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

Abstract

In a medical image processing device that performs image processing to generate medical images using complex images captured of a subject, the value of pixels with low signal intensity is corrected to a state suitable for image processing. [Solution] The medical image processing apparatus of the embodiment is a medical image processing apparatus that generates a medical image based on a first complex image of a subject, and has an image generation unit. The image generation unit generates a second complex image by removing the background phase component from the first complex image, and uses the noise contained in the second complex image to correct the pixel value of the first pixel in each pixel constituting the second complex image in which the pixel value of the signal of interest representing the image of the subject is "0" or less, thereby generating the medical image.

Description

[Technical Field]

[0001] Embodiments of the present invention relate to a medical image processing apparatus, a medical image processing method, and a program. [Background technology]

[0002] Conventionally, diffusion-weighted imaging (DWI) acquired using medical imaging diagnostic devices such as magnetic resonance imaging (MRI) machines has been used for the diagnosis of patients (subjects). However, it is known that in diffusion-weighted imaging, different phase fluctuations occur from image to image due to factors such as motion, respiration, pulsation, perfusion, and magnetic field inhomogeneity during the diffusion encoding image processing. Furthermore, when generating diffusion-weighted images, image processing is performed by adding and averaging multiple complex images acquired repeatedly in order to improve the signal-to-noise ratio (S / N ratio). However, since each complex image used to generate the diffusion-weighted image has the different phase fluctuations mentioned above, simply adding the complex images together results in a medical image unsuitable for diagnosis. For this reason, conventionally, each complex image used to generate the diffusion-weighted image is converted to a magnitude image, and then the respective magnitude images are added and averaged to generate the diffusion-weighted image. Incidentally, complex images have a real component and an imaginary component. Therefore, in the calculation to convert a complex image to a magnitude image, the sum of squares (in other words, the calculation of absolute values) is performed for both the real and imaginary components of the complex image. At this time, Gaussian noise (noise containing both positive and negative values) that follows a Gaussian distribution and can be considered to be "0" on average, contained in both the real and imaginary components of the complex image, is converted by the magnitude transformation calculation into Lysian noise (noise containing only positive values) that follows a Rice distribution. Therefore, when multiple magnitude images that have undergone magnitude transformation are added together to generate a diffusion-weighted image, the Lysian noise in each magnitude image becomes noise that generates a noise floor, causing a decrease in the contrast of the diffusion-weighted image. In other words, if the noise contained in the magnitude images remains Gaussian noise, the noise in the positive and negative values ​​will cancel each other out by adding the magnitude images together, and the level of noise contained in the generated diffusion-weighted image will decrease.In contrast, if the noise in the magnitude image is lysian noise, adding the magnitude image will increase the noise level to a positive value (the noise level will approach the level of the signal of interest representing the subject's image), resulting in a decrease in the contrast between light and dark areas across the entire area of ​​the generated diffusion-weighted image. As a result, the generated diffusion-weighted image may become, for example, a blurred image that is generally white, making it unsuitable for diagnosis.

[0003] Therefore, it has been proposed to add the real part image generated from the real part component of a complex image obtained by removing the background phase. Conventional background phase removal techniques estimate the background phase from the complex image and remove only the estimated phase component from the complex image's phase. As a result, the real part component of the complex image contains both the signal of interest, which is a true signal and not noise, and Gaussian noise, while the imaginary part component generated from the imaginary part of the complex image contains only Gaussian noise. Therefore, the imaginary part component of the complex image can be discarded, that is, it can be removed from the complex image and not used when generating a diffusion-weighted image. This makes it possible to simply average the real part component of the complex image obtained by applying the background phase removal technique, and solves the problem of averaging after performing the magnitude transformation calculation mentioned above when generating a diffusion-weighted image (the problem that Gaussian noise is converted into Lysian noise, which causes a decrease in contrast).

[0004] However, in the real part image components of a complex image obtained by applying conventional background phase removal techniques, the level of Gaussian noise is high, and in pixels where the signal intensity (level) of the signal of interest is low, the signal level may fall below the noise level, resulting in the value of the signal of interest (i.e., the pixel value) being "0" or less (pixel value = "0" or negative pixel value). Images containing pixel values ​​of "0" or less may cause problems in subsequent image processing, such as the inability to generate a suitable image (in this case, a diffusion-weighted image). For this reason, conventional proposals, including those concerning real part addition by background phase removal, suggest excluding pixel values ​​of "0" or less from image processing, or correcting (replacing) them with a value greater than "0" before image processing. However, conventional proposals do not specify what value to correct pixel values ​​of "0" or less to. Therefore, if all pixel values ​​of "0" or less are simply corrected to a predetermined value greater than "0" and image processing is performed, the resulting image will be unnatural. In image processing that generates diffusion-weighted images, such as Isotropic Diffusion Weighted Images (IsoDWI), pixels will have small values ​​only along a specific MPG (Motion Probing Gradient) axis, resulting in an image with unnaturally missing pixels. Similarly, in image processing that generates diffusion tensor images, such as diffusion-weighted images, logarithmic calculations are used in the process. This results in abnormal values ​​for pixel values ​​less than or equal to "0," which are far removed from the values ​​of surrounding pixels, leading to an unnatural image. [Prior art documents] [Patent Documents]

[0005] [Patent Document 1] U.S. Patent No. 10605882 [Non-patent literature]

[0006] [Non-Patent Document 1] Hakon Gudbjartsson, Samuel Patz, “The Rician Distribution of Noisy MRI Data”, HHS Public Access, Author manuscript, Magn Reson Med. Author manuscript; available in PMC 2008 February 25. [Non-Patent Document 2] Douglas E. Prah, Eric S. Paulson, Andrew S. Nencka, Kathleen M. Schmainda, “A Simple Method for Rectified Noise Floor Suppression: Phase-Corrected Real Data Reconstruction With Application to Diffusion-Weighted Imaging”, Magnetic Resonance in Medicine 64:418-429 (2010) [Non-Patent Document 3] Cornelius Eichner, Stephen F Cauley, Julien Cohen-Adad, Harald E Moller, Robert Turner, Kawin Setsompop, Lawrence L Wald, “Real Diffusion-Weighted MRI Enabling True Signal Averaging and Increased Diffusion Contrast”, HHS Public Access, Author manuscript, Neuroimage. Author manuscript; available in PMC 2016 November 15. [Overview of the project] [Problems that the invention aims to solve]

[0007] The problem that the embodiments disclosed herein and in the drawings aim to solve is to correct the values ​​of pixels with low signal intensity in a medical image processing apparatus that performs image processing to generate a medical image using a complex image captured of a subject, so that the values ​​of pixels with low signal intensity become suitable for image processing. However, the problem that the embodiments disclosed herein and in the drawings aim to solve is not limited to the above problem. Problems corresponding to the effects of each configuration shown in the embodiments described later can also be positioned as other problems. [Means for solving the problem]

[0008] The medical image processing apparatus of the embodiment is a medical image processing apparatus that generates a medical image based on a first complex image of a subject, and has an image generation unit. The image generation unit generates a second complex image by removing the background phase component from the first complex image, and uses the noise contained in the second complex image to correct the pixel value of the first pixel in each pixel constituting the second complex image, such that the pixel value of the signal of interest representing the image of the subject is "0" or less, thereby generating the medical image. [Brief explanation of the drawing]

[0009] [Figure 1] A diagram showing an example of the configuration of a medical image diagnostic device equipped with a medical image processing device according to the embodiment. [Figure 2] A diagram showing an example of the functional configuration of a medical image processing apparatus according to the embodiment. [Figure 3] A schematic diagram showing an example of an image from which the background phase of a complex image has been removed in a medical image processing device according to the embodiment. [Figure 4] This figure schematically illustrates an example of an image in which the background phase of a complex image has been removed by conventional magnitude transformation calculations. [Figure 5] A figure showing an example of an image generated by the medical image processing apparatus according to the embodiment. [Figure 6] A flowchart showing an example of the process flow for generating a medical image in a medical image processing apparatus according to the embodiment.

Best Mode for Carrying Out the Invention

[0010] Hereinafter, a medical image processing apparatus, a medical image processing method, and a program according to an embodiment will be described with reference to the drawings. In the following description, the case where the medical image processing apparatus according to the embodiment is provided in a Magnetic Resonance Imaging (MRI) apparatus (hereinafter referred to as "MRI apparatus") will be described as an example.

