated trained model M4 in the memory 141.

Then, when the scan for the subject P1 is executed, the processing function 144b reads out the learned model M4 from the memory 141, and uses the read out learned model M4 to improve the quality of the projection data set collected from the subject P1. and thus improve the quality of CT image data based on projection data sets.

For example, the scan function 144a first collects a projection data set B25 by performing a scan A25 on the subject P1. For example, the scan function 144a collects the projection data set B25 by irradiating the range R1 shown in FIG. 3 with X-rays. Furthermore, the processing function 144b generates image data C25 by executing a reconstruction process based on the collected projection data set B25.

Next, the control function 144c sets a range R2, which is the scan range of the scan A22, by accepting an input operation from the user who referred to the reference image based on the image data C25. Next, the scan function 144a executes a scan A22 on the range R2 set in the reference image, and collects a projection data set B22.

Processing function 144b then evaluates the quality of projection data set B22. Furthermore, the processing function 144b determines whether or not to perform correction processing on the projection data set B22, depending on the evaluation result of the quality of the projection data set B22. That is, if the quality of the projection data set B22 is sufficiently high, the processing function 144b may omit the correction processing described later. In this case, the processing function 144b can reconstruct high quality image data C22 based on the projection data set B22. Note that the process of determining whether or not to perform the correction process on the projection data set B22 may be omitted as appropriate.

When performing correction processing on the projection data set B22, the processing function 144b inputs the projection data set B25 and the projection data set B22 to the trained model M4 read from the memory 141. As a result, the learned model M4 outputs a projection data set B26 having higher quality than the projection data set B22. That is, the learned model M4 generates the projection data set B26 by correcting the projection data set B22 while maintaining consistency with the projection data set B25. Note that the projection data set B26 is an example of the third projection data set. Furthermore, the processing function 144b reconstructs image data C27, which is high-quality CT image data, based on the projection data set B26 generated using the learned model M4.

Note that the processing function 144b may input the entire projection data set B25 to the trained model M4, or may input only a part of it. For example, the processing function 144b may input only the portion included in the range R2 of the projection data set B25 to the learned model M4. In other words, the processing function 144b inputs at least a portion of the projection data set B25 collected from the range R1 to the learned model M4.

Further, although the learned models M1 to M4 have been described as being configured by neural networks, the processing function 144b may generate the learned models M1 to M4 by a machine learning method other than the neural network. Furthermore, although the processing function 144b has been described as generating the learned models M1 to M4, the learned models M1 to M4 may be generated by other devices.

Further, although the X-ray CT system 10 has been described as executing the correction processing of the projection data set or CT image data, the embodiment is not limited to this. For example, the correction process may be executed in another device different from the X-ray CT system 10. This point will be explained below using the medical information processing system 1 shown in FIG. 5 as an example. FIG. 5 is a block diagram showing an example of the configuration of the medical information processing system 1 according to the fifth embodiment. The medical information processing system 1 includes an X-ray CT system 10 and a medical processing device 20 that performs correction processing.

As shown in FIG. 5, the X-ray CT system 10 and the medical processing device 20 are interconnected via a network NW. Here, the X-ray CT system 10 and the medical processing device 20 can be installed at any location as long as they can be connected via the network NW. For example, the medical processing device 20 may be installed in a different hospital than the X-ray CT system 10. That is, the network NW may be configured as a local network closed within the hospital, or may be a network via the Internet. Furthermore, although one X-ray CT system 10 is shown in FIG. 5, the medical information processing system 1 may include a plurality of X-ray CT systems 10.

The medical processing device 20 is realized by computer equipment such as a workstation. For example, the medical processing device 20 includes a memory 21, a display 22, an input interface 23, and a processing circuit 24, as shown in FIG.

The memory 21 is realized by, for example, a RAM, a semiconductor memory device such as a flash memory, a hard disk, an optical disk, or the like. For example, the memory 21 stores various data transmitted from the X-ray CT system 10. Further, for example, the memory 21 stores a program for a circuit included in the medical processing device 20 to realize its function. Note that the memory 21 may be realized by a server group (cloud) connected to the medical processing device 20 via the network NW.

The display 22 displays various information. For example, the display 22 displays a CT image for display generated by the processing circuit 24, a GUI, etc. for accepting various operations from the user. For example, the display 22 is a liquid crystal display or a CRT display. The display 22 may be of a desktop type, or may be composed of a tablet terminal or the like that can communicate wirelessly with the main body of the medical processing device 20.

