JP6510193B2 - Magnetic resonance imaging apparatus and image processing apparatus - Google Patents

Description

Embodiments of the present invention relate to a magnetic resonance imaging (MRI) apparatus and an image processing apparatus .

  An MRI apparatus magnetically excites nuclear spins of an object placed in a static magnetic field with a radio frequency (RF) signal at a Larmor frequency, and generates magnetic resonance (MR) accompanying this excitation. It is an imaging diagnostic apparatus that reconstructs an image from a signal.

  In the examination of coronary arteries using an MRI apparatus, it is important to detect a cardiac rest period in which the motion of the coronary arteries is reduced. The stationary phase is confirmed as a period of several tens of ms to several hundreds of ms at the end systole and the end diastole, respectively. Once the stationary phase is identified, it is possible to obtain a clear image by performing data collection in the stationary phase.

  As a conventional stationary phase detection method, a method of tracking a coronary artery by local template matching of a coronary artery part, and a method of using a difference value or a cross correlation coefficient between adjacent image data as an index for the whole image It has been known.

  When tracking a coronary artery by template matching, the region of interest (ROI) is manually located at the position of the left and right coronary arteries in the image of the first frame of the cine image of the 4-chamber view of the heart. It is set. Then, the coronary artery is tracked by template matching of the ROI portion. When the trajectory of the coronary artery is obtained by tracking the coronary artery, a period in which the change in the position of the coronary artery is small can be extracted as the stationary phase.

  On the other hand, in the method of using the difference value or the cross correlation coefficient between adjacent images for the entire image as an index, the period in which the number of residual pixels is below a certain value or the cross correlation coefficient value is a certain value or more The stationary phase can be identified.

JP, 2011-147560, A

An object of the present invention is to provide a magnetic resonance imaging apparatus and an image processing apparatus capable of automatically detecting the resting period of a coronary artery with good accuracy.

A magnetic resonance imaging apparatus according to an embodiment of the present invention includes an image acquisition unit, a feature point detection unit, a coronary artery detection unit, and a rest period detection unit. The image acquisition unit collects magnetic resonance image data covering the heart of the subject and magnetic resonance cine image data depicting the coronary artery of the subject. The feature point detection unit automatically detects the position of the feature point of the heart based on magnetic resonance image data covering the heart. The coronary artery detection unit automatically sets at least an initial spatial search region in the magnetic resonance cine image data based on the position of the feature point, and targets a plurality of initial and subsequent search regions corresponding to a plurality of frames. A change in frame direction of the representative position of the coronary artery is detected from the magnetic resonance cine image data by template matching. The stationary period detection unit detects a period during which the coronary artery can be regarded as stationary based on a change in the frame direction of the representative position of the coronary artery. The coronary artery detection unit is configured to select a template having the largest matching score as a result of the template matching for the initial spatial search area among the plurality of templates simulating the shape of the coronary artery as the template for the initial spatial search area. It is configured to be adopted as a template for matching.
An image processing apparatus according to an embodiment of the present invention includes an image acquisition unit, a feature point detection unit, a coronary artery detection unit, and a rest period detection unit. The image acquisition unit collects image data covering the heart of the subject and cine image data depicting the coronary artery of the subject. The feature point detection unit automatically detects the position of the feature point of the heart based on the image data covering the heart. The coronary artery detection unit automatically sets at least an initial spatial search area in the cine image data based on the position of the feature point, and performs template matching for a plurality of initial search areas corresponding to a plurality of frames. The change in the frame direction of the representative position of the coronary artery is detected from the cine image data by the The stationary period detection unit detects a period during which the coronary artery can be regarded as stationary based on a change in the frame direction of the representative position of the coronary artery. The coronary artery detection unit is configured to select a template having the largest matching score as a result of the template matching for the initial spatial search area among the plurality of templates simulating the shape of the coronary artery as the template for the initial spatial search area. It is configured to be adopted as a template for matching.

FIG. 1 is a block diagram of a magnetic resonance imaging apparatus according to an embodiment of the present invention. FIG. 2 is a functional block diagram of a computer shown in FIG. The figure which shows the acquisition period of MR image data and MR cine image data which cover a heart. The figure explaining the setting method of the initial spatial search area | region for template matching of a coronary artery. The figure which shows an example of the template for template matching in an initial spatial search area | region. The figure explaining the update method of the template for template matching. The figure which shows the example which detected the stationary period of the coronary artery based on the change in the flame | frame direction of the representative position of a coronary artery. 7 is a flowchart showing an example of the operation of the magnetic resonance imaging apparatus shown in FIG.

A magnetic resonance imaging apparatus and an image processing apparatus according to an embodiment of the present invention will be described with reference to the attached drawings.

  FIG. 1 is a block diagram of a magnetic resonance imaging apparatus according to an embodiment of the present invention.

  The magnetic resonance imaging apparatus 20 includes a cylindrical static magnetic field magnet 21 that forms a static magnetic field, a shim coil 22 provided inside the static magnetic field magnet 21, a gradient magnetic field coil 23, and an RF coil 24.