[0011] The MRI apparatus irradiates an RF (Radio Frequency) pulse while applying a strong magnetic field to a subject (e.g., a human body), receives electromagnetic waves generated from hydrogen atomic nuclei in the body of the subject by the RF coil due to the nuclear magnetic resonance phenomenon, and reconstructs a tomographic image (hereinafter referred to as "MR image") of the subject from a nuclear magnetic resonance signal (hereinafter referred to as "MR signal") based on the received electromagnetic waves. The MRI apparatus can also capture an MR image of the subject by reconstructing an MR signal based on electromagnetic waves received by an RF coil attached to the subject. By displaying the MR image of the subject, an operator (such as a doctor or a technician) of the MRI examination can visually confirm whether there is a lesion in the subject or not.

[0012] FIG. 1 is a diagram showing an example of the configuration of a medical image diagnostic apparatus (MRI apparatus) including the medical image processing apparatus according to the embodiment. The MRI apparatus 1 includes, for example, a gantry apparatus 10, a bed apparatus 20, a control apparatus 30, and a console apparatus 40. In the present embodiment, the control apparatus 30 and the console apparatus 40 will be described as being separate from the gantry apparatus 10, but a part or all of the components of the control apparatus 30 and the console apparatus 40 may be included in the gantry apparatus 10.

[0013] The gantry device 10 includes, for example, a static magnetic field magnet 12, a gradient magnetic field coil 14, and an RF coil 16. Further, in the gantry device 10, as a part of the components of the RF coil 16, for example, an RF coil 17 that can be attached to the subject P is provided.

[0014] The static magnetic field magnet 12 is a magnet formed in a hollow substantially cylindrical shape. The static magnetic field magnet 12 generates a uniform static magnetic field in the internal space. The static magnetic field magnet 12 is, for example, a permanent magnet or a superconducting magnet. When the static magnetic field magnet 12 is a superconducting magnet, it receives power supply from a static magnetic field power supply (not shown) to generate a static magnetic field.

[0015] The gradient magnetic field coil 14 is a coil formed in a hollow substantially cylindrical shape. The gradient magnetic field coil 14 is disposed inside the static magnetic field magnet 12. The gradient magnetic field coil 14 is formed by combining three coils corresponding to the X-axis, Y-axis, and Z-axis that are orthogonal to each other. Each of the three coils corresponding to the directions of each axis receives current supply individually from a gradient magnetic field power supply 32, and in the imaging space (i.e., inside the bore) of the MRI apparatus 1 into which the subject P is introduced, a gradient magnetic field whose magnetic field strength changes along the X-axis, Y-axis, and Z-axis is generated. In the present embodiment, the central axis of the gantry device 10 or the longitudinal direction of the top plate 24 of the bed device 20 is defined as the Y-axis direction, an axis that is orthogonal to the Y-axis direction and horizontal with respect to the floor surface of the room where the MRI apparatus 1 is installed is defined as the X-axis direction, and a direction that is orthogonal to the Y-axis direction and perpendicular to the floor surface is defined as the Z-axis direction, respectively. And in the present embodiment, the Y-axis direction is the same direction as the static magnetic field.

[0016] Here, the gradient magnetic fields along the X-axis, Y-axis, and Z-axis generated by the gradient magnetic field coil 14 respectively correspond to, for example, a slice selection gradient magnetic field, a phase encoding gradient magnetic field, and a readout gradient magnetic field. The slice selection gradient magnetic field is used to determine an arbitrary imaging section in the MRI apparatus 1. The phase encoding gradient magnetic field is used to change the phase of the MR signal according to the spatial position in the MRI apparatus 1. The readout gradient magnetic field is used to change the frequency of the MR signal according to the spatial position in the MRI apparatus 1.

[0017] The RF coil 16 is a whole-body coil housed within the mounting device 10 and configured to surround the subject P in the imaging space. The RF coil 16 includes a transmitting coil that receives RF pulses from the transmitting circuit 33 to generate a high-frequency magnetic field, and a receiving coil that receives the MR signal emitted from the subject P due to the influence of the high-frequency magnetic field. When the receiving coil of the RF coil 16 receives the MR signal, it outputs the received MR signal to the receiving circuit 34. The RF coil 16 may be composed of different coils for the transmitting and receiving coils, or it may be composed of the same coil, that is, a coil that can be used for both transmitting and receiving. In this case, the RF coil 16 may be, for example, a birdcage coil.

[0018] The RF coil 17 is a local coil attached to the subject P. The RF coil 17 comes in various shapes depending on the area of ​​the subject P being imaged (hereinafter referred to as the "imaging area"). Figure 1 shows an example of an RF coil 17 attached to the torso of the subject P. The RF coil 17 receives the MR signal emitted from the subject P due to the influence of the high-frequency magnetic field generated by the RF coil 16. Upon receiving the MR signal, the RF coil 17 outputs the received MR signal to the receiving circuit 34. The RF coil 17 may be a wired type that outputs the received MR signal to the receiving circuit 34 via a wire, or a wireless type that outputs (transmits) the received MR signal to the receiving circuit 34 wirelessly. Figure 1 shows, for example, an example of a wired RF coil 17 that outputs the received MR signal to the receiving circuit 34 via a coil cable connected to a coil port (not shown) located on the top plate 24. The RF coil 17 may be, for example, a coil array composed of multiple coil elements. The function of the RF coil 17 may also be realized by the RF coil 16.

[0019] The patient bed device 20 is a device that introduces the subject P to be imaged into the interior of the rigging device 10, that is, into the bore of the rigging device 10, by moving the top plate 24 on which the subject P to be imaged is placed. In other words, the patient bed device 20 is a device that moves the top plate 24 so that the imaging area of ​​the subject P is in a position suitable for imaging within the magnetic field generated in the cavity of the static magnetic field magnet 12, gradient magnetic field coil 14, and RF coil 16, that is, within the imaging opening. The patient bed device 20 comprises, for example, a base 22 and a top plate 24.

[0020] The base 22 moves the tabletop 24 on which the subject P is placed horizontally (in the X-axis and Y-axis directions) or vertically (in the Z-axis direction) by the operation of a table drive device (not shown) which operates in response to a control signal output by the table control circuit 35. The base 22 includes a housing that movably supports the tabletop 24. The table drive device (not shown) includes, for example, a motor or actuator. The table drive device (not shown) may move not only the tabletop 24 but also the base 22 itself in the longitudinal direction (Y-axis direction) of the tabletop 24. If the support device 10 is configured to move in the Y-axis direction, the table drive device (not shown) may also operate to move the support device 10 so that the subject P is introduced into the support device 10. If the bed drive device (not shown) is configured such that both the frame device 10 and the top plate 24 and base 22 are movable, it may operate by moving the frame device 10, the top plate 24, and the base 22 respectively so that the subject P is introduced into the frame device 10.

[0021] The top plate 24 is a plate-shaped component on which the subject P is placed. The top plate 24 is made of a material with low conductivity (less affected by magnetic fields), such as glass fiber.

[0022] The control device 30 controls the operation of the frame device 10 and the bed device 20 in response to control from the console device 40. The control device 30 includes, for example, a sequence control circuit 31, a gradient magnetic field power supply 32, a transmitting circuit 33, a receiving circuit 34, and a bed control circuit 35. The control device 30 may be located within the frame device 10 or within the console device 40.

