JP7213099B2 - Medical image processing device, magnetic resonance imaging device, medical image processing method, and program - Google Patents

Description

TECHNICAL FIELD Embodiments of the present invention relate to a medical image processing apparatus, a magnetic resonance imaging apparatus, a medical image processing method, and a program.

An MRI apparatus using a magnet that generates a high magnetic field of 3 Tesla or more can obtain a high SNR (Signal to Noise Ratio). As the magnetic field increases, unevenness tends to occur, which may affect the image quality. Since the transmission unevenness of the high-frequency magnetic field changes depending on the body size of the subject, it is necessary to deal with it.

JP 2013-176553 A

The problem to be solved by the present invention is to improve the image quality of medical images.

A medical image processing apparatus according to an embodiment includes an acquisition unit and a generation unit. The acquisition unit acquires a medical image based on magnetic resonance signals obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF (Radio Frequency) coil. When the image is input, the generation unit is configured to output a corrected image that reduces the influence of non-uniformity of the high-frequency magnetic field from the input image. By inputting the medical image, a corrected image in which the influence of the non-uniformity is reduced is generated from the medical image acquired by the acquisition unit.

1 is a diagram showing an example of the configuration of a medical image processing system 1 including a medical image processing apparatus 200 according to an embodiment; FIG. The figure which shows an example of the MRI apparatus 100 which concerns on embodiment. 1 is a diagram showing an example of a medical image processing apparatus 200 according to an embodiment; FIG. FIG. 4 is a diagram showing an example of teacher information 234; FIG. 4 is a flow chart showing the flow of a series of processes during runtime of the processing circuit 210 on the side of the medical image processing apparatus 200 according to the embodiment. 4 is a diagram showing an example of the configuration of a correction model MDL at runtime according to the embodiment; FIG. 4 is a flow chart showing a series of processes during training of the processing circuit 210 on the side of the medical image processing apparatus 200 according to the embodiment. FIG. 4 is a diagram showing an example of the configuration of a correction model MDL during training according to the embodiment; FIG. 4 shows another example of the MRI apparatus 100 according to the embodiment;

Hereinafter, embodiments of a medical image processing apparatus, a magnetic resonance imaging apparatus, a medical image processing method, and a program will be described in detail with reference to the drawings.

FIG. 1 is a diagram showing an example of the configuration of a medical image processing system 1 including a medical image processing apparatus 200 according to an embodiment. For example, as shown in FIG. 1, the medical image processing system 1 includes an MRI apparatus 100 and a medical image processing apparatus 200. FIG. The MRI apparatus 100 and the medical image processing apparatus 200 are connected via a network NW. The network NW includes, for example, a WAN (Wide Area Network), a LAN (Local Area Network), the Internet, a leased line, a radio base station, a provider, and the like.

The MRI apparatus 100 is, for example, an apparatus that generates an MR image, which is one of medical images, from MR (Magnetic Resonance) signals obtained by a nuclear magnetic resonance phenomenon caused by applying a magnetic field to a subject (eg, human body).

The medical image processing apparatus 200 is implemented by one or more processors. For example, the medical image processing apparatus 200 may be a computer included in a cloud computing system, or a computer that operates independently without depending on other devices (standalone computer).

[Configuration example of MRI apparatus]
FIG. 2 is a diagram showing an example of the MRI apparatus 100 according to the embodiment. As shown in FIG. 2, the MRI apparatus 100 includes a static magnetic field magnet 101, a gradient magnetic field coil 102, a gradient magnetic field power supply 103, a bed 104, a bed control circuit 105, one or more transmission coils 106, and a transmission circuit. 107 , one or more receive coils 108 , receive circuitry 109 , sequence control circuitry 110 and console device 120 . Transmit coil 106 and receive coil 108 are also referred to as "RF coils."

The static magnetic field magnet 101 is a hollow, substantially cylindrical magnet that generates a uniform static magnetic field in the internal space. The static magnetic field magnet 101 is, for example, a permanent magnet or a superconducting magnet. The gradient magnetic field coil 102 is a coil formed in a hollow, substantially cylindrical shape, and is arranged inside the static magnetic field magnet 101 . The gradient magnetic field coil 102 is formed by combining three coils corresponding to the mutually orthogonal x, y, and z axes. The z-axis direction represents the longitudinal direction of the top plate 104a of the bed 104, and the x-axis direction represents the axial direction orthogonal to the z-axis direction and parallel to the floor surface of the room in which the MRI apparatus 100 is installed. , the y-axis direction represents the axial direction perpendicular to the floor surface. The three coils corresponding to the respective axial directions receive individual currents from the gradient magnetic field power supply 103 and generate gradient magnetic fields whose magnetic field strengths vary along the respective x, y, and z axes. Note that the z-axis direction is the same direction as the static magnetic field.

A gradient magnetic field power supply 103 supplies current to the gradient magnetic field coil 102 . Here, the x-, y-, and z-axis gradient magnetic fields generated by the gradient magnetic field coil 102 correspond to, for example, the slice selection gradient magnetic field Gs, the phase encoding gradient magnetic field Ge, and the readout gradient magnetic field Gr, respectively. . The slice selection gradient magnetic field Gs is used to arbitrarily determine an imaging section. The phase-encoding gradient magnetic field Ge is used to change the phase of the magnetic resonance signal according to the spatial position. The readout gradient magnetic field Gr is used to change the frequency of the magnetic resonance signal according to the spatial position.