  The magnetic resonance imaging apparatus 20 further includes a control system 25. The control system 25 includes a static magnetic field power supply 26, a gradient magnetic field power supply 27, a shim coil power supply 28, a transmitter 29, a receiver 30, a sequence controller 31 and a computer 32. The gradient magnetic field power supply 27 of the control system 25 is composed of an X axis gradient magnetic field power supply 27 x, a Y axis gradient magnetic field power supply 27 y and a Z axis gradient magnetic field power supply 27 z. The computer 32 further includes an input device 33, a display device 34, an arithmetic device 35, and a storage device 36.

  The static magnetic field magnet 21 is connected to the static magnetic field power supply 26 and has a function of forming a static magnetic field in the imaging region by the current supplied from the static magnetic field power supply 26. The static magnetic field magnet 21 is often constituted by a superconducting coil, and is connected to the static magnetic field power supply 26 at the time of excitation to supply a current, but after being excited once, it is not connected. Is common. In addition, the static magnetic field magnet 21 may be a permanent magnet, and the static magnetic field power supply 26 may not be provided.

  Further, a cylindrical shim coil 22 is provided coaxially inside the static magnetic field magnet 21. The shim coil 22 is connected to a shim coil power supply 28 and is configured such that a current is supplied from the shim coil power supply 28 to the shim coil 22 to make the static magnetic field uniform.

  The gradient magnetic field coil 23 is composed of an X axis gradient magnetic field coil 23 x, a Y axis gradient magnetic field coil 23 y and a Z axis gradient magnetic field coil 23 z, and is formed in a tubular shape inside the static magnetic field magnet 21. A bed 37 is provided inside the gradient magnetic field coil 23 to be an imaging region, and the subject P is set on the bed 37. The RF coil 24 includes a whole body coil (WBC: whole body coil) for transmitting and receiving an RF signal built in the gantry, a bed 37 and a local coil for receiving an RF signal provided in the vicinity of the subject P.

  The gradient magnetic field coil 23 is connected to the gradient magnetic field power supply 27. The X-axis gradient coil 23x, Y-axis gradient coil 23y and Z-axis gradient coil 23z of the gradient coil 23 are respectively the X-axis gradient power supply 27x, Y-axis gradient power supply 27y and Z-axis gradient of the gradient power supply 27 It is connected to the magnetic field power supply 27z.

  Then, currents supplied from the X-axis gradient power supply 27x, the Y-axis gradient power supply 27y and the Z-axis gradient power supply 27z to the X-axis gradient coil 23x, the Y-axis gradient coil 23y and the Z-axis gradient coil 23z, respectively. A gradient magnetic field Gx in the X-axis direction, a gradient magnetic field Gy in the Y-axis direction, and a gradient magnetic field Gz in the Z-axis direction can be formed in the imaging region.

  The RF coil 24 is connected to at least one of the transmitter 29 and the receiver 30. The RF coil 24 for transmission has a function of receiving an RF signal from the transmitter 29 and transmitting it to the subject P, and the RF coil 24 for reception is associated with excitation of nuclear spins in the subject P by the RF signal. It has a function of receiving the generated MR signal and giving it to the receiver 30.

  On the other hand, the sequence controller 31 of the control system 25 is connected to the gradient power supply 27, the transmitter 29 and the receiver 30. The sequence controller 31 controls information necessary to drive the gradient power supply 27, the transmitter 29 and the receiver 30, for example, operation control information such as the intensity and application time of pulse current to be applied to the gradient power supply 27. X-axis gradient magnetic field Gx, Y-axis gradient magnetic field Gy, Z-axis gradient magnetic field by driving the gradient magnetic field power source 27, the transmitter 29 and the receiver 30 according to the function of storing sequence information describing It has a function to generate Gz and RF signals.

  Also, the sequence controller 31 is configured to receive raw data, which is complex data obtained by detection and A / D (analog to digital) conversion of an MR signal in the receiver 30, and to provide it to the computer 32. Ru.

  For this reason, the transmitter 29 has a function of applying an RF signal to the RF coil 24 based on the control information received from the sequence controller 31, while the receiver 30 detects the MR signal received from the RF coil 24. Then, by performing necessary signal processing and performing A / D conversion, a function of generating raw data that is digitized complex data and a function of providing the generated raw data to the sequence controller 31 are provided.

  Further, the magnetic resonance imaging apparatus 20 is provided with an ECG unit 38 for acquiring an ECG (electro cardiogram) signal of the subject P. The ECG signal acquired by the ECG unit 38 is configured to be output to the computer 32 via the sequence controller 31.

  In addition, the computer 32 is provided with various functions by causing the computing device 35 to execute the program stored in the storage device 36 of the computer 32. However, in place of at least a part of the program, a specific circuit having various functions may be provided in the magnetic resonance imaging apparatus 20.

  FIG. 2 is a functional block diagram of the computer 32 shown in FIG.