[0023] The sequence control circuit 31 is a sequencer that performs imaging of the subject P by driving the gradient power supply 32, the transmitting circuit 33, and the receiving circuit 34 based on sequence information set by the console device 40. The sequence control circuit 31 may be a processing circuit having a processor such as a CPU (Central Processing Unit). The sequence information is information that defines the procedure for performing imaging processing to image the subject P in the MRI device 1. The sequence information is defined in advance for each imaging process performed in the MRI device 1. For example, the sequence information shows the operation and timing of the operation of the gradient power supply 32, the transmitting circuit 33, and the receiving circuit 34 when imaging the subject P in chronological order (hereinafter referred to as "events"). More specifically, the sequence information indicates events such as the magnitude and timing of the current supplied to the gradient coil 14 by the gradient power supply 32, the strength and timing of the RF pulses transmitted (supplied) to the RF coil 16 by the transmitting circuit 33, and the timing of the RF pulse supply to the receiving circuit 34 to receive (detect) the MR signals output by the RF coil 16 and RF coil 17. The sequence control circuit 31 drives the gradient power supply 32, the transmitting circuit 33, and the receiving circuit 34 by sequentially executing the events indicated in the sequence information at timings based on a predetermined clock signal. When the receiving circuit 34 receives an MR signal, it transfers the received MR signal (more specifically, data representing the MR signal received by the receiving circuit 34 (hereinafter referred to as "MR data")) to the console device 40. The clock signal is generated, for example, by a clock generation circuit (not shown) including a clock oscillator, and represents the timing used as a reference for the operation of imaging the subject P in the MRI device 1. The clock signal is supplied to each component of the control device 30. The sequence control circuit 31 executes events in sequence based on the timing of the clock signal, causing the gradient power supply 32, the transmitting circuit 33, and the receiving circuit 34 to operate in synchronous manner.

[0024] The gradient power supply 32 supplies current individually to each of the three coils in the gradient coil 14, corresponding to the direction of each axis.

[0025] The transmitting circuit 33 supplies RF pulses to the RF coil 16. The RF pulses supplied by the transmitting circuit 33 to the RF coil 16 are pulses corresponding to the Larmor frequency, which is determined by the type of atomic nucleus being targeted and the strength of the magnetic field.

[0026] The receiving circuit 34 detects the MR signal output by the RF coil 16 and RF coil 17 and generates MR data representing the detected MR signal. The receiving circuit 34 generates the MR data, for example, by converting the MR signal into digital data. The receiving circuit 34 outputs the generated MR data to the sequence control circuit 31. The sequence control circuit 31 transfers the MR data output by the receiving circuit 34 to the console device 40.

[0027] The bed control circuit 35 outputs a control signal to a bed drive device (not shown) provided in the bed device 20 that moves the base 22 and the top plate 24 on which the subject P is placed, in response to control from the console device 40. The bed control circuit 35 may be provided in the pedestal device 10 or in the bed device 20. In this case, the bed control circuit 35 outputs a control signal to a bed drive device (not shown) provided in the bed device 20 that corresponds to an input signal input from an input interface (not shown) provided in the device in which the bed control circuit 35 is provided, when the operator of the MRI device 1, such as a doctor or technician, or the person performing the MRI examination (hereinafter referred to as "the person performing the MRI examination") operates the input interface (not shown) provided in the device in which the bed control circuit 35 is provided, in response to an input signal input from the input interface (not shown).

[0028] The console device 40 controls the entire MRI device 1 and collects MR data. The console device 40 includes, for example, a memory 41, a display 42, an input interface 43, and a processing circuit 50.

[0029] Memory 41 can be implemented using semiconductor memory elements such as ROM (Read Only Memory), RAM (Random Access Memory), or flash memory, or by a hard disk drive (HDD), optical disc, etc. Memory 41 stores data such as MR data output by the sequence control circuit 31 and reconstructed images (MR images) generated based on the MR data. This data may be stored in an external memory that the MRI device 1 can communicate with, rather than in memory 41 (or in addition to memory 41). The external memory may be a NAS (Network Attached Storage) or a cloud server that manages the external memory and accepts read / write requests, and is controlled by the cloud server. The external memory can be implemented using a system called PACS (Picture Archiving and Communication Systems). PACS is a medical image management system that systematically stores medical images acquired by various medical imaging diagnostic devices.

[0030] The display 42 displays various types of information. For example, the display 42 displays medical images generated by the processing circuit 50, or GUI (Graphical User Interface) images that accept various operations from the person performing the MRI examination. The display 42 may be, for example, a liquid crystal display (LCD), a CRT (Cathode Ray Tube) display, or an organic EL (Electroluminescence) display. The display 42 may be mounted on the stand device 10. The display 42 may be a desktop type, or it may be a display device (for example, a tablet terminal) that can communicate wirelessly with the main unit of the console device 40.

[0031] The input interface 43 receives various input operations from the MRI examiner and outputs an electrical signal indicating the content of the received input operation to the processing circuit 50. For example, the input interface 43 receives input operations such as acquisition conditions when acquiring MR data, generation conditions when generating MR data, reconstruction conditions when reconstructing reconstructed images, and image processing conditions when generating post-processed images from reconstructed images. The input interface 43 can be implemented by, for example, a mouse, keyboard, touch panel, trackball, switch, button, joystick, camera, infrared sensor, microphone, etc. If the input interface 43 is a touch panel, the display 42 may be formed integrally with the input interface 43. The input interface 43 may be provided on the rigging device 10. The input interface 43 may also be implemented by a display device (e.g., a tablet terminal) that can communicate wirelessly with the main body of the console device 40. In this specification, the input interface 43 is not limited to those equipped with physical operating components such as the mouse or keyboard described above. For example, an electrical signal processing circuit that receives an electrical signal corresponding to an input operation from an external input device provided separately from the console device 40 and outputs this electrical signal to the processing circuit 50 is also an example of an input interface 43.

[0032] The processing circuit 50 controls the overall operation of the MRI device 1. The processing circuit 50 sets sequence information in the sequence control circuit 31. For example, when acquiring a diffusion-weighted image (DWI) of a subject P in the MRI device 1, the processing circuit 50 sets sequence information corresponding to the diffusion-weighted image acquisition process in the sequence control circuit 31. The processing circuit 50 performs functions such as acquisition function 51, reconstruction function 52, image processing function 53, and output control function 54. The processing circuit 50 realizes these functions, for example, by having a hardware processor in a computer device execute a program (software) stored in a memory 41, which is a memory device (storage circuit).

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

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

[0035] The acquisition function 51 acquires the MR data transferred by the sequence control circuit 31. The MR data is obtained by converting the MR signal into digital data by the receiving circuit 34. The acquisition function 51 stores the acquired MR data in the memory 41.

[0036] The reconstruction processing function 52 performs a predetermined reconstruction process on the MR data acquired by the acquisition function 51 (MR data stored in memory 41) to generate a reconstructed image. For example, the reconstruction processing function 52 arranges the MR data in two or three dimensions corresponding to the gradient magnetic field for slice selection, the gradient magnetic field for phase encoding, and the gradient magnetic field for readout, and then performs a reconstruction process using Fourier transform or the like to generate a reconstructed image. The reconstruction processing function 52 stores the generated reconstructed image in memory 41.

[0037] The image processing function 53 generates an MR image for presentation to the MRI examiner by applying predetermined image processing to the reconstructed image stored in the memory 41 based on the input operation received by the input interface 43. The MR images generated by the image processing function 53 include, for example, T1-weighted images (Weighted Imaging: WI), T2-weighted images, proton density-weighted images, FLAIR (Fluid Attenuated Inversion Recovery) images, T2*-weighted images, susceptibility-weighted images (SWI), diffusion-weighted images (DWI), MR angiography (MRA) images, and MR perfusion (MRP) images. In particular, when generating a diffusion-weighted image, the image processing function 53 generates a more suitable diffusion-weighted image by removing the background phase from the original reconstructed image (a complex signal image (hereinafter referred to as "complex image")). Details of the image processing at this time will be described later. The image processing function 53 stores the converted MR image in the memory 41.

[0038] Image processing function 53 is an example of a "medical image processing device." A complex image is an example of a "first complex image," and a diffusion-weighted image is an example of a "medical image."

[0039] The output control function 54 controls, for example, the display mode on the display 42. The output control function 54 outputs and displays the MR image generated by the image processing function 53 and stored in the memory 41 on the display 42. This allows the person performing the MRI examination to visually confirm the MR image displayed on the display 42 and perform diagnoses and examinations, such as whether or not there is a lesion in the subject P. The output control function 54 may also transmit the MR image to, for example, a tablet terminal connected to the main body of the console device 40 via a network (not shown) and display it on the display device. The output control function 54 may also display GUI images or the like to accept various operations from the person performing the MRI examination.