The input interface 23 accepts various input operations from the user, converts the received input operations into electrical signals, and outputs the electrical signals to the processing circuit 24 . For example, the input interface 23 receives from the user reconstruction conditions when reconstructing CT image data, image processing conditions when generating a CT image for display from CT image data, and the like. For example, the input interface 23 may include a mouse, a keyboard, a trackball, a switch, a button, a joystick, a touchpad that performs input operations by touching the operation surface, a touchscreen that integrates a display screen and a touchpad, and an optical sensor. This is realized by using a non-contact input circuit, a voice input circuit, etc. Note that the input interface 23 may be configured with a tablet terminal or the like that can communicate wirelessly with the main body of the medical processing device 20. Furthermore, the input interface 23 is not limited to one that includes physical operating components such as a mouse and a keyboard. For example, an example of the input interface 23 is an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the medical processing device 20 and outputs this electrical signal to the processing circuit 24. include.

The processing circuit 24 controls the overall operation of the medical processing device 20 by executing a processing function 24a and a control function 24b. Note that the processing function 24a is an example of a processing section. Processing by the processing circuit 24 will be described later.

In the medical processing device 20 shown in FIG. 5, each processing function is stored in the memory 21 in the form of a computer-executable program. The processing circuit 24 is a processor that reads programs from the memory 21 and executes them to implement functions corresponding to each program. In other words, the processing circuit 24 in a state where the program has been read has a function corresponding to the read program.

Although the processing function 24a and the control function 24b are described as being implemented in a single processing circuit 24 in FIG. 5, the processing circuit 24 is configured by combining a plurality of independent processors, and each processor executes a program. The function may be realized by executing the . Further, each processing function of the processing circuit 24 may be realized by being appropriately distributed or integrated into a single or multiple processing circuits.

Furthermore, the processing circuit 24 may realize its functions using a processor of an external device connected via the network NW. For example, the processing circuit 24 reads and executes a program corresponding to each function from the memory 21, and also uses a server group (cloud) connected to the medical processing device 20 via the network NW as a computing resource. Each function shown in FIG. 5 is realized.

For example, first, the scan function 144a in the X-ray CT system 10 executes a first scan to collect a first projection data set by irradiating an X-ray to a range R1 along the body axis direction of the subject P1. . Furthermore, after the first scan, the scan function 144a performs a second projection data set that collects a second projection data set by irradiating an X-ray to a range R2 along the body axis of the subject P1, which is narrower than the range R1. Run a scan.

The processing function 144b also generates first image data based on at least a portion of the first projection data set and second image data based on the second projection data set. Next, the control function 144c transmits the first image data and the second image data to the medical processing device 20 via the network NW. Alternatively, the control function 144c sends the first projection data set and the second projection data set to the medical processing device 20. The processing functionality 24a then generates first image data based at least in part on the first projection data set and second image data based on the second projection data set.

The processing function 24a then executes correction processing on the second image data. For example, the processing function 24a generates third image data having higher quality than the second image data by inputting the first image data and the second image data to the above-mentioned trained model M1 or trained model M3. do. Note that the processing function 24a may evaluate the quality of the second image data prior to the correction process, and determine whether or not to generate the third image data according to the evaluation result.

As another example, the control function 144c sends the first projection data set and the second projection data set to the medical processing device 20. The processing function 24a then executes a correction process on the second projection data set. For example, the processing function 24a inputs at least a portion of the first projection data set and the second projection data set to the above-mentioned trained model M2 or trained model M4, so that the quality is higher than that of the second projection data set. Generate a high third projection data set. Note that the processing function 24a may evaluate the quality of the second projection data set prior to the correction process, and determine whether or not to generate the third projection data set based on the evaluation result.

The processing function 24a then reconstructs, based on the third projection data set, third image data having higher quality than the CT image data based on the second projection data set. Alternatively, the control function 24b transmits the third projection data set to the X-ray CT system 10 via the network NW. In this case, the processing function 144b reconstructs the third image data based on the third projection data set.

The term "processor" used in the above description refers to, for example, a CPU, a GPU (Graphics Processing Unit), an application specific integrated circuit (ASIC), a programmable logic device (for example, a simple programmable logic device), or an application specific integrated circuit (ASIC). Simple Programmable Logic Device (SPLD), Complex Programmable Logic Device (CPLD), and Field Programmable Gate Array ：FPGA)) and other circuits. The processor realizes functions by reading and executing programs stored in the memory 141 or the memory 21.