  The arithmetic unit 35 of the computer 32 functions as an imaging condition setting unit 40 and a data processing unit 41 by executing a program stored in the storage unit 36. The imaging condition setting unit 40 includes a cardiac morphology imaging condition setting unit 40A and a cardiac cine imaging condition setting unit 40B. The data processing unit 41 includes an image generation unit 41A, a feature point detection unit 41B, a coronary artery detection unit 41C, and a rest period detection unit 41D. In addition, the storage device 36 functions as a k-space data storage unit 42, an image data storage unit 43, and a dictionary file storage unit 44.

  The imaging condition setting unit 40 has a function of setting an imaging condition including a pulse sequence and outputting the set imaging condition to the sequence controller 31. The cardiac morphology imaging condition setting unit 40A has a function of setting an imaging condition for acquiring MR image data that covers the heart of the subject P. The cardiac cine imaging condition setting unit 40B has a function of setting an imaging condition for collecting MR cine image data in which the coronary artery of the subject P is depicted.

  FIG. 3 is a diagram showing acquisition periods of MR image data and MR cine image data covering the heart.

  In FIG. 3A, the horizontal axis indicates time (cardiac phase direction), and the vertical axis indicates the voltage of the ECG signal. MR image data and MR cine image data covering the heart are collected in synchronization with ECG signals, respectively. A systole and a diastole exist between R waves which are reference waves of the ECG signal. During systole and diastole, stationary phases of the coronary artery exist respectively. The diastolic stationary phase exists in the middle of diastole, specifically about the cardiac phase corresponding to about 75% of the period (RR) between adjacent R waves.

  Therefore, the MR signals for generation of MR image data covering the heart of the subject P are acquired in the systolic or diastolic resting phase of the heart in ECG synchronization from the viewpoint of reducing the influence of the movement of the heart. It is appropriate to set imaging conditions. In particular, as shown in FIG. 3B, it is practical to collect MR signals for generation of MR image data covering the heart of the subject P in the middle diastole of the heart. Therefore, hereinafter, an example will be described in which MR signals for generation of MR image data mainly covering the heart are acquired in the middle diastole.

  As a simple method, MR image data covering the heart of the subject P can be collected as axial cross-sectional image data, coronal cross-sectional image data, or sagittal cross-sectional image data of a plurality of frames. That is, MR image data covering the heart of the subject P can be acquired as any of two dimensional (2D) simple orthogonal three-sectional image data.

  In this case, acquisition of MR image data covering the heart of the subject P is multi-slice imaging in an axial direction, coronal direction or sagittal direction targeting a three-dimensional (3D: three dimensional) region including the heart. . Therefore, MR volume image data of a 3D region including the heart is acquired as multi-slice image data.

  However, MR volume image data of a 3D area including the heart may be acquired by a 3D scan which acquires MR signals from the 3D area including the heart. Further, all or two of axial cross-sectional image data, coronal cross-sectional image data, and sagittal cross-sectional image data may be acquired as MR image data covering the heart.

  When collecting axial cross-sectional image data and the like of a plurality of frames as MR image data covering the heart, cardiac phases can be regarded as identical as possible during the resting period of the heart so that the shapes of the heart match as closely as possible. It is desirable to collect the MR signal at Therefore, a scan for acquiring cross-sectional image data of a plurality of frames is a multi-slice data acquisition scan for acquiring MR signals repetitively in cardiac phase which can be regarded as identical between R waves in ECG synchronization. That is, as shown in FIG. 3B, a fixed delay time is set so that an MR signal of the ECG signal is triggered and the MR signal is acquired at the same cardiac phase during the resting period of the heart.

  If it is difficult to acquire an MR signal for one frame of cross-sectional image data during one heartbeat, it may be acquired over a plurality of heartbeats such as two heartbeats. Practically, the data acquisition conditions are set such that MR signals of one frame are acquired during one heartbeat or two heartbeats.

  If the MR signals for one frame are divided into a plurality of heartbeats and collected, the data collection period in one heartbeat can be shortened. Thus, by performing data collection while holding a breath, the effects of respiratory movement can also be reduced. Of course, even in the case where MR signals for one frame are collected during one heartbeat, the effects of respiratory movement can be reduced if data collection is performed during breath holding.

  In addition, it is important from the viewpoint of avoiding the influence of respiratory motion to collect MR signals by a single shot sequence that collects MR signals for one frame after application of one excitation pulse. In addition, high-speed imaging methods such as 2D steady state free precession (SSFP) are used to reduce the influence of cardiac phase shift and respiratory movement between MR signals. It is important to collect MR signals.

  On the other hand, MR cine image data is acquired in ECG synchronization after acquisition of MR image data covering the heart. Specifically, as shown in FIG. 3C, the MR cine image data has a period in which at least a stationary phase between different R waves is covered, ie, a period including a period from one stationary phase to the next stationary phase. The imaging conditions are set to be continuously collected.