The bed 104 has a tabletop 104a on which the subject OB is placed. Under the control of the bed control circuit 105, the tabletop 104a is moved into the cavity ( (imaging opening). The bed 104 is usually installed so that its longitudinal direction is parallel to the central axis of the static magnetic field magnet 101 . The bed control circuit 105 drives the bed 104 under the control of the console device 120 to move the top board 104a in the longitudinal direction and the vertical direction.

A transmit coil 106 is arranged inside the gradient coil 102 . Transmit coil 106 is, for example, a Whole Body Coil housed in a cradle containing static field magnet 101 . The transmission coil 106 is supplied with current from the transmission circuit 107 and generates a high-frequency magnetic field for exciting nuclear spins inside the object OB.

The transmission circuit 107 supplies the transmission coil 106 with a current corresponding to the Larmor frequency determined by the type of nuclei of interest and the strength of the magnetic field, and causes the transmission coil 106 to generate a high-frequency magnetic field. Further, the transmission circuit 107 may independently control currents supplied to a plurality of feeding points of the transmission coil 106 to correct spatial non-uniformity of the high-frequency magnetic field. Such spatial homogenization of the magnetic field is called B1 shimming (RF shimming). In B1 shimming, the amplitude and phase of the high-frequency magnetic field generated from each transmission coil 106 are adjusted to spatially homogenize the magnetic field acting on the subject OB. Hereinafter, the high-frequency magnetic field generated from the transmission coil will be referred to as "transmission RF".

A receiving coil 108 is arranged inside the gradient coil 102 . The receiving coil 108 receives magnetic resonance signals emitted from the subject OB under the influence of the transmission RF. A magnetic resonance signal includes, for example, a signal intensity component and a phase component. Upon receiving the magnetic resonance signal, the reception coil 108 outputs the received magnetic resonance signal to the reception circuit 109 . The receiving coil 108 may be implemented by a phased array coil in which at least two or more coil elements are arranged in an array. Note that the receiving coil 108 is not limited to a phased array coil, and may be realized by one RF coil that combines transmission and reception.

The receiving circuit 109 generates magnetic resonance data based on the magnetic resonance signal output from the receiving coil 108 . Specifically, the receiving circuit 109 generates magnetic resonance data, which is a digital signal, by performing analog/digital conversion on the magnetic resonance signal, which is an analog signal output from the receiving coil 108 . The receiving circuit 109 transmits the generated magnetic resonance data to the sequence control circuit 110 . Further, the receiving circuit 109 may be provided on the side of the gantry device including the static magnetic field magnet 101, the gradient magnetic field coil 102, and the like. The magnetic resonance signals output from each coil element of the receiving coil 108 are appropriately distributed and combined and output to the receiving circuit 109 .

The sequence control circuit 110 drives the gradient magnetic field power supply 103, the transmission circuit 107, and the reception circuit 109 based on the sequence information transmitted from the console device 120, thereby imaging (scanning) the subject OB. Sequence information is information that defines a procedure for performing imaging processing. The sequence information includes the intensity and timing of the current supplied by the gradient magnetic field power supply 103 to the gradient magnetic field coil 102, the amplitude and phase of the transmission RF generated from the transmission coil 106 by the transmission circuit 107, the phase of the transmission RF generated by the transmission coil 106, the reception circuit 109 includes information defining the timing of detecting magnetic resonance signals.

The sequence control circuit 110 drives the gradient magnetic field power supply 103, the transmission circuit 107, and the reception circuit 109 to image the subject OB, and when receiving the magnetic resonance data from the reception circuit 109, the received magnetic resonance data is sent to the console device. 120.

The console device 120 controls the entire MRI apparatus 100 and collects magnetic resonance data. For example, console device 120 includes communication interface 122 , input interface 124 , display 126 , processing circuitry 130 and memory (storage) 150 .

The communication interface 122 includes, for example, a communication interface such as a NIC (Network Interface Card). The communication interface 122 communicates with the medical image processing apparatus 200 via the network NW and receives information from the medical image processing apparatus 200 . Communication interface 122 outputs the received information to processing circuitry 130 . The communication interface 122 may also transmit information to other devices connected via the network NW under the control of the processing circuitry 130 .

The input interface 124 is an interface that receives various input operations from the operator. Upon receiving an input operation, the input interface 124 converts the received input operation into an electrical signal and outputs the electrical signal to the processing circuit 130 . For example, the input interface 124 is implemented by a mouse, keyboard, trackball, switch, button, joystick, touch panel, and the like. Also, the input interface 124 may be implemented by, for example, a user interface that accepts voice input such as a microphone. When the input interface 124 is a touch panel, the display 126 (to be described later) may be formed integrally with the input interface 124 .

The display 126 displays various information. For example, the display 126 displays an image generated by the processing circuit 130, a GUI (Graphical User Interface) for accepting various input operations from the operator, and the like. For example, the display 126 is an LCD (Liquid Crystal Display), a CRT (Cathode Ray Tube) display, an organic EL (Electroluminescence) display, or the like.

Processing circuitry 130 performs, for example, acquisition function 132 , generation function 134 , communication control function 136 , and display control function 138 . The processing circuit 130 implements these functions by, for example, executing a program stored in a memory 150, which is a storage device (storage circuit), by a hardware processor provided in a computer.