[0040] [Functional Configuration of Medical Image Processing Equipment] Next, the configuration and operation for realizing the function of performing image processing to generate a diffusion-weighted image in the image processing function 53 will be described. Figure 2 is a diagram showing an example of the functional configuration of the medical image processing apparatus (image processing function 53) according to the embodiment. Figure 3 is a diagram schematically showing an example of an image from which the background phase of a complex image has been removed in the medical image processing apparatus (image processing function 53) according to the embodiment. The image processing function 53 performs, for example, a background phase estimation function 532, a background phase removal function 534, an image generation function 536, etc. Figure 4 is a diagram schematically showing an example of an image from which the background phase of a complex image has been removed by conventional magnitude transformation calculation. Figure 4 is a diagram for comparing image processing by the image processing function 53 with image processing by conventional magnitude transformation. In the following description, the example image shown in Figure 3 will be referred to as appropriate, and each function performed by the image processing function 53 shown in Figure 2 will be described while comparing it with the example of a conventional image shown in Figure 4.

[0041] Hereinafter, in the following explanation, it is assumed that multiple complex images, which are the basis for the diffusion-weighted image generated by the reconstruction processing function 52, are stored in memory 41. Each complex image is a medical image acquired in the MRI device 1 by changing the strength and direction of the gradient magnetic field. In other words, each complex image is a medical image in which the diffusion motion of water molecules (the strength of diffusion of water molecules by Brownian motion) is captured as a change in phase by transmitting (supplying) RF pulses of MPG intensity in the direction of the MPG (Motion Probing Gradient) in the MRI device 1. More specifically, each complex image is a medical image obtained in the MRI apparatus 1 when, in accordance with the sequence information corresponding to the diffusion-weighted image acquisition process, the sequence control circuit 31 generates a magnetic field in the imaging port (bore) of the gantry apparatus 10, and the transmission circuit 33 transmits (supplies) an RF pulse of MPG intensity in the MPG direction from the RF coil 16, and the receiving circuit 34 receives (detects) the MR signal output by the RF coil 16 and RF coil 17. Each complex image has a real component and an imaginary component.

[0042] Figure 3(a) shows an example of a complex image processed by the image processing function 53. In Figure 3(a), the complex image ZI is shown separately as the real component image (hereinafter referred to as the "real image") Re(ZI) and the imaginary component image (hereinafter referred to as the "imaginary image") Im(ZI). Here, both the real image Re(ZI) and the imaginary image Im(ZI) contain a signal of interest (a true signal that is not noise) representing the image of the subject P, and Gaussian noise (noise containing both positive and negative values) that follows a Gaussian distribution. The Gaussian noise is at a level that can be considered to be "0" on average. Furthermore, both the real image Re(ZI) and the imaginary image Im(ZI) contain phase fluctuations during imaging.

[0043] The complex image ZI is an example of a "first complex image". Gaussian noise is an example of "noise" and "noise containing both positive and negative noise values".

[0044] Conventional image processing using magnitude transformation converts each complex image to its magnitude. This method of magnitude transformation of complex images is well known. In the following explanation, an image obtained by representing a complex image as a magnitude will be called a "magnitude image," and an image obtained by representing a complex image as a phase will be called a "phase image."

[0045] Figure 4(a) shows an example of the complex image ZI shown in Figure 3(a) represented by magnitude and phase. In Figure 4(a), the complex image ZI is shown separately as the magnitude image |ZI| and the phase image Arg(ZI). Here, the magnitude image |ZI| contains the signal of interest and lysian noise (noise with only positive values) that follows a Rice distribution, which is obtained by transforming the Gaussian noise through the magnitude transformation. Lysian noise is noise that generates a noise floor and causes a decrease in the contrast of the resulting diffusion-weighted image.

[0046] The background phase estimation function 532 acquires each complex image used to generate the diffusion-weighted image from memory 41, and calculates (estimates) the background phase contained in the complex image based on the image calculated by smoothing each acquired complex image. The method for estimating the background phase in the background phase estimation function 532 is a known method. The calculation (estimate) of the background phase in the background phase estimation function 532 may be performed using, for example, AI (Artificial Intelligence) technology. The background phase is information that represents the phase component of the smoothed complex image. The background phase estimation function 532 outputs an image representing the estimated background phase (hereinafter referred to as the "background phase image") to the background phase removal function 534.

[0047] The background phase removal function 534 removes the background phase from the complex signal of each complex image acquired from memory 41, based on the background phase image output by the background phase estimation function 532. More specifically, the background phase is removed by applying a coefficient calculated based on the complex conjugate of the background phase and the magnitude of the background phase image to the complex signal, so that the signal intensity (level) of the signal of interest, which is represented by the real and imaginary components in the complex signal, can be represented entirely by the real component. The background phase removal function 534 outputs each complex image from which the background phase has been removed (hereinafter referred to as the "complex image after background phase removal") to the image generation function 536.

[0048] Figure 3(b) shows an example of a complex image RZI obtained by removing the background phase from the complex image ZI using the background phase removal function 534. In Figure 3(b), the complex image RZI after background phase removal is shown separately as the real part image Re(RZI) and the imaginary part image Im(RZI). The real part image Re(RZI) contains the signal of interest and Gaussian noise, while the imaginary part image Im(RZI) contains only Gaussian noise. In other words, the real part image Re(RZI) and the imaginary part image Im(RZI) are images from which the background phase that was present in the real part image Re(ZI) and the imaginary part image Im(ZI) has been removed. Therefore, by simply averaging the complex image RZI, the level of Gaussian noise can be reduced (for example, to a level that can be considered almost "0").

[0049] The complex image RZI is an example of a "second complex image". The real part image Re(RZI) is an example of a "second component image", and the imaginary part image Im(RZI) is an example of a "first component image".

[0050] Figure 4(b) shows an example of images representing the complex image RZI shown in Figure 3(b) as both a magnitude image and a phase image. In Figure 4(b), the complex image RZI is shown as both a magnitude image |RZI| and a phase image Arg(RZI). Comparing the complex image ZI shown in Figure 4(a) with the complex image RZI shown in Figure 4(b), the magnitude image |RZI| after background phase removal does not change significantly from the magnitude image |ZI|. The phase image Arg(RZI) after background phase removal has the background phase contained in the phase image Arg(ZI) removed, and although some noise components remain, it is a nearly uniform image.

[0051] Figure 3(c) shows an example of an image representing the removal of background phase from the complex signal of a complex image ZI on the complex plane of the complex signal. In Figure 3(c), the signal intensity (level) of the signal of interest represented by the complex signal CS on the complex plane, which is represented by the real axis R and the imaginary axis I, is represented as the length of an arrow with a background phase component θ. Figure 3(c) shows that the background phase removal function 534 removes the background phase (i.e., rotates the arrow to make the background phase component θ "0°"), thereby converting the complex signal CS into a complex signal RCS in which all of the signal intensity (level) of the signal of interest is represented as the length of an arrow on the real axis R.

[0052] The configuration of the image processing function 53 up to this point, that is, the configuration of the background phase estimation function 532 and the background phase removal function 534 provided by the image processing function 53, is the same as the configuration of the conventional method of performing real part addition by removing the background phase. Therefore, by discarding the imaginary part image Im(RZI) which contains only Gaussian noise from the complex image RZI shown in Figure 3(b) (without using the imaginary part image Im(RZI)) and performing averaging of the real part image Re(RZI), the image processing function 53 can also generate a diffusion-weighted image similar to the conventional method. In other words, with the configuration up to this point, the image processing function 53 can generate a diffusion-weighted image similar to the conventional method without converting the Gaussian noise contained in the complex image ZI into lysian noise, which is a factor that causes a decrease in contrast.

[0053] The configuration in the image processing function 53 that performs real part addition by removing the background phase is not limited to the configuration of the background phase estimation function 532 and the background phase removal function 534 described above, but any configuration that generates a complex image equivalent to the complex image RZI shown in Figure 3(b) may be used.

[0054] In the image processing function 53, the image generation function 536 differs from the conventional configuration that performs real part addition by removing background phase, in that it generates a diffusion-weighted image by averaging complex images.

[0055] The image generation function 536 corrects the complex image after background phase removal output by the background phase removal function 534 to generate a diffusion-weighted image. More specifically, the image generation function 536 generates a diffusion-weighted image by correcting pixels in the real part of the complex image after background phase removal where the signal intensity (level) of the signal of interest is lower than a predetermined level, using the signal intensity of pixels in the imaginary part of the complex image. At this time, the image generation function 536 corrects the pixel values ​​of pixels in the real part of the real part of the image after background phase removal where the signal intensity (level) of the signal of interest, that is, the pixel value is "0" or less (pixel value = "0" or a negative pixel value), by replacing them with the pixel values ​​of pixels at the same position in the imaginary part of the imaginary part of the imaginary image after background phase removal. In other words, the image generation function 536 corrects the real part of the complex image after background phase removal using the pixel values ​​of pixels in the imaginary part of the imaginary image (the imaginary part of the complex image after background phase removal that contains only Gaussian noise), which would conventionally be discarded. As a result, the image generation function 536 can improve the noise level in the diffusion-weighted images it generates, producing more natural (less unnatural) medical images.