  MR cine image data can be collected as time-series arbitrary 2D cross-sectional image data as long as a coronary artery is depicted. In the following, as MR cine image data, although the case where time-series four-chamber cross-sectional image (4-chamber view) data in which both the left coronary artery and the right coronary artery are depicted is acquired is described as an example Desired image data such as axial cross-sectional image data can be collected as MR cine image data, regardless of whether it is cross-sectional image data of a depicted heart.

  The data processing unit 41 acquires from the sequence controller 31 the MR signals acquired by the imaging scan under the imaging conditions set in the imaging condition setting unit 40 and arranges them in the k space formed in the k space data storage unit 42. Function, function to reconstruct image data by taking k-space data from k-space data storage unit 42 and performing image reconstruction processing including Fourier transform (FT: Fourier transform), image data obtained by reconstruction In the image data storage unit 43, and a function for performing necessary image processing on the image data fetched from the image data storage unit 43 and causing the display device 34 to display the image data. In addition, the data processing unit 41 has a function of detecting the resting period of the coronary artery as shown in FIG. 3 (A) based on the MR image data covering the heart of the subject P and the MR cine image data depicting the coronary artery. have.

  The image generation unit 41A has a function of generating MR image data by image reconstruction processing on the MR signal and necessary image processing. For example, when an MR signal is acquired under the imaging condition set in the cardiac morphology imaging condition setting unit 40A, the image generation unit 41A generates MR image data that covers the heart based on the MR signal. When an MR signal is acquired under the imaging condition set in the cardiac cine imaging condition setting unit 40B, the image generation unit 41A generates MR cine image data in which a coronary artery is drawn based on the MR signal.

  For this reason, the magnetic resonance imaging apparatus 20 includes the subject P by causing the image generation unit 41A of the computer 32 to cooperate with hardware such as the static magnetic field magnet 21, the shim coil 22, the gradient magnetic field coil 23, and the RF coil 24. A function as an image acquisition unit for acquiring MR image data covering the heart of the subject and MR cine image data in which the coronary artery of the subject P is depicted. However, if similar functions are provided, the image acquisition unit may be configured by other components.

  The feature point detection unit 41B has a function of detecting the position of a feature point of the heart based on MR image data covering the heart. When detecting the resting period of the right coronary artery in the data processing unit 41, the spatial positions of the apex and the tricuspid valve are automatically detected as feature points of the heart. On the other hand, when detecting the resting period of the left coronary artery in the data processing unit 41, the spatial positions of the apex and the mitral valve are automatically detected as feature points of the heart. Therefore, at least the spatial coordinates of the apex are automatically detected as feature points of the heart, and when detecting the resting periods of the left and right coronary arteries, the spatial coordinates of three points of the mitral valve, the apex and the tricuspid valve are automatically detected. Ru.

  When MR image data covering the heart is acquired as multi-slice image data, the slice interval may be coarser than the resolution in the slice. Therefore, in order to remove anisotropy related to the resolution, prior to the detection of the feature points of the heart, it is possible to perform isotropic processing to make the resolution of the multi-slice image data equal in three axial directions. The isotropic process can be performed by an interpolation process and a process of removing an unnecessary point. When the isotropic process is performed, the feature points of the heart are detected based on volume image data that covers the heart after the isotropic process.

  The method of detecting heart feature points is arbitrary. A method based on machine learning may be mentioned as a method for automatically detecting heart feature points with high accuracy. In a method based on machine learning, multiple dictionary file data are used. For that purpose, the dictionary file storage unit 44 stores a plurality of dictionary file data for detecting heart feature points.

  The dictionary file data is data representing, as a combination of parameters, determination criteria for specifying the positions of feature points such as the apex of the heart, the mitral valve, and the tricuspid valve from volume image data including the heart. On the other hand, by referring to the parameters specified by the dictionary file, it is possible to prepare an algorithm or function for determining heart feature points.

  As a specific example, using a matrix representing pixel values at each position of volume image data including the heart as an input and referring to a dictionary file, features such as the apex of the apical portion, the apex of the mitral valve and the apex of the tricuspid valve It is possible to create non-linear determinants that return the position vector of a point as an output. Alternatively, a function that takes a matrix representing pixel values at each position of volume image data as an input, refers to a dictionary file, and returns a position vector corresponding to a feature point of the heart based on information such as statistics and probability distribution You can also create

  Therefore, by referring to the dictionary file stored in the dictionary file storage unit 44, it is possible to determine the spatial position of the feature point. In this case, the feature point detection unit 41B functions as a classifier that determines the spatial position of the feature point with reference to the dictionary file.

  The dictionary file is an expression of the anatomical form of the heart as a combination of a plurality of parameters as described above, and anatomical knowledge about the form of each part of the heart and morphological image data of the heart of many subjects. It can be created based on For example, it is possible to calculate a combination of parameters constituting a dictionary file by so-called learning theory, based on a large number of heart shape information classified by disease.