A hardware processor that realizes each function of the processing circuit 130 is, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an application specific integrated circuit (ASIC), a programmable logic device, etc. Means circuit. The programmable logic device is, for example, a Simple Programmable Logic Device (SPLD), a Complex Programmable Logic Device (CPLD), a Field Programmable Gate Array (FPGA), or the like. Instead of storing the program in memory 150, the program may be configured to be directly embedded within the circuitry of the hardware processor. In this case, the hardware processor implements each of the above functions by reading and executing a program incorporated in the circuit. The hardware processor is not limited to being configured as a single circuit, and may be configured as one hardware processor by combining a plurality of independent circuits to implement each function. Also, a plurality of components may be integrated into one hardware processor to realize each function.

The memory 150 is implemented by, for example, a RAM (Random Access Memory), a semiconductor memory device such as a flash memory, a hard disk, an optical disk, or the like. These non-transitory storage media may be realized by other storage devices such as NAS (Network Attached Storage) and external storage server devices connected via the network NW. The non-transitory storage medium that implements the memory 150 may include a storage medium such as a ROM (Read Only Memory) and a register.

Acquisition function 132 acquires magnetic resonance data from sequence control circuit 110 . As described above, the magnetic resonance data is data obtained by analog/digital conversion of electromagnetic wave signals (magnetic resonance signals) generated in the subject OB by the nuclear magnetic resonance phenomenon. The data arranged according to the phase encoding amount and the frequency encoding amount applied by the gradient magnetic field described above is also called k-space data. The k-space represents a frequency space in which a one-dimensional waveform is acquired when a magnetic resonance signal is repeatedly acquired by the receiving coil 108 as a one-dimensional waveform.

The generation function 134 performs reconstruction processing including processing such as Fourier transform (for example, inverse Fourier transform) on the k-space data acquired by the acquisition function 132, thereby generating an MR image reconstructed from the k-space data. to generate The generation function 134 is an example of a "first generation unit". These medical images may be two-dimensional images or three-dimensional images.

When the generation function 134 generates the MR image by reconstruction, the communication control function 136 transmits the reconstructed MR image to the medical image processing apparatus 200 via the communication interface 122 . Also, the communication control function 136 may receive various information from the medical image processing apparatus 200 via the communication interface 122 .

A display control function 138 causes the display 126 to display the MR image generated by the generation function 134 . Further, when the communication interface 122 receives medical images such as MR images from the medical image processing apparatus 200 , the display control function 138 may cause the display 126 to display the medical images received by the communication interface 122 .

[Configuration example of medical image processing apparatus]
FIG. 3 is a diagram showing an example of the medical image processing apparatus 200 according to the embodiment. As shown in FIG. 3, the medical image processing apparatus 200 includes, for example, a communication interface 202, an input interface 204, a display 206, processing circuitry 210, and memory 230. FIG.

Communication interface 202 includes, for example, a communication interface such as a NIC. For example, the communication interface 202 communicates with the MRI apparatus 100 via the network NW and receives reconstructed MR images and the like from the MRI apparatus 100 . The communication interface 202 outputs the received medical images to the processing circuitry 210 . Further, the communication interface 202 may be controlled by the processing circuit 210 to transmit information to the MRI apparatus 100 or other apparatuses connected via the network NW. The other device may be, for example, a terminal device that can be used by an image reader such as a doctor.

The input interface 204 receives various input operations from the operator, converts the received input operations into electrical signals, and outputs the electrical signals to the processing circuit 210 . For example, the input interface 204 is implemented by a mouse, keyboard, trackball, switch, button, joystick, touch panel, and the like. Also, the input interface 204 may be realized by, for example, a user interface that accepts voice input such as a microphone. When the input interface 204 is a touch panel, a display 206 (to be described later) may be formed integrally with the input interface 204 .

The display 206 displays various information. For example, the display 206 displays an image generated by the processing circuit 210, and displays a GUI for accepting various input operations from the operator. For example, the display 206 is an LCD, CRT display, organic EL display, or the like.

Processing circuitry 210 performs, for example, acquisition function 212 , correction function 214 , power control function 216 , and learning function 218 . The acquisition function 212 is an example of an “acquisition unit”, the correction function 214 is an example of a “generation unit” or a “second generation unit”, and the learning function 218 is an example of a “learning unit”.

The processing circuit 210 implements these functions by, for example, executing a program stored in a memory 230, which is a storage device (storage circuit), by a hardware processor provided in the computer.

A hardware processor that implements each function of the processing circuit 210 means circuitry such as, for example, a CPU, a GPU, an application specific integrated circuit, a programmable logic device, and the like. Instead of storing the program in memory 230, the program may be configured to be directly embedded within the circuitry of the hardware processor. In this case, the hardware processor implements each of the above functions by reading and executing a program incorporated in the circuit. The hardware processor is not limited to being configured as a single circuit, and may be configured as one hardware processor by combining a plurality of independent circuits to implement each function. Also, a plurality of components may be integrated into one hardware processor to realize each function.

The memory 230 is implemented by, for example, a RAM, a semiconductor memory device such as a flash memory, a hard disk, an optical disk, or the like. These non-transitory storage media may be implemented by other storage devices connected via the network NW, such as NAS and external storage server devices. The memory 230 may also include non-transitory storage media such as ROMs and registers. The memory 230 stores, for example, correction model information 232, teacher information 234, sensitivity map 236, and the like.

The correction model information 232 is information (program or data structure) defining a correction model MDL, which will be described later. The correction model MDL is a model trained to output an image in which the influence of non-uniformity of the transmitted high-frequency magnetic field is reduced with respect to the input image. Hereinafter, the non-uniformity of the transmitted high-frequency magnetic field will be referred to as "Tx unevenness", and the image with reduced Tx unevenness will be referred to as "corrected image".