[0056] The image generation function 536 is an example of an "image generation unit". Among the pixels that make up the real image after background phase removal, a pixel whose signal intensity (level = pixel value) of the signal of interest is 0 or less (pixel value = 0, or a negative pixel value) is an example of a "first pixel".

[0057] Here, we will explain in more detail the method for generating diffusion-weighted images while correcting pixel values ​​in the image generation function 536. In the following explanation, we will assume that the number of complex images (complex images after background phase removal) that are averaged when generating the diffusion-weighted image, in other words, the number of times the complex images after background phase removal are added, is N (where N is a natural number greater than or equal to 2).

[0058] The image generation function 536 calculates the summation mean Avg of the pixel values ​​of the signal of interest (including Gaussian noise) for the complex image Z (complex image after background phase removal) for each MPG intensity bval and MPG direction bvec. This calculation of the summation mean Avg is expressed as shown in equation (1) below.

[0059]

number

[0060] In equation (1) above, nex (number of excitations) is the number of excitations performed when acquiring multiple complex images Z (i.e., the number of images taken), and is the number of complex images Z that are averaged to generate a diffusion-weighted image. In equation (1) above, Re(Z) is the value of the real part component of the complex image Z. That is, Re(Z nex=i,bval,bvec ) is the value of the real part component that constitutes the complex image Z for the i-th acquisition number nex, with MPG intensity bval and MPG direction bvec. Therefore, in equation (1) above, the real part component Re of the same MPG intensity bval and MPG direction bvec in multiple complex images Z from 1 to N acquisition numbers nex is added together and averaged to obtain the average value Avg bval,bvec This calculates the mean value Avg, which is the sum of the pixel values ​​of pixels at the same position in multiple complex images Z.

[0061] A pixel at the same position in multiple complex images Z is an example of a "first pixel." The summation average Avg is an example of a "summation average."

[0062] Here, the number of MPG axes is "3" (for example, the three axes X, Y, and Z shown in Figure 1), the number of imaging cycles nex is "2", and the MPG intensity bval of the complex image Z is 1000 [s / mm²]. 2 Let's consider the case where ]. In this case, the image generation function 536 calculates the summation average Avg as shown in equation (2) below.

[0063]

number

[0064] On the other hand, the image generation function 536 calculates a correction value Corr for correcting the summation mean Avg for each MPG intensity bval using the imaginary component Im for the complex image Z, i.e., Gaussian noise. This calculation of the correction value Corr is expressed as shown in equation (3) below.

[0065]

number

[0066] In equation (3) above, V is the MPG direction, i.e., the number of MPG axes, and Im(Z) is the value of the imaginary component of the complex image Z. In equation (3) above, the number of images captured nex is the i-th image, MPG intensity bval, and the value of the imaginary component (Im(Z)) that constitutes the complex image Z with MPG direction bvec=vj. nex=i,bval,bvec=vj The absolute value obtained by averaging the number of imaging counts nex from 1 to N is then averaged for the number of MPG axes on which the complex image Z was captured, and then smoothed to obtain the correction value Corr bval It is calculating the correction value Corr. bval In this calculation, smoothing may be omitted. That is, the correction value Corr may be obtained by averaging the absolute values ​​of the Gaussian noise values ​​of pixels at the same position corresponding to the average value Avg of the signal of interest across multiple complex images Z, and then averaging these values ​​for the number of MPG axes.

[0067] The Gaussian noise value of the pixel at the same position corresponding to the summation mean Avg of the signal of interest in multiple complex images Z is an example of "the noise value of the noise represented by the second pixel". The correction value Corr is an example of "the correction value". The calculation of the correction value Corr using equation (3) above is an example of "the calculation of the correction value".

[0068] Here, we consider the case where the number of MPG axes is "3" (for example, the three axes X, Y, and Z shown in Figure 1), the number of imaging cycles nex is "2", and the MPG intensity bval of the diffusion-weighted image is "1000". In this case, the image generation function 536 calculates the correction value Corr as shown in equation (4) below.

[0069]

number

[0070] Then, the image generation function 536 corrects the complex image output by the background phase removal function 534, after background phase removal, for each MPG intensity bval and MPG direction bvec, based on the summation average Avg calculated by equation (1) above and the correction value Corr calculated by equation (3) above. More specifically, the image generation function 536 compares the summation average Avg and the correction value Corr, and takes the larger one (not the smaller one) as the summation average Avg', which is the corrected summation average Avg in the complex image Z (complex image after background phase removal). The calculation of this summation average Avg' is expressed as shown in equation (5) below.

[0071]

number

[0072] The average value Avg' is an example of the corrected "pixel value of the first pixel".

[0073] In this way, the image generation function 536 generates a diffusion-weighted image by correcting the signal intensity (level) of the signal of interest in the real part component of the complex image Z by replacing it with the signal intensity (level) of the Gaussian noise in the imaginary part component if the signal intensity (level) is lower than the signal intensity (level) of the Gaussian noise in the imaginary part component. In other words, the image generation function 536 generates a diffusion-weighted image by correcting the pixel value of each pixel that makes up the real part image after background phase removal by replacing it with the noise value of the Gaussian noise in the imaginary part image after background phase removal if the pixel value is lower than the noise value of the Gaussian noise in the imaginary part image after background phase removal. As a result, the diffusion-weighted image generated by the image generation function 536 is a more natural-looking medical image than a diffusion-weighted image generated by averaging complex images that have undergone real part addition by background phase removal in the conventional method.

[0074] In the example described above, we considered the case where the number of MPG axes is three, the X, Y, and Z axes as shown in Figure 1, and showed an example of how to calculate the average value Avg and the correction value Corr. However, in the MRI device 1, the number of MPG axes can be greater than three by controlling, for example, the direction of the axis of the gradient magnetic field generated by the gradient magnetic field coil 14. In this case, the calculation of the average value Avg and the correction value Corr can be easily considered from the calculation method described above, depending on the number of MPG axes, and can be performed in the same way. Therefore, a detailed explanation of the calculation method for the average value Avg and the correction value Corr when the number of MPG axes is greater than three will be omitted.

[0075] Here, an example of an image generated by the image generation function 536 will be described. Figure 5 shows an example of an image generated by the medical image processing apparatus (image processing function 53, more specifically, the image generation function 536) according to the embodiment. Figure 5(a) shows an example of an isotropic diffusion-weighted image (IsoDWI) generated by the image generation function 536 as a diffusion-weighted image, and Figure 5(b) shows an example of a diffusion tensor image calculated by the image generation function 536 from the diffusion-weighted image. The geometric mean image and the diffusion tensor image are MR images (medical images) that can be generated based on a complex image after background phase removal, similar to the same background phase removal. In Figure 5, for comparison, the respective images generated based on a complex image after background phase removal using real part addition by conventional background phase removal (see the complex image RZI shown in Figure 3(b)) and the respective images generated by the image generation function 536 while correcting the complex image after background phase removal are shown. In the following explanation, the geometric mean image generated based on the complex image after background phase removal before correction is referred to as the "pre-improvement geometric mean image," and the geometric mean image generated while correcting the complex image after background phase removal is referred to as the "improvement geometric mean image." Furthermore, the diffusion tensor image generated based on the complex image after background phase removal before correction is referred to as the "pre-improvement diffusion tensor image," and the diffusion tensor image generated while correcting the complex image after background phase removal is referred to as the "improvement diffusion tensor image."