In addition, in FIG. 1, the single memory 141 has been described as storing programs corresponding to each processing function of the processing circuit 144. Furthermore, in FIG. 5, the explanation has been made assuming that the single memory 21 stores programs corresponding to each processing function of the processing circuit 24. However, embodiments are not limited thereto. For example, a configuration may be adopted in which a plurality of memories 141 are arranged in a distributed manner, and the processing circuit 144 reads a corresponding program from each individual memory 141. Similarly, a configuration may be adopted in which a plurality of memories 21 are arranged in a distributed manner and the processing circuit 24 reads the corresponding program from each individual memory 21. Further, instead of storing the program in the memory 141 or the memory 21, the program may be directly incorporated into the circuit of the processor. In this case, the processor realizes its functions by reading and executing a program built into the circuit.

Each component of each device according to the embodiments described above is functionally conceptual, and does not necessarily need to be physically configured as shown in the drawings. In other words, the specific form of distributing and integrating each device is not limited to what is shown in the diagram, and all or part of the devices can be functionally or physically distributed or integrated in arbitrary units depending on various loads and usage conditions. Can be integrated and configured. Furthermore, all or any part of each processing function performed by each device can be realized by a CPU and a program that is analyzed and executed by the CPU, or can be realized as hardware using wired logic.

Further, the processing method described in the above embodiment can be realized by executing a processing program prepared in advance on a computer such as a personal computer or a workstation. This processing program can be distributed via a network such as the Internet. In addition, this processing program is recorded on a computer-readable non-transitory recording medium such as a hard disk, flexible disk (FD), CD-ROM, MO, DVD, etc., and is executed by being read from the recording medium by the computer. You can also.

According to at least one embodiment described above, high quality CT image data can be generated.

Although several embodiments of the invention have been described, these embodiments are presented by way of example and are not intended to limit the scope of the invention. These embodiments can be implemented in various other forms, and various omissions, substitutions, and changes can be made without departing from the gist of the invention. These embodiments and their modifications are included within the scope and gist of the invention as well as within the scope of the invention described in the claims and its equivalents.

1 Medical information processing system 10 X-ray CT system 110 Frame device 140 Console device 144 Processing circuit 144a Scan function 144b Processing function 144c Control function 20 Medical processing device 24 Processing circuit 24a Processing function 24b Control function

Claims (6)

first learning data that is image data reconstructed based on a projection data set collected by a scan in which the position of the X-ray tube is not rotated around the scan target; and second learning data, which is image data reconstructed based on the projection data set collected by the scan performed at the first dose while rotating the X-ray tube, and the position of the X-ray tube is outputting third learning data, which is image data reconstructed based on a projection data set collected by a scan performed at a second dose higher than the first dose while rotating around the scan target; a processing unit that generates a trained model by executing machine learning using the data;
A first scan in which a first projection data set is collected by irradiating a first range along the body axis of the subject with X-rays without rotating the position of the X-ray tube around the subject. After the first scan, while rotating the position of the X-ray tube around the subject, X-rays are applied to a second range narrower than the first range along the body axis direction. a scanning unit that performs a second scan that collects a second projection data set by irradiating the
The processing unit may process the first image data reconstructed based on at least a portion of the first projection data set and the second image data reconstructed based on the second projection data set into the learned data set. An X-ray CT system that generates third image data having higher quality than the second image data by inputting the data to a model. The scanning unit collects the first projection data set while moving at least one of a position of an X-ray tube that generates X-rays and the subject along the body axis direction. X-ray CT system. The X-ray CT system according to claim 2, wherein the scanning unit collects the second projection data set without moving the position of the X-ray tube and the subject along the body axis direction. 4. The processing unit according to claim 1 , wherein the processing unit evaluates the quality of the second image data and determines whether or not to generate the third image data according to the evaluation result. The X-ray CT system described. The second range is generated based on at least a portion of the first projection data set and is set by an operation on positioning image data displayed on a display unit. The X-ray CT system described. first learning data that is image data reconstructed based on a projection data set collected by a scan in which the position of the X-ray tube is not rotated around the scan target; and second learning data, which is image data reconstructed based on the projection data set collected by the scan performed at the first dose while rotating the X-ray tube, and the position of the X-ray tube is outputting third learning data, which is image data reconstructed based on a projection data set collected by a scan performed at a second dose higher than the first dose while rotating around the scan target; Equipped with a processing unit that generates a trained model by executing machine learning on the data,
The processing unit is configured to generate a first projection collected by a first scan in which a first range along the body axis of the subject is irradiated with X-rays without rotating the position of the X-ray tube around the subject. first image data reconstructed based on at least a part of the data set; and first image data narrower than the first range along the body axis direction while rotating the position of the X-ray tube around the subject. inputting second image data reconstructed based on a second projection data set collected after the first scan by a second scan in which X-rays are irradiated to a range of 2 to the learned model; A medical processing device that generates third image data having a higher quality than the second image data.

Images