A sensitivity map is data of the spatial distribution of sensitivity that represents the sensitivity non-uniformity of the receiving coil 108 .

The correction model MDL may be realized by, for example, a DNN (Deep Neural Network(s)) such as a CNN (Convolutional Neural Network).

When the correction model MDL is implemented by a DNN, the correction model information 232 includes, for example, an input layer, one or more hidden layers (intermediate layers), and an output layer that constitute each DNN included in the correction model MDL. It includes connection information on how the included neurons (units or nodes) are connected to each other, and weight information on how many connection coefficients are given to data input and output between connected neurons. be The connection information includes, for example, the number of neurons included in each layer, information specifying the type of neuron to which each neuron is connected, an activation function that realizes each neuron, information such as gates provided between neurons in the hidden layer. including. The activation function that realizes the neuron may be, for example, a function that switches the operation according to the input code (ReLU (Rectified Linear Unit) function, ELU (Exponential Linear Units) function, clipping function), a sigmoid function, , a step function, a hyperpolic tangent function, or an identity function. A gate selectively passes or weights data transferred between neurons, for example, depending on the value (eg, 1 or 0) returned by the activation function. A coupling coefficient is a parameter of an activation function. For example, in a hidden layer of a neural network, a weight given to output data when data is output from a neuron in a certain layer to a neuron in a deeper layer. including. The coupling coefficients may also include bias components unique to each layer, and the like.

The teacher information 234 is information used for supervised learning of the correction model MDL described above. FIG. 4 is a diagram showing an example of the teacher information 234. As shown in FIG. As illustrated, the teacher information 234 is information in which a model teacher image to be output by the corrected model MDL is associated as a teacher label with a learning image to be input to the corrected model MDL. A learning image is a medical image generated when the transmission RF is not spatially uniform within the subject OB, and is a medical image including Tx unevenness, for example. A teacher image is a medical image generated under conditions in which the transmission RF is spatially and uniformly irradiated within the subject OB. This is a medical image with little unevenness. Medical images used as teacher images are, for example, medical images acquired by a low-field MRI system with a magnetic field strength of about 1.0 Tesla, or medical images obtained using an MP2RAGE (Magnetization Prepared 2 Rapid Acquisition Gradient Echoes) sequence. is.

The Rx unevenness is caused by causing the receiving coil 108 realized by the phased array coil to receive the magnetic resonance signal generated inside the subject OB when the transmitting coil 106 realized by the whole body coil is caused to transmit the transmission RF. This is unevenness in pixel values on a sensitivity map (hereinafter referred to as a reception sensitivity map) obtained when

A sensitivity map 236 is a sensitivity map that indicates the spatial distribution of Rx unevenness. The sensitivity map 236 is obtained by causing the receiving coil 108 realized by the phased array coil to receive the magnetic resonance signal, and the transmitting coil 106 realized by the whole body coil to receive the magnetic resonance signal. generated by dividing by the transmit sensitivity map obtained when The purpose of dividing by the transmission sensitivity map is to eliminate the influence of Tx unevenness from the reception sensitivity map. The sensitivity map 236 is an example of "receiver sensitivity distribution".

Returning to the description of FIG. 3 , the acquisition function 212 acquires medical images, ie MR images, from the MRI apparatus 100 via the communication interface 202 .

The correction function 214 uses the correction model MDL indicated by the correction model information 232 to correct the Tx unevenness caused by the transmitting coil 106 and the Rx unevenness caused by the receiving coil 108 from the MR image acquired by the acquiring function 212. Remove. For example, the correction model MDL may be implemented as part of the correction function 214 by the processor executing the correction model MDL.

The output control function 220 controls the communication interface 202 to transmit to the MRI apparatus 100 a corrected image from which the Tx unevenness and Rx unevenness have been removed by the correction function 214 . Also, the output control function 220 may cause the display 206 to display a corrected image from which the Tx unevenness and the Rx unevenness have been removed by the correction function 214 .

The learning function 222 inputs the learning image included in the teacher information 234 to the correction model MDL, and the image output by the correction model MDL is used as a teacher label in advance for the learning image input to the correction model MDL. The correction model MDL is learned so as to approach the associated teacher image.

For example, when the correction model MDL is a neural network, the learning function 222 performs correction so that the difference between the image output by the correction model MDL and the teacher image (for example, the difference in pixel value for each pixel) becomes small. The parameters of the model MDL are learned using stochastic gradient descent methods such as SGD (Stochastic Gradient Descent), Momentum SGD, AdaGrad, RMSprop, AdaDelta, and Adam (Adaptive moment estimation).

[Processing Flow of Medical Image Processing Apparatus at Runtime]
The flow of a series of processes of the processing circuit 210 on the side of the medical image processing apparatus 200 according to the embodiment will be described below with reference to the flowchart. FIG. 5 is a flow chart showing the flow of a series of processes during runtime of the processing circuit 210 on the side of the medical image processing apparatus 200 according to the embodiment. Runtime means executing processing using a well-trained correction model MDL. The processing of this flowchart may be performed repeatedly at a predetermined cycle, for example. Note that when the processor realizing the processing circuit 210 of the medical image processing apparatus 200 is configured by combining a plurality of independent circuits, that is, when the processing circuit 210 of the medical image processing apparatus 200 is realized by a plurality of processors. , a part or all of the processing of this flowchart may be processed in parallel by a plurality of processors.