  Specifically, for example, if the site to be identified is a mitral valve, the signal at each point with respect to the area including the mitral valve and the blood area surrounding the mitral valve and the myocardium etc. A multi-dimensional vector information consisting of values, gradients between signal values, signal values and histogram information on gradients can be used as a dictionary file.

  Then, the feature point detection unit 41B is used as a discriminator (discriminator) for discriminating feature points such as an apical portion of the heart by creating a large number of dictionary file data and storing the data in the dictionary file storage unit 44. can do. In other words, each dictionary file stored in the dictionary file storage unit 44 can be used as reference data for calculating the position of the feature point in the feature point detection unit 41B.

  That is, the feature point detection unit 41B performs anatomical knowledge and other plural objects based on simple plural cross-sectional image data such as axial cross-sectional image data or volume image data generated from plural cross-sectional image data. The detection processing of the position information of the feature points of the heart can be performed with reference to the reference data generated from at least one of the morphological image data of the heart.

  The dictionary file storage unit 44 may be added with a dictionary file based on heart shape image data collected by another magnetic resonance imaging apparatus or another image diagnostic apparatus such as an X-ray computed tomography (CT) apparatus. . In addition, it is preferable to prepare a dictionary file that is statistically processed and created for each feature such as the physical type of the subject and the type of disease.

  As another example of a method of detecting heart feature points, a method using signal processing based on anatomical knowledge on volume image data generated from a plurality of simple cross-sectional image data such as axial cross-sectional image data or a plurality of cross-sectional image data Can be mentioned.

  For example, when axial cross-sectional image data is collected under conditions where signals from blood flow are emphasized, anatomical knowledge that signal values from blood flow from the aorta to the left ventricle and apex have continuity. On the basis of the above, it is possible to analytically automatically detect the feature points of the heart by detecting the large part of the pixel value from the volume image data.

  More specifically, in certain axial cross-sectional image data, the search range of the aorta is determined based on the known information of the spatial position with respect to the aorta. Next, the position where the pixel value is maximum in the search range of the aorta is detected. Further, by tracing the point having the maximum pixel value in the axial direction and connecting the detected point group, the center line of the aorta can be obtained. By detecting the end of the center line of the aorta thus determined, it is possible to automatically detect the apex. Also, by calculating the gradient of pixel values around the center line of the blood vessel and detecting the gradient of pixel values corresponding to the shape of the mitral valve, the tip position of the mitral valve can be automatically detected. Furthermore, the pointed position of the tricuspid valve can also be automatically detected in the same manner. That is, the continuity of the signal from the blood can be used to automatically detect the feature point of the heart as well as the position of the blood vessel.

  Furthermore, the detection process of the position information of the feature points of the heart referring to the dictionary file data and the detection process of the position information of the feature points of the heart by signal processing based on anatomical knowledge are switched according to the desired conditions. It is good. As a specific example, detection processing of position information of the feature points of the heart is performed by the first algorithm including signal processing based on anatomical knowledge, and according to detection accuracy of position information of the feature points by the first algorithm, anatomy There is a method of performing processing of detecting position information of feature points according to a second algorithm with reference to dictionary file data generated from at least one of a plurality of other subject's cardiac morphological images.

  In addition, as another example of the detection method of the feature points of the heart, correct positions of the feature points are corrected by correcting the positions of the feature points of the heart provisionally calculated by an automatic calculation algorithm that performs 2D pattern matching. There is a way to calculate.

  Specifically, the apex position of the apex and mitral valve on the initial cross section is calculated from the volume image data by an automatic calculation algorithm that performs 2D pattern matching on the initial cross section of the volume image data and the reference data.

  Next, an equation is calculated that represents the position of the first major axis connecting the apex of the apex and the apex of the mitral valve on the initial cross-section. However, since the initial cross-section is a cross-section provisionally set for 2D pattern matching in the automatic calculation algorithm, the calculation accuracy of the positions of the apex, the apex of the mitral valve, and the first long axis is rough. Therefore, the initial cross section is rotated about the first major axis with low accuracy.

  Next, the rotation angle of the initial cross section is detected such that the resolution of the plurality of axial cross section image data before the isotropic processing is maximized. That is, since the axial cross-sectional image data has anisotropy, the resolution differs depending on the normal direction of the cross section intersecting the axial cross-sectional image data. Therefore, a high resolution cross section for detecting the apex position of the apex and mitral valve with high accuracy is detected.

  Next, accurate apical and mitral valve positions are detected by 2D pattern matching of axial cross-sectional image data and reference data on the detected high-resolution cross section or feature point detection of pixel values of axial cross-sectional image data be able to.

  Also, the apex position of the tricuspid valve can be detected in the same manner as the apex position of the mitral valve. Alternatively, for the apex position of the tricuspid valve, left ventricular short-axis image data perpendicular to the long axis of the heart determined as a straight line connecting the apex and the apex of the mitral valve is generated. It can also be detected by feature point detection processing such as template matching or pixel value determination.