[0076] First, let's explain the geometric mean image shown in Figure 5(a). Figure 5(a-1) shows an example of the geometric mean image before improvement, and Figure 5(a-2) shows an example of the geometric mean image after improvement. To facilitate explanation, the geometric mean images shown in Figure 5(a-1) and Figure 5(a-2) are magnified images of the same position in the geometric mean image of the same location, respectively: the geometric mean image before improvement, which was generated based on the complex image after background phase removal before correction, and the geometric mean image after improvement, which was generated while correcting the complex image after background phase removal. Furthermore, Figure 5(a-1a) and Figure 5(a-2a) show magnified images of the same region Ra at a further location. As can be seen by comparing the pre-improvement geometric mean image shown in Figure 5(a-1) with the improved geometric mean image shown in Figure 5(a-2), or by comparing the magnified image of region Ra shown in Figure 5(a-1a) with the magnified image of region Ra shown in Figure 5(a-2a), the improved geometric mean image shows an improvement in the noise content, resulting in a more natural-looking medical image than the pre-improvement geometric mean image. More specifically, in the complex image after background phase removal before correction, pixels with a pixel value of "0" or less appear as unnaturally missing pixels (black) in the generated pre-improvement geometric mean image, such as pixel P1 shown in Figure 5(a-1a). In contrast, in the improved geometric mean image, pixels with a pixel value of "0" or less are corrected with the pixel value at the same position in the imaginary part image (see the imaginary part image Im(RZI) shown in Figure 3(b)), which contains only Gaussian noise that would normally be discarded. In other words, the signal intensity (level) of the Gaussian noise is used for correction. Therefore, pixels do not appear unnaturally missing, such as pixel P1 shown in Figure 5(a-2a). Moreover, in the improved geometric mean image, the correction is performed by replacing pixel values ​​of "0" or less with the noise value of the Gaussian noise, so the correction is not excessive. As a result, the improved geometric mean image shown in Figure 5(a-2) is a more natural medical image with less noticeable distortion than the unimproved geometric mean image shown in Figure 5(a-1).

[0077] Next, we will explain the diffusion tensor image shown in Figure 5(b). Figure 5(b-1) shows an example of a diffusion tensor image before improvement, and Figure 5(b-2) shows an example of a diffusion tensor image after improvement. Figures 5(b-1a) and 5(b-2a) show enlarged views of the same region Rb in the diffusion tensor image before improvement and the diffusion tensor image after improvement, respectively. A diffusion tensor image is a medical image that represents the direction in which water molecules are restricted by diffusion, for example, the left-right direction in the diffusion tensor image shown in Figure 5(b) is "red", the front-back direction (depth direction) is "green", and the up-down direction is "blue". The region Rb mainly contains the part where "water" is imaged. In a diffusion tensor image, the part where "water" is mainly imaged is a pixel with a low signal intensity (level) of the signal of interest, that is, a pixel value of "0" or less, or a pixel where negative values ​​are easily obtained. However, as can be seen from the uncorrected diffusion tensor image shown in Figure 5(b-1), or the magnified image of region Rb shown in Figure 5(b-1a), the uncorrected diffusion tensor image is an unnatural image in which adjacent pixels in the area where "water" is mainly captured are different colors, meaning that the continuity of diffusion restriction is lost. This is because the image processing to generate the diffusion tensor image involves logarithmic calculations that do not take negative pixel values ​​into account. As a result of these logarithmic calculations, noise values ​​of "0" or less, or negative noise values, in the Gaussian noise contained in the complex image after background phase removal before correction (more specifically, the real part image after background phase removal) become abnormal values ​​that are far removed from the pixel values ​​of the surrounding pixels, representing various directions. In contrast, as can be seen from the corrected diffusion tensor image shown in Figure 5(b-2), or the magnified image of region Rb shown in Figure 5(b-2a), the corrected diffusion tensor image is a more natural medical image in which the area where "water" is mainly captured is darker, and the continuity of diffusion restriction is observed.This is because, in the improved diffusion tensor image, pixels with pixel values ​​of "0" or less are corrected by Corr, the noise correction value of the Gaussian noise, which became a positive value when its absolute value was calculated in the calculation of equation (3) above. Therefore, even if logarithmic calculations are performed in the image processing to generate the diffusion tensor image, the pixel values ​​will not become abnormally large or far removed from those of the surrounding pixels.

[0078] In this way, the image generation function 536 generates a diffusion-weighted image (in the example above, a geometric mean image or a diffusion tensor image) by averaging and correcting the complex image after background phase removal output by the background phase removal function 534. The image generation function 536 outputs the generated diffusion-weighted image to the output control function 54. As a result, in the MRI device 1, the output control function 54 displays the diffusion-weighted image output by the image generation function 536, i.e., the image processing function 53, on the display 42, presenting the diffusion-weighted image to the person undergoing the MRI examination.

[0079] [Processing by medical image processing equipment] Next, the image processing flow for generating a diffusion-weighted image in the image processing function 53 will be described. Figure 6 is a flowchart showing an example of the processing flow for generating a medical image (diffusion-weighted image) in the medical image processing apparatus (image processing function 53) according to the embodiment. In the following description, it will be assumed that multiple complex images for generating the diffusion-weighted image generated by the reconstruction processing function 52 are stored in the memory 41.

[0080] When the image processing function 53 starts the process of generating a diffusion-weighted image, the background phase estimation function 532 retrieves each complex image used to generate the diffusion-weighted image from memory 41, and calculates (estimates) the background phase to be removed from the complex image based on the image calculated by smoothing each of the retrieved complex images (step S100). The background phase estimation function 532 outputs the background phase image representing the calculated (estimated) background phase to the background phase removal function 534.

[0081] Next, the background phase removal function 534 removes the background phase from the complex signal of each complex image acquired from the memory 41 based on the background phase image output by the background phase estimation function 532 (step S110). The background phase removal function 534 outputs each complex image after background phase removal to the image generation function 536.

[0082] Next, the image generation function 536 generates a diffusion-weighted image by averaging the pixel values ​​of the complex image output by the background phase removal function 534 while correcting them (step S120). The image generation function 536 stores the generated diffusion-weighted image in the memory 41.

[0083] The output control function 54 displays the diffusion-weighted image generated by the image processing function 53 (image generation function 536) and stored in the memory 41 on the display 42 and presents it to the person undergoing the MRI examination (step S130). Then, the MRI device 1 (image processing function 53 of the processing circuit 50) completes the processing of this flowchart.

[0084] With this configuration and processing, the image processing function 53 within the processing circuit 50 of the MRI device 1 generates a complex image (complex image after background phase removal) in the same way as the conventional configuration that performs real part addition by removing background phase. The image processing function 53 then uses the pixel values ​​of pixels in the imaginary image (imaginary image after background phase removal that contains only Gaussian noise) that would conventionally be discarded to correct pixels in the real image (real image after background phase removal) where the signal intensity (level) of the signal of interest is lower than a predetermined level, thereby generating a diffusion-weighted image. As a result, the MRI device 1 equipped with the image processing function 53 as a medical image processing device can improve the noise state contained in each of the original complex images generated by averaging, and can generate a more natural (less unnatural) diffusion-weighted image than the diffusion-weighted image generated by averaging complex images that have undergone real part addition by removing background phase in the conventional method. This allows the person performing the MRI examination to visually confirm the diffusion-weighted image (MR image) displayed on the display 42 and perform diagnoses and examinations such as whether or not there is a lesion in the subject P in a natural manner.

[0085] [Example of correction value] In the example described above, the correction value Corr, used to correct (replace) pixels with low signal intensity (level) of the signal of interest in the real part component of the complex image Z, was calculated for each MPG intensity bval using equation (3) above. In other words, the correction value Corr was calculated by averaging the absolute value of the Gaussian noise noise (imaginary part value) of pixels at the same position corresponding to the average value Avg of the signal of interest across multiple complex images Z, and then averaging this value for the number of MPG axes. Incidentally, in order to calculate the correction value Corr, it is necessary to calculate the absolute value of the Gaussian noise, which contains both positive and negative values. This is because if the result of averaging the Gaussian noise values ​​becomes a negative value, these negative pixels will cause the diffusion-weighted image to appear unnatural. However, the calculation of the absolute value of the Gaussian noise used to calculate the correction value Corr can be performed at any time. In other words, the timing at which a positive calculation result is obtained by calculating the absolute value can be at any point in the calculation of the correction value Corr. For example, the correction value Corr may be calculated as shown in equation (6) or equation (7) below. The correction value Corr bval When calculating this, smoothing may be omitted, but to facilitate comparison with equation (3) above, equations (6) and (7) below show an example where smoothing is performed. However, in the following explanation, the explanation of smoothing will be omitted to simplify the explanation of the timing of the calculation of the absolute value.