First, the correction function 214 waits until the acquisition function 212 acquires the medical image from the MRI apparatus 100 via the communication interface 202 (step S100). is obtained, the medical image is input to the sufficiently learned correction model MDL (step S102).

Next, the correction function 214 causes the correction model MDL to which the medical image is input to output a corrected image in which Tx unevenness and Rx unevenness are reduced from the input medical image (step S104).

Next, the output control function 216 transmits the corrected image output by the correction model MDL to the MRI apparatus 100 via the communication interface 202 (step S106). In response, the display control function 138 of the processing circuit 130 of the MRI apparatus 100 causes the display 126 to display the corrected image received from the medical image processing apparatus 200 . Note that the output control function 216 may cause the display 206 to display the corrected image output by the corrected model MDL instead of or in addition to transmitting it to the MRI apparatus 100 .

The image correction processing by the correction model MDL at runtime will be described below with reference to the configuration diagram of the correction model MDL. FIG. 6 is a diagram showing an example of the configuration of the correction model MDL at runtime according to the embodiment. As shown in the illustrated example, the runtime correction model MDL according to the embodiment may include a first layer 310 , a second layer 320 , a third layer 330 , and a CNN (Convolutional Neural Network) 340 .

The first layer 310 uses the sensitivity map 236 to correct the brightness of the input medical image IMG_IN. This reduces the Rx unevenness contained in the input medical image IMG_IN. The first layer 310 outputs the luminance-corrected medical image IMG_IN, that is, the medical image IMG_IN with reduced Rx unevenness, to the second layer 320 and the third layer 330 .

The second layer 320 smoothes the medical image IMG_IN input from the first layer 310 to generate a medical image from which high frequency components are removed (hereinafter referred to as a low frequency component image IMG_A). For example, the second layer 320 may be an LPF (Low-Pass Filter). The second layer 320, which is the LPF, repeats the sum-of-products operation on the medical image IMG_IN, for example, while sliding a linear transformation matrix called a filter or kernel by a predetermined stride amount. The linear transformation matrix may be a box-shaped filter in which the weight coefficients (filter coefficients) are normalized to the same value for all elements (averaging), or a Gaussian filter that changes the weight coefficients according to the distance to a certain target pixel It may be a filter (Gaussian Filtering) or other filters such as a median filter or a bilateral filter. The second layer 320 may also perform morphological processing including various types of processing such as Erosion, Dilation, Opening, Closing, and Gradient on the medical image IMG_IN to smooth the medical image IMG_IN.

The 2nd layer 320 will output the produced|generated low frequency component image IMG_A to CNN340, if low frequency component image IMG_A is produced|generated. The low frequency component image IMG_A is an example of a "low frequency component image".

The third layer 330 sharpens the medical image IMG_IN input from the first layer 310 to generate a medical image from which low frequency components are removed (hereinafter referred to as high frequency component image IMG_B). For example, the third layer 330 may be a HPF (High-Pass Filter). The third layer 330, which is an HPF, outputs a difference image obtained by subtracting the low-frequency component image IMG_A generated by the second layer 320 from the medical image IMG_IN input from the first layer 310, as a high-frequency component image IMG_B. . The high frequency component image IMG_B is an example of a "high frequency component image".

CNN 340 includes, for example, an input layer 342 , multiple convolutional layers 344 , multiple activation layers 346 , and an output layer 348 .

The input layer 342 receives the low-frequency component image IMG_A from the second layer 320 . For example, when the low-frequency component image IMG_A is a matrix having an element corresponding to each pixel, the input layer 342 adds a bias component to the matrix corresponding to the low-frequency component image IMG_A as appropriate, and Output to convolutional layer 344 .

The convolution layer 344 repeats product-sum operations while sliding a linear transformation matrix called a filter or kernel with respect to the input matrix by a predetermined stride amount, and converts the input matrix to the linear transformation matrix. Generates a matrix containing multiple elements with associated sum of products as element values. At this time, the convolutional layer 344 performs padding (for example, zero padding) to interpolate elements of arbitrary values around the input matrix, and the matrix input to the convolutional layer 344 is input to the input layer 342. The matrix may be converted into a matrix having the same number of rows and columns as the matrix of the low-frequency component image IMG_A. Convolutional layer 344 then outputs the generated matrix to activation layer 346 .

The activation layer 346 calculates an activation function for each element of the matrix input from the convolution layer 344, and outputs the calculated matrix to the subsequent layer. The activation function of activation layer 346 may be, for example, a ReLU function.

The output layer 348 outputs the matrices processed by the previous convolutional layer 344 and activation layer 346 .

In this way, by inputting the low-frequency component image IMG_A to the CNN 340 in which the convolution layer 344 and the activation layer 346 are configured in multiple layers, a low-frequency component image (hereinafter referred to as , a corrected low-frequency component image IMG_A#) can be generated. The corrected low-frequency component image IMG_A# is an example of the "second corrected image".

Correction function 214 obtains the matrix output by output layer 348 of CNN 340 as corrected low frequency component image IMG_A#. Then, the correction function 214 adds the high-frequency component image IMG_B output by the third layer 330 and the corrected low-frequency component image IMG_A# output by the CNN 340 to obtain the image from the medical image IMG_IN input to the correction model MDL. Generate a corrected image IMG_GEN from which the Tx unevenness and the Rx unevenness are removed or reduced. The corrected image IMG_GEN is an example of a “corrected image”.