  The position of each feature point can be automatically detected from the volume image data by template matching using a large number of templates of each feature point prepared for each of the features and the diseases of the subject P in addition to the method described above.

  According to the various methods described above, not only heart feature points, but also six reference cross sections required for cardiac examination, ie, vertical long-axis view, horizontal long-axis view ), 2-chamber view, 3-chamber view, 4-chamber view and left ventricle short axis view are also detected automatically can do.

  Therefore, the spatial position of the reference cross section can be automatically detected in the feature point detection unit 41B, and the imaging cross section for MR cine image data can be positioned using the detected spatial position of the reference cross section. In practice, the spatial position of the reference cross section automatically detected by the feature point detection unit 41B can be given to the cardiac cine imaging condition setting unit 40B and can be displayed on the display device 34. Then, the imaging section for MR cine image data can be set by performing the necessary fine adjustment on the spatial position of the reference section by the operation of the input device 33.

  The reference cross-sectional image of the heart in which the left and right coronary arteries are depicted is a four-chamber cross-sectional image. Therefore, for example, the feature point detection unit 41B automatically detects the spatial position of the four-chamber cross-sectional image from the MR volume image data covering the entire heart, and performs necessary adjustment on the automatically detected four-chamber cross-sectional image Thus, the four-chamber cross section of the heart can be set as an imaging cross section for MR cine image data. The position of the four-chamber cross section includes, for example, the long axis of the heart determined as a straight line connecting the apex and apex of the mitral valve, and the section passing through the apex position of the tricuspid valve, that is, the apex position of the mitral valve, apex It can be detected as a cross section passing through three points of the tip of the head and tricuspid valve.

  The coronary artery detection unit 41C has a function of detecting a change in the frame direction of the representative position of the coronary artery by template matching for each frame with respect to the MR cine image data. Template matching is image recognition processing that performs pattern matching of a target image using a template. In particular, the coronary artery detection unit 41C is configured to set at least an initial spatial search area for template matching in MR cine image data based on the position of the feature point detected by the feature point detection unit 41B. There is. In addition, about the search area after an initial stage, if the position of the 2D cross section used as search object is specified, it can be set as a 2D search area.

  FIG. 4 is a diagram for explaining a method of setting an initial spatial search region for template matching of a coronary artery.

  The heart has four chambers: right ventricle (RV), right atrium (RA), left atrium (LA) and left ventricle (LV). The LA and LV are separated by a mitral valve. On the other hand, RA and RV are separated by a tricuspid valve. Therefore, on the four-chamber cross section in which four chambers of RV, RA, LV and LA are drawn, three points of a mitral valve, an apex and a tricuspid valve are drawn as feature points.

  Then, as shown in FIG. 4, the space coordinates of the mitral valve, the apex and the tricuspid valve can be detected by the feature point detection unit 41B as a 2D position on the four-chamber cross section. When the positions of the mitral valve, the apex and the tricuspid valve are detected as the positions of the feature points, an initial spatial search of at least the right coronary artery based on the positions of the detected apex, the tricuspid valve and the mitral valve An area and an initial spatial search area of the left coronary artery can be set.

  Specifically, a line segment connecting the apex and the tricuspid valve is a first line segment, a line segment connecting the tricuspid valve and the estimated position of the right coronary artery is a second line segment, and the apex and the mitral valve Where the line segment connecting the two is the third line segment, and the line segment connecting the mitral valve and the estimated position of the left coronary artery is the fourth line segment, the length of the first line segment (apex and tricuspid valve Ratio A / B of the distance A) to the length of the second line segment (the distance between the tricuspid valve and the estimated position of the right coronary artery) B, the first line segment and the second line Of the third segment (the distance between the apex and the mitral valve) C and the length of the fourth segment (the estimated position of the mitral valve and the left coronary artery) The distance C) between the distance D) and the angle θ2 formed between the third line segment and the fourth line segment take similar values regardless of the object P.

  Therefore, the ratio A / B, the ratio C / D, the angle θ1 and the angle θ2 can be measured in advance for a plurality of subjects P. Then, the positions of the right and left coronary arteries can be estimated by using the ratio A / B, the ratio C / D, and the average value of the angles θ1 and θ2 as experience values. That is, rough positions of the right coronary artery and left coronary artery are respectively estimated based on the coordinates of the mitral valve, apex and tricuspid valve, and a predetermined range based on each estimated position of the right coronary artery and left coronary artery is the right coronary artery The initial spatial search region of the left coronary artery and the initial spatial search region of the left coronary artery can be set respectively.

  That is, according to the empirical value of the ratio A / B of the length A of the first line segment and the length B of the second line segment and the empirical value of the angle θ1 between the first line segment and the second line segment. An initial spatial search region of the right coronary artery can be set as a region based on the identified point. On the other hand, the empirical value of the ratio C / D of the length C of the third line segment to the length D of the fourth line segment and the empirical value of the angle θ2 between the third line segment and the fourth line segment An initial spatial search region of the left coronary artery can be set as a region based on the identified point.