[0086]

number

[0087]

number

[0088] In equation (6) above, the value of the imaginary component (Im(Z)) that constitutes the complex image Z where the number of imaging steps nex is the i-th image, MPG intensity bval, and MPG direction bvec = vj.nex=i,bval,bvec=vj After calculating the absolute value of )), the number of imaging times nex is averaged from 1 to N, and then, the complex image Z is averaged by the number of MPG axes for which the complex image Z was imaged, to calculate the correction value Corr bval is being calculated. That is, in the above equation (6), first, the absolute value of the noise value of Gaussian noise of pixels at the same position corresponding to the average value Avg of the interest signal in a plurality of complex images Z is calculated, and then, the correction value Corr is calculated by performing two types of averaging. In the correction value Corr calculated by the above equation (6), since all the noise values of Gaussian noise are made positive values and averaged, it becomes a larger value than the correction value Corr calculated by the above equation (3). Therefore, in the diffusion-enhanced image generated by the image generation function 536, the pixel value at the corrected (replaced) position becomes larger than the diffusion-enhanced image generated by correcting (replacing) the pixels with the correction value Corr calculated by the above equation (3), and it becomes a medical image in which the noise is conspicuous. In other words, the diffusion-enhanced image generated by correcting (replacing) the pixels with the correction value Corr calculated by the above equation (6) becomes a medical image similar to the diffusion-enhanced image generated by a configuration that performs real part addition by removing the conventional background phase.

[0089] The calculation of the correction value Corr by the above equation (6) is an example of the "second calculation method".

[0090] On the other hand, in the above equation (7), when the number of imaging times nex is the i-th, the value of the imaginary part component (Im(Z nex=i,bval,bvec=vj )) that constitutes the complex image Z of the MPG intensity bval and the MPG direction bvec = vj is averaged from the number of imaging times nex from 1 to N, and further, after averaging by the number of MPG axes for which the complex image Z was imaged, the absolute value is calculated to obtain the correction value Corr bvalThis is how it calculates the correction value Corr. In other words, in equation (7) above, two types of summation averaging are performed on the noise values ​​of the Gaussian noise of the pixels at the same position corresponding to the summation average Avg of the signal of interest in multiple complex images Z, and then the absolute value is calculated to calculate the correction value Corr. In the correction value Corr calculated by equation (7) above, all the noise values ​​of the Gaussian noise, including positive and negative values, are summed and averaged until they become close to "0" before becoming positive, so it is a smaller value than the correction value Corr calculated by equation (3) above. For this reason, the diffusion-weighted image generated by the image generation function 536 has smaller pixel values ​​at the corrected (replaced) position than the diffusion-weighted image generated by correcting (replacing) pixels with the correction value Corr calculated by equation (3) above, resulting in a medical image in which the noise is corrected (removed) more naturally (without any sense of unnaturalness).

[0091] The calculation of the correction value Corr using equation (7) above is an example of the "first calculation method".

[0092] The image generation function 536 may be configured to switch between the correction values ​​Corr calculated by equations (3), (6), and (7) above when generating a diffusion-weighted image while correcting the complex image after background phase removal output by the background phase removal function 534. In other words, the image generation function 536 may be configured to change the intensity of the correction of the complex image after background phase removal performed when generating a diffusion-weighted image for each diffusion-weighted image to be generated. The image generation function 536 may be configured to change the intensity of the correction of the complex image after background phase removal performed when generating a diffusion-weighted image for each pixel that makes up the same diffusion-weighted image. Here, the intensity of the correction increases in the order of equations (7), (3), and (6) above.

[0093] [Another example of a correction value] In the example described above (including another example), the correction value Corr was calculated for each MPG intensity bval using Gaussian noise. However, the correction value Corr may also be calculated for each MPG intensity bval and MPG direction bvec, similar to the summation mean Avg. In this case, the calculation of the correction value Corr is expressed as shown in equations (8) and (9) below. The correction value Corr is calculated using equation (8) or (9) below. bval,bvec When calculating this, smoothing may be omitted, but in order to facilitate comparison with equations (3), (6), and (7) above, equations (8) and (9) below show an example of when smoothing is performed. However, in the following explanation as well, in order to simplify the explanation of the timing of the calculation of the absolute value, the explanation of smoothing will be omitted.

[0094]

number

[0095]

number

[0096] In equation (8) above, the number of imaging steps nex is the i-th step, and the value of the imaginary component (Im(Z)) that constitutes the complex image Z of MPG intensity bval and MPG direction bvec is the value of the imaginary component. nex=i,bval,bvec The values ​​are averaged from the number of images nex from 1 to N, and then the absolute value is calculated to obtain the correction value Corr. bval,bvec It is calculating...

[0097] On the other hand, in equation (9) above, the number of imaging steps nex is the i-th step, and the value of the imaginary component (Im(Z)) that constitutes the complex image Z of MPG intensity bval and MPG direction bvec is the i-th step. nex=i,bval,bvec )) After calculating the absolute value, the number of imaging cycles nex is averaged from 1 to N, and the correction value Corr bval,bvec It is calculating...

[0098] When generating a diffusion-weighted image, the image generation function 536 calculates a correction value Corr from the complex image after background phase removal output by the background phase removal function 534 using the above equation (8) or (9). bval,bvec The image is then corrected to generate a diffusion-weighted image. In this case as well, the diffusion-weighted image generated by the image generation function 536 is a more natural-looking medical image than the diffusion-weighted image generated by summing complex images after removing the background phase, as opposed to the diffusion-weighted image generated by summing complex images.

[0099] The image generation function 536 may be configured to switch between the correction value Corr calculated by equation (8) and equation (9) above when generating a diffusion-weighted image while correcting the complex image after background phase removal output by the background phase removal function 534. In other words, even when the image generation function 536 calculates the correction value Corr for each MPG intensity bval and MPG direction bvec, it may be configured to change the intensity of the correction of the complex image after background phase removal performed when generating the diffusion-weighted image on a pixel-by-pixel basis. Here, the intensity of the correction increases in the order of equation (8) and equation (9) above.

[0100] As described above, in the medical image processing apparatus of the embodiment, the image processing function 53 generates a complex image (complex image after background phase removal) in the same way as the conventional configuration that performs real part addition by removing background phase. Then, in the medical image processing apparatus of the embodiment, the image processing function 53 uses the pixel values ​​of pixels in the imaginary image (imaginary image after background phase removal that contains only Gaussian noise) that would conventionally be discarded, and generates a diffusion-weighted image while correcting pixels in the real image (real image after background phase removal) where the signal intensity (level) of the signal of interest is lower than a predetermined level. As a result, in the medical image diagnostic apparatus (MRI apparatus 1) equipped with the medical image processing apparatus of the embodiment, the noise state contained in each of the original complex images generated by additive averaging is improved, and a diffusion-weighted image that is more natural and less unnatural can be displayed on the display 42 and presented to the person undergoing the MRI examination than a diffusion-weighted image generated by additive averaging of complex images that have undergone real part addition by removing background phase in the conventional method. This allows the MRI examiner to visually confirm the presented diffusion-weighted images (MR images) and naturally perform diagnoses and examinations, such as determining whether or not there are lesions in the subject P.

[0101] In the embodiment described above, the image processing function 53 removes the background phase component contained in the complex image, similar to the configuration that performs real part addition by removing the background phase in the conventional method, and corrects the pixels of the real part component of the complex image using the noise value of the Gaussian noise of the imaginary part component that constitutes the complex image from which the background phase component has been removed. However, the noise value for correcting the pixels of the real part component of the complex image is not limited to the imaginary part component of the complex image from which the background phase component has been removed, and the pixels of the real part component of the complex image may be corrected using a noise value that can be estimated by applying various methods to the captured image. For example, the image processing function 53 may correct the pixels of the real part component of the complex image using a noise value that can be estimated from various MR images captured by the MRI device 1 (noise specific to the MRI device 1), such as g-Factor (geometry-Factor) or sensitivity map. In this case, the MR image used for noise estimation is not limited to the image (complex image) of subject P, the subject of the MRI examination in this case. For example, it may be an image of a different subject taken from the same imaging area, or an image taken when the subject is not placed on the top plate 24. In other words, the MR image used for noise estimation is not limited to the image taken in this MRI examination, but may also be an image of a different subject or taken at a different time. The functional configuration, operation, and processing of the image processing function 53 in these cases can be easily considered based on the functional configuration, operation, and processing of the image processing function 53 in the processing circuit 50 of the MRI apparatus 1 of the above-described embodiment. Therefore, a detailed explanation of the functional configuration, operation, and processing of the image processing function 53 when correcting the pixels of the real part component of the complex image using a value other than the noise value of the imaginary part component of the complex image from which the background phase component has been removed is omitted.