[Processing Flow of Medical Image Processing Apparatus During Training]
The flow of a series of processes of the processing circuit 210 on the side of the medical image processing apparatus 200 according to the embodiment will be described below with reference to the flowchart. FIG. 7 is a flow chart showing a series of processes during training of the processing circuit 210 of the medical image processing apparatus 200 according to the embodiment. Training is learning the correction model MDL. The processing of this flowchart may be performed repeatedly at a predetermined cycle, for example. Note that when the processor realizing the processing circuit 210 of the medical image processing apparatus 200 is configured by combining a plurality of independent circuits, that is, when the processing circuit 210 of the medical image processing apparatus 200 is realized by a plurality of processors. , a part or all of the processing of this flowchart may be processed in parallel by a plurality of processors.

First, the learning function 218 selects one learning image IMG_LR from a plurality of learning images IMG_LR included in the teacher information 234, and inputs the selected learning image IMG_LR to the correction model MDL (step S200). The learning image IMG_LR is an example of a "first medical image".

Next, the learning function 218 smoothed the corrected low-frequency component image IMG_A# output by the CNN 340 of the corrected model MDL to which the learning image IMG_LR was input, and the teacher image IMG_TR associated with the learning image IMG_LR as a teacher label. A difference from an image (hereinafter referred to as a low-frequency component teacher image IMG_TR#) is derived (step S202). The teacher image IMG_TR is an example of a "second medical image".

Next, the learning function 218 uses the element values of the linear transformation matrix of the convolution layer 344 and the Various parameters such as activation function coefficients and bias components are determined by learning using the stochastic gradient descent method or the like (step S204).

Next, the learning function 218 selects all the learning images IMG_LR included in the teacher information 234 to determine whether or not the correction model MDL has been learned, and selects all the learning images IMG_LR to learn the correction model MDL. If it is determined that it has not, the process returns to S200, and a learning image IMG_LR different from the previously selected learning image IMG_LR is selected.

When the learning function 218 determines that all the learning images IMG_LR have been selected and the correction model MDL has been learned, the processing of this flowchart ends.

FIG. 8 is a diagram showing an example of the configuration of the correction model MDL during training according to the embodiment. As in the illustrated example, during training, the corrected model MDL includes the above-described first layer 310, second layer 320, third layer 330, and CNN 340, as well as a fourth layer 360 and a fifth layer 370. including.

During training, the first layer 310 uses the sensitivity map 236 to correct the brightness of the input learning image IMG_LR.

The second layer 320 smoothes the learning image IMG_LR luminance-corrected by the first layer 310 to generate a low-frequency component image IMG_A.

The third layer 330 sharpens the learning image IMG_LR luminance-corrected by the first layer 310 to generate a high-frequency component image IMG_B.

CNN 340 removes or reduces Tx unevenness from low frequency component image IMG_A generated by second layer 320 to generate corrected low frequency component image IMG_A#.

A teacher image IMG_TR associated with the learning image IMG_LR is input to the fourth layer 360 . When the teacher image IMG_TR is input, the fourth layer 360 uses the sensitivity map 236 to correct the brightness of the input teacher image IMG_TR.

The fifth layer 370 smoothes the teacher image IMG_TR whose brightness has been corrected by the fourth layer 360, and generates a low-frequency component teacher image IMG_TR# by removing high-frequency components from the teacher image IMG_TR. For example, the fifth layer 370, like the second layer 320, can be an LPF.

The learning function 218 obtains the difference between the low-frequency component teacher image IMG_TR# generated by the fifth layer 370 and the corrected low-frequency component image IMG_A# generated by the CNN 340, and stochastically The parameters of the CNN 340 are learned using error backpropagation based on gradient descent or the like.

According to the embodiment described above, the processing circuit 210 of the medical image processing apparatus 200 acquires a medical image generated by transmitting the transmission RF from the transmission coil 106 toward the subject OB. When an image is input, the processing circuit 210 performs the correction model MDL learned to output a corrected image in which image unevenness caused by the non-uniformity of the high-frequency magnetic field is removed or reduced from the input image. By inputting the acquired medical image, a corrected image is generated by removing or reducing image unevenness from the acquired medical image. As a result, the image quality of medical images can be improved.

Further, according to the above-described embodiment, by using the low-frequency component teacher image IMG_TR# obtained by removing the high-frequency component from the teacher image IMG_TR, the correction model MDL can be learned with high accuracy regardless of the resolution of the MRI apparatus 100. can do. Further, by using the low-frequency component teacher image IMG_TR# obtained by removing the high-frequency component from the teacher image IMG_TR, the subject OB whose medical image to be the teacher image and the learning image IMG_LR to be input to the CNN 340 are different from each other can be scanned. Even with the obtained medical image, the correction model MDL can be learned with high accuracy.

In general, when the subject OB is different, there is an individual difference in the structure of the part even if the part to be scanned is the same. Therefore, unevenness that can be included in a medical image also fluctuates according to individual differences. is preferably However, when both the teacher image IMG_TR and the learning image IMG_LR to be input to the correction model MDL are obtained from the same object OB, the same object OB is scanned many times in order to prepare many teacher images IMG_TR. As a result, the burden on the subject OB tends to increase. In addition, since the same medical image of the subject OB is used for learning the correction model MDL, the generalization of the correction model MDL tends to deteriorate.

In contrast, in the present embodiment, even if the teacher image IMG_TR and the learning image IMG_LR input to the CNN 340 are medical images obtained by scanning different subjects OB, both images are smoothed. Transformation processing is performed to remove high-frequency components related to detailed structures such as edges. Generalizability can be improved.