  As another method, the ratio A / B, the ratio C / D, the ratio C / D, the angle θ1 and the angle θ2 are measured by measuring the ratio A / B, the ratio C / D, the angle θ1 and the angle θ2 for a plurality of subjects P in advance. It is possible to determine the possible range of Then, the range that the ratio A / B of the length A of the first line segment and the length B of the second line segment can take and the angle θ1 between the first line segment and the second line segment can take A possible range of the position of the right coronary artery specified corresponding to the range can be set in the initial spatial search region of the right coronary artery. On the other hand, the possible range of the ratio C / D of the length C of the third line segment to the length D of the fourth line segment and the angle θ2 between the third line segment and the fourth line segment A possible range of the position of the left coronary artery specified corresponding to the range can be set in the initial spatial search region of the left coronary artery.

  As described above, the coronary artery detection unit 41C has a function of narrowing down at least the initial spatial search region for template matching based on the position of the feature point detected by the feature point detection unit 41B. Therefore, template matching makes it possible to reliably track the coronary artery from the MR cine image data.

  If the heart feature points are automatically detected based on MR volume image data collected in the middle diastole, each position of the heart feature points will be detected as the position in the middle diastole . Therefore, an initial spatial search region is set in the MR cine image data in the middle diastole of the heart. That is, the frame in the middle diastole of the heart of MR cine image data is taken as an initial frame to be subjected to template matching.

  When an initial spatial search area of the coronary artery is set in the first frame of the MR cine image data to be subjected to template matching, template matching can be started. A template for template matching targeting an initial space search area can be prepared in advance. However, it is preferable to prepare a plurality of templates simulating the typical shape of the coronary artery, in consideration of the temporal change of the shape of the coronary artery, the variation of the shape of the coronary artery for each subject P, and the like.

  FIG. 5 is a diagram showing an example of a template for template matching in an initial space search area.

  As shown in FIG. 5, it is possible to prepare a plurality of templates T1, T2, T3 and T4 which simulate the typical shape of the coronary artery in the middle diastole of the heart. That is, preparing a plurality of templates T1, T2, T3 and T4 corresponding to the same cardiac phase or a predetermined time phase section leads to the detection of a coronary artery with template matching.

  Therefore, when performing template matching for both the left and right coronary arteries, templates for a plurality of left coronary arteries corresponding to the same cardiac phase or predetermined temporal phase segment, and the same cardiac phase or predetermined temporal phase It is preferable to prepare a plurality of templates for the right coronary artery corresponding to the sections. The coronary artery templates T1, T2, T3 and T4 become image data of about 25 pixels square when searching for a coronary artery depicted in four-chamber cross-sectional image data of 320 pixels square. The plurality of templates T1, T2, T3 and T4 shown in FIG. 5 are templates for the right coronary artery.

  As a template selection method when preparing a plurality of templates, template matching is actually performed using a plurality of templates, and a template whose correlation coefficient after normalization is closest to 1 is selected as an initial template The way to do it is practical. That is, of the plurality of templates simulating the shape of the coronary artery, as a result of template matching targeting the initial spatial search region, adopting the template with the highest degree of coincidence as the template for template matching of the initial spatial search region Can.

The normalized correlation coefficient r _xy calculated as an index of the degree of coincidence in template matching can be calculated, for example, by the evaluation function represented by equation (1).

The normalized correlation coefficient r _xy calculated by equation (1) takes a value from −1 to 1. If the brightness value of the template completely matches the brightness value of the corresponding area in the search area, the correlation coefficient r _xy after normalization is 1. On the other hand, when there is no relationship between the luminance value of the template and the luminance value of the corresponding area in the search area, the correlation coefficient r _xy after normalization is zero. Also, the pattern of the luminance value of the template completely matches the pattern of the luminance value of the corresponding area in the search area, but the correlation coefficient r _xy after normalization is -1 when the gradation is reversed. It becomes.

  By performing template matching using the value of such correlation coefficient as an index, representative positions of the left and right coronary arteries can be specified with high accuracy as the relative position of the template in the search area.

  In order to detect changes in the frame direction of the representative position of the coronary artery, the representative position of the coronary artery is detected for each frame from the MR cine image data by template matching targeting a plurality of search areas after the initial stage corresponding to the plurality of frames. It is necessary. Therefore, it is important to narrow down the search area after the initial space search area as well as the initial space search area.

  Therefore, as a simple method, the search area after the initial space search area can be made the same area as the initial space search area or a fixed area which is slightly larger. In this case, all search regions for template matching are set based on the positions of the feature points detected by the feature point detection unit 41B. In addition, the size of the search area is determined to at least the range in which the position of the coronary artery can be taken so that the position or the shape of the coronary artery changes and the area is surely within the search area.

  As a specific example, if a coronary artery depicted in 320-pixel square four-chamber cross-sectional image data is searched by performing template matching using a 25-pixel square template, the initial position of the coronary artery is considered in consideration of the change in the coronary artery position. The space search area can be set to an area of about 50 pixels square, and the search area after that can be set to an area of about 70 pixels square.