[0102] In the above-described embodiment, a medical image processing device was provided in the MRI device 1, and the image processing function 53 within the medical image processing device generated a diffusion-weighted image based on a complex image captured by the MRI device 1. However, the medical images that the medical image processing device generates are not limited to diffusion-weighted images. Any image generated by applying a magnitude transformation process to a complex image can be generated using the same approach as the medical image processing device (image processing function 53) in the above-described embodiment, thereby similarly improving the noise level and generating a more natural (less unnatural) medical image.

[0103] In the embodiment described above, an example was shown in which the image processing function 53 (more specifically, the background phase removal function 534) removes the background phase component by setting the background phase component θ to "0°", thereby treating the entire signal intensity (level) of the signal of interest as a complex signal on the real axis R (see Figure 3(c)). However, the method for removing the background phase component is not limited to setting the background phase component θ to "0°". For example, the background phase removal function 534 may remove the background phase component by setting the background phase component θ to "90°", thereby treating the entire signal intensity (level) of the signal of interest as a complex signal on the imaginary axis I. In other words, the entire signal intensity (level) of the signal of interest may be represented on either the real axis R or the imaginary axis I, and the other of the real axis R or the imaginary axis I may be removed as the background phase component. In this case, the functional configuration, operation, and processing of the image processing function 53 can be easily understood by swapping the real and imaginary components in the functional configuration, operation, and processing of the image processing function 53 within the processing circuit 50 of the MRI apparatus 1 of the above-described embodiment. Therefore, a detailed explanation of the functional configuration, operation, and processing of the image processing function 53 when the background phase component θ is removed as a complex signal on the imaginary axis I of the signal intensity (level) of the signal of interest is omitted.

[0104] In the embodiments described above, the case in which the medical image processing device is provided in a magnetic resonance imaging (MRI) device (MRI device 1) was used as an example. However, this is merely one example, and the medical image processing device can be applied to any medical imaging diagnostic device that handles complex signals (complex images) and generates medical images by processing them. For example, the medical imaging diagnostic device may be an ultrasound diagnostic device that captures complex signals (complex images). Even in this case, the medical image processing device can generate a more natural (less unnatural) medical image with improved noise levels. In this case, the configuration, operation, and processing of the medical image processing device should be equivalent to the configuration, operation, and processing of the medical image processing device in the embodiments described above, so as to be suitable for a medical imaging diagnostic device (e.g., an ultrasound diagnostic device) equipped with the medical image processing device. Therefore, a detailed explanation of the configuration, operation, and processing of the medical image processing device provided in medical imaging diagnostic devices other than MRI devices will be omitted.

[0105] The embodiments described above can be expressed as follows. A medical image processing device that generates a medical image based on a first complex image of a subject is equipped with processing circuitry. The aforementioned processing circuit is A second complex image is generated by removing the background phase component from the first complex image. Using the noise contained in the second complex image, the pixel values ​​of the first pixels in each pixel constituting the second complex image, where the pixel value of the signal of interest representing the image of the subject is "0" or less, are corrected to generate the medical image. Medical image processing equipment.

[0106] According to at least one embodiment described above, a medical image processing device (53) that generates a medical image (diffusion-weighted image) based on a first complex image (CI) of a subject (P) is provided, and an image generation unit (535) that generates a second complex image (RCI) by removing the background phase component from the first complex image, and uses the noise (Gaussian noise) contained in the second complex image to correct the pixel value of the first pixel in each pixel constituting the second complex image whose pixel value of the signal of interest representing the image of the subject is "0" or less, thereby generating the medical image, is provided, and in a medical image processing device that performs image processing for generating a medical image using a complex image of a subject, the values ​​of pixels with low signal intensity can be corrected to a state suitable for image processing.

[0107] While several embodiments have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. These embodiments can be carried out in a variety of other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents. [Explanation of Symbols]

[0108] 1...MRI device, 10...Pantry device, 12...Static magnetic field magnet, 14...Gradient field coil, 16...RF coil, 17...RF coil, 20...Patient table device, 22...Base, 24...Tabletop, 30...Control device, 31...Sequence control circuit, 32...Gradient field power supply, 33...Transmitting circuit, 34...Receiver circuit, 35...Patient table control circuit, 40...Console device, 41...Memory, 42...Display, 43...Input interface, 50...Processing circuit, 51...Acquisition function, 52...Reconstruction processing function, 53...Image processing function, 532...Background phase estimation function, 534...Background phase removal function, 536...Image generation function, 54...Output control function

Claims

1. A medical image processing device that generates a medical image based on a first complex image of a subject, An image generation unit generates a second complex image by removing the background phase component from the first complex image, and uses the noise contained in the second complex image to correct the pixel value of the first pixel in each pixel constituting the second complex image, such that the pixel value of the signal of interest representing the image of the subject is "0" or less, thereby generating the medical image. A medical image processing device equipped with [a specific feature].

2. The image generation unit, A second complex image is generated in which the signal of interest, represented by the real and imaginary components in the first complex image, is represented entirely by either the real or imaginary component. In each pixel constituting the first component image, which is represented by either the real or imaginary component of the second complex image, the pixel value of the first pixel is corrected using the noise value of the noise represented by the second pixel at the same position as the first pixel constituting the second component image, which is represented by either the real or imaginary component of the second complex image, to generate the medical image. The medical image processing apparatus according to claim 1.

3. The aforementioned medical image is a medical image generated by adding and averaging a plurality of the second component images. The image generation unit, The averaged pixel values ​​of the first pixels at the same position in multiple second component images are added together and averaged to calculate the pixel value representing the signal of interest in the pixels constituting the medical image. The noise values ​​represented by the second pixels at the same position in multiple first component images are added together and averaged to obtain a correction value for correcting the added average value. If the correction value at the same position is greater than the average value, the pixel value of the first pixel is corrected by replacing the average value with the correction value. The medical image processing apparatus according to claim 2.

4. The noise in question is noise that includes both positive and negative noise values. The calculation of the correction value includes calculating the absolute value of the noise value of the noise represented by the second pixel, The image generation unit, in calculating the correction value, switches the timing of calculating the correction value. The medical image processing apparatus according to claim 3.

5. The calculation of the aforementioned correction value is as follows: A first calculation method, which involves averaging the noise values ​​represented by the second pixels at the same position in multiple first component images, and then calculating the absolute value, A second calculation method involves calculating the absolute value of the noise value represented by the second pixel at the same position in each of the first component images, and then averaging the results of each calculation of the absolute value, Includes, The image generation unit switches between the first calculation method and the second calculation method in calculating the correction value. The medical image processing apparatus according to claim 4.

6. The aforementioned medical image is a diffusion-weighted image generated by a magnetic resonance imaging apparatus. The image generation unit generates a plurality of second complex images corresponding to each of the plurality of first complex images captured by changing the intensity and direction of the gradient magnetic field in the magnetic resonance imaging apparatus, and uses the noise contained in each of the second complex images to correct the pixel values ​​of the first pixels and generate the medical image. A medical image processing apparatus according to any one of claims 1 to 5.

7. The computer of a medical image processing device that generates a medical image based on a first complex image of a subject, A second complex image is generated by removing the background phase component from the first complex image. Using the noise contained in the second complex image, the pixel values ​​of the first pixels in each pixel constituting the second complex image, where the pixel value of the signal of interest representing the image of the subject is "0" or less, are corrected to generate the medical image. Medical image processing methods.

8. The computer of a medical image processing device that generates a medical image based on a first complex image of the subject, A second complex image is generated by removing the background phase component from the first complex image. Using the noise contained in the second complex image, the pixel values ​​of the first pixels in each pixel constituting the second complex image, where the pixel value of the signal of interest representing the image of the subject is "0" or less, are corrected to generate the medical image. program.

Images