(Modification of embodiment)
Modifications of the embodiment will be described below. In the above-described embodiment, the MRI apparatus 100 and the medical image processing apparatus 200 are different apparatuses, but the present invention is not limited to this. For example, the medical image processing apparatus 200 may be realized by one function of the console device 120 of the MRI apparatus 100. FIG. That is, the medical image processing apparatus 200 may be a virtual machine virtually implemented by the console device 120 of the MRI apparatus 100 .

FIG. 9 is a diagram showing another example of the MRI apparatus 100 according to the embodiment. As shown in FIG. 9, the processing circuit 130 of the console device 120 includes the above-described acquisition function 132, generation function 134, communication control function 136, and display control function 138, as well as a correction function 214 and a learning function 218. and

Also, the memory 150 of the console device 120 may store correction model information 232 , teacher information 234 , and a sensitivity map 236 .

With such a configuration, the MRI apparatus 100 alone can improve the image quality of medical images.

Further, in the above-described embodiment, the correction model MDL is learned based on the low-frequency component teacher image IMG_TR# obtained by removing the high-frequency component from the teacher image IMG_TR, but the present invention is not limited to this. For example, the learning function 218 is based on a low-frequency component teacher image IMG_TR# obtained by removing high-frequency components from the teacher image IMG_TR, a sequence parameter, a statistical value indicating the degree of unevenness contained in the medical image, a B1 map, and the like. , may learn the correction model MDL.

The sequence parameters are parameters dependent on the transmission RF, such as the type of sequence in the main scan, the type of the transmission coil 106, the calibration value of the transmission RF, and the adjustment values such as the phase and amplitude of the transmission RF in B1 shimming. .

Statistical values indicating the degree of unevenness include, for example, standard deviation, RMS value, PP value, and the like.

The B1 map includes, for example, a sensitivity map based on the signal strength of magnetic resonance signals obtained by a sequence in which the transmission RF flip angle is α degrees (eg, 45 degrees), and a transmission RF flip angle of 2α degrees (eg, 90 degrees). ) is obtained by substituting a sensitivity map based on the signal intensity of the magnetic resonance signal obtained by the sequence of ) into a certain theoretical formula.

In this way, in addition to the teacher image IMG_TR, by learning the correction model MDL using the sequence parameter, statistical values indicating the degree of unevenness, the B1 map, etc., the Tx unevenness that can be included in the low-frequency component image can be accurately corrected. can be removed. In particular, when the correction model MDL is learned using the B1 map, if there is a region with locally low or high signal intensity on the input image IMG_IN at run time, it is possible to determine whether the region is a change in luminance due to a lesion. , or the change in luminance due to Tx unevenness can be identified by the correction model MDL.

For example, the learning function 218 may determine, for a region, if the B1 value is significantly greater than a reference value (e.g., 1.0) on the B1 map and the brightness is high on the input image IMG_IN, or if the B1 map If the B1 value is significantly less than the reference value and the luminance is low on the input image IMG_IN, the correction model MDL is trained to identify the area as a blocky area.

On the other hand, the learning function 218 determines that a region has a lesion when the B1 value on the B1 map is approximately the same as the reference value and the luminance is high or low on the input image IMG_IN. Train the correction model MDL to identify that there is. By learning the correction model MDL using the B1 map in this way, it is possible to appropriately correct the brightness of a local region having different brightness compared to other regions as Tx unevenness, so the probability of the presence of a lesion can be reduced. A region with high Tx can be left as it is on the image without being removed as Tx unevenness.

The embodiment described above can be expressed as follows.
a storage for storing programs;
a processor;
By executing the program, the processor
Acquiring a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil,
When an image is input, inputting the acquired medical image to a model that has been trained to output a corrected image that reduces the influence of the inhomogeneity of the high-frequency magnetic field from the input image, generating a corrected image that reduces the effects of the non-uniformity from the acquired medical image;
A medical image processing apparatus configured as follows.

According to at least one embodiment described above, the processing circuit 210 of the medical image processing apparatus 200 acquires a medical image generated by transmitting a transmission RF from the transmission coil 106 toward the subject OB, When an image is input, the acquired medical image is transferred to the correction model MDL that has been trained to output a corrected image in which image unevenness caused by the non-uniformity of the high-frequency magnetic field is removed or reduced from the input image. By inputting, a corrected image is generated by removing or reducing image unevenness from the acquired medical image, so that the image quality of the medical image can be improved.

While several embodiments of the invention have been described, these embodiments have been 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, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and spirit of the invention, as well as the scope of the invention described in the claims and equivalents thereof.

Reference Signs List 1 medical image processing system 100 MRI apparatus 101 static magnetic field magnet 102 gradient magnetic field coil 104 bed 105 bed control circuit 106 transmission coil 107 transmission circuit 108 reception coil 109 Reception circuit 110 Sequence control circuit 120 Console device 200 Medical image processing device 202 Communication interface 204 Input interface 206 Display 210 Processing circuit 212 Acquisition function 214 Correction function , 216... output control function, 218... learning function, 230... memory

Claims (10)

an acquisition unit that acquires a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil;
When an image is input, the medical image acquired by the acquisition unit is applied to a model that has been trained to output a corrected image in which the influence of the non-uniformity of the high-frequency magnetic field is reduced from the input image. a generation unit that generates a corrected image in which the influence of the non-uniformity is reduced from the medical image acquired by the acquisition unit , by inputting
The image input to the model is a medical image whose brightness has been corrected using the reception sensitivity distribution of the magnetic resonance signal in the RF coil.
Medical image processing equipment. The generating unit
generating a high-frequency component image obtained by extracting high-frequency components and a low-frequency component image obtained by extracting low-frequency components from the medical image obtained by the obtaining unit;
inputting the low-frequency component image to the model to generate a second corrected image in which the influence of the non-uniformity is reduced from the low-frequency component image;
generating a corrected image in which the influence of the non-uniformity is reduced from the medical image acquired by the acquisition unit, based on the second corrected image and the high-frequency component image ;
The medical image processing apparatus according to claim 1 . The low-frequency component image of the low-frequency component of the second medical image generated when the high-frequency magnetic field is uniform is associated with the first medical image generated when the high-frequency magnetic field is non-uniform. Further comprising a learning unit that learns the model based on the information obtained,
The medical image processing apparatus according to claim 2 . an acquisition unit that acquires a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil;
When an image is input, the medical image acquired by the acquisition unit is applied to a model that has been trained to output a corrected image in which the influence of the non-uniformity of the high-frequency magnetic field is reduced from the input image. a generation unit that generates a corrected image in which the influence of the non-uniformity is reduced from the medical image acquired by the acquisition unit, by inputting
The generating unit
generating a high-frequency component image obtained by extracting high-frequency components and a low-frequency component image obtained by extracting low-frequency components from the medical image obtained by the obtaining unit;
inputting the low-frequency component image to the model to generate a second corrected image in which the influence of the non-uniformity is reduced from the low-frequency component image;
generating a corrected image in which the influence of the non-uniformity is reduced from the medical image acquired by the acquisition unit, based on the second corrected image and the high-frequency component image;
Medical image processing equipment. a first generator that generates a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil;
When an image is input, the medical image generated by the first generation unit is applied to a model trained to output a corrected image in which the influence of non-uniformity of a high-frequency magnetic field is reduced from the input image. a second generating unit that generates a corrected image in which the influence of the non-uniformity is reduced from the medical image generated by the first generating unit by inputting an image ;
The image input to the model is a medical image whose brightness has been corrected using the reception sensitivity distribution of the magnetic resonance signal in the RF coil.
Magnetic resonance imaging equipment. a first generator that generates a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil;
When an image is input, the medical image generated by the first generation unit is applied to a model trained to output a corrected image in which the influence of non-uniformity of a high-frequency magnetic field is reduced from the input image. a second generation unit that generates a corrected image in which the influence of the non-uniformity is reduced from the medical image generated by the first generation unit by inputting an image;
The second generator,
generating a high-frequency component image obtained by extracting high-frequency components and a low-frequency component image obtained by extracting low-frequency components from the medical image generated by the first generating unit;
inputting the low-frequency component image to the model to generate a second corrected image in which the influence of the non-uniformity is reduced from the low-frequency component image;
generating a corrected image in which the influence of the non-uniformity is reduced from the medical image generated by the first generating unit, based on the second corrected image and the high-frequency component image;
Magnetic resonance imaging equipment. the computer
Acquiring a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil,
When an image is input, inputting the acquired medical image to a model that has been trained to output a corrected image that reduces the influence of the inhomogeneity of the high-frequency magnetic field from the input image, generating a corrected image in which the influence of the non-uniformity is reduced from the acquired medical image;
The image input to the model is a medical image whose brightness has been corrected using the reception sensitivity distribution of the magnetic resonance signal in the RF coil.
Medical image processing method. the computer
Acquiring a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil,
When an image is input, inputting the acquired medical image to a model that has been trained to output a corrected image that reduces the influence of the inhomogeneity of the high-frequency magnetic field from the input image, generating a corrected image in which the influence of the non-uniformity is reduced from the acquired medical image;
generating a high-frequency component image obtained by extracting high-frequency components and a low-frequency component image obtained by extracting low-frequency components from the acquired medical image;
inputting the low-frequency component image to the model to generate a second corrected image in which the influence of the non-uniformity is reduced from the low-frequency component image;
generating a corrected image in which the influence of the non-uniformity is reduced from the acquired medical image based on the second corrected image and the high-frequency component image;
Medical image processing method. A program for a computer to execute ,
A process of acquiring a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil;
When an image is input, inputting the acquired medical image to a model that has been trained to output a corrected image that reduces the influence of the inhomogeneity of the high-frequency magnetic field from the input image, a process of generating a corrected image from the acquired medical image that reduces the effect of the non-uniformity ,
The image input to the model is a medical image whose brightness has been corrected using the reception sensitivity distribution of the magnetic resonance signal in the RF coil.
program . A program for a computer to execute,
A process of acquiring a medical image based on a magnetic resonance signal obtained by exciting a subject with a high-frequency magnetic field transmitted by an RF coil;
When an image is input, inputting the acquired medical image to a model that has been trained to output a corrected image that reduces the influence of the inhomogeneity of the high-frequency magnetic field from the input image, and generating a corrected image from the acquired medical image that reduces the effect of the non-uniformity,
a process of generating a high-frequency component image obtained by extracting high-frequency components and a low-frequency component image obtained by extracting low-frequency components from the acquired medical image;
A process of generating a second corrected image in which the influence of the non-uniformity is reduced from the low-frequency component image by inputting the low-frequency component image to the model;
a process of generating a corrected image in which the influence of the non-uniformity is reduced from the acquired medical image based on the second corrected image and the high-frequency component image;
program.

Images