  Another method is to update the search area after the initial spatial search area at a predetermined frame interval based on the representative position of the coronary artery detected in the past. In this case, the initial spatial search area is set based on the position of the feature point detected by the feature point detection unit 41B, and the search areas for the second and subsequent frames are based on the representative position of the coronary artery detected in the past. It will be set. Specifically, the search area for the second and subsequent frames can be set to a predetermined range centered on the representative position of the coronary artery before the previous frame. In this case, the size of the search area can be reduced compared to the case where the search area is not changed.

  For example, in the case of searching for a coronary artery depicted in 320-pixel square four-chamber cross-sectional image data by performing template matching using a 25-pixel square template, all search areas are set to about 50-pixel square can do. Therefore, the time required for template matching can be shortened.

  The positions and shapes of left and right coronary arteries to be searched by template matching change according to the cardiac phase. In addition, between the inflow and outflow times of blood, the luminance values in the left and right coronary arteries also change. Therefore, using the same template may make it difficult to track the left and right coronary arteries.

  Therefore, a plurality of templates corresponding to different cardiac phases can be prepared in advance. Specifically, at least a diastolic template for the left coronary artery, a diastolic template for the right coronary artery, a systolic template for the left coronary artery, and a systolic template for the right coronary artery are prepared in advance. It is preferable to update the template according to the phase. In this case, the ability to trace the coronary artery can be secured by preparing a template corresponding to more cardiac phases.

  However, a method of using a part of MR cine image data as at least one template for template matching may be mentioned as a method of more reliably tracking the coronary artery. Specifically, a template for template matching for a search area after the initial spatial search area is set as MR cine image data of an area where a coronary artery is drawn, specified by the past template matching. It can be updated at a predetermined frame interval. This makes it possible to dispense with the preparation of a large number of templates and to update the templates following changes in the shape of the coronary artery.

  FIG. 6 is a view for explaining a method of updating a template for template matching.

  As shown in FIG. 6, template matching can be performed on an initial spatial search area of cine image data using an initial template prepared in advance. The initial spatial search area can be an area within a certain range centered on the position of the coronary artery estimated based on the positions of feature points such as the mitral valve, apex and tricuspid valve.

  When template matching in the initial spatial search area using the initial template is completed, the representative position of the coronary artery in the initial spatial search area is specified as the position of the initial template. Therefore, a rectangular area surrounded by an alternate long and short dash line corresponding to the position of the initial template in the initial space search area can be cut out and used as a template for the next search area.

  Therefore, a part of cine image data of the previous frame is used as a template for template matching in the next search area. Such cutting out of the area corresponding to the position of the template after template matching in the search area of cine image data, updating of the template to the cut out cine image data, and search area of the next frame by the updated template Template matching within can be repeated until the last frame targeted for template matching.

  Although FIG. 6 shows an example in which the template is updated for each frame, the template may be updated for each of a plurality of frames. Moreover, it is effective for reliable tracking of a coronary artery that the template corresponding to the area | region containing the tissue around a coronary artery is used for template matching, without restricting to the area which a coronary artery occupies.

  By sequentially executing such template matching limited to a search region on cine image data of a plurality of frames in time series, it is possible to obtain the position of the template at the time when the cross correlation coefficient becomes maximum for each frame. The change in position of the template in the frame direction can then be plotted.

  A change in position of the template in the frame direction can be regarded as a change in the representative position of the coronary artery. Therefore, if template matching of search regions corresponding to left and right coronary arteries is sequentially performed on cine image data of a plurality of frames in time series using templates corresponding to left and right coronary arteries, the frame direction of the representative position of the right coronary artery Changes and changes in the frame direction of the representative position of the left coronary artery can be detected.

  The stationary period detection unit 41D has a function of detecting a period during which the coronary artery can be regarded as stationary based on the change in the frame direction of the representative position of the coronary artery.

  FIG. 7 is a diagram showing an example in which the resting period of the coronary artery is detected based on the change in the frame direction of the representative position of the coronary artery.

  In FIG. 7, the vertical axis indicates the movement distance [pixel] between the frames of the template, and the horizontal axis indicates the frame numbers of time series. If the linear distance which the template moved between frames is calculated and plotted, a line graph representing the moving distance of the template as shown in FIG. 7 is obtained. The travel distance of the template represents the travel distance of the coronary artery. Therefore, the section of the frame in which the movement distance of the template is minimum can be determined as the stationary phase.

  In the example shown in FIG. 7, the section of the frame in which the movement distance of the template is zero appears twice. The two sections are considered to correspond to rest periods of the coronary arteries in diastole and systole. Thus, the cardiac phase periods corresponding to the two frame sections can be detected as the stationary period of the coronary artery.

  When a graph representing the same template movement distance was created for a plurality of subjects P, a frame section in which the template movement distance is zero appeared in all graphs.

  It is known that the coronary arteries move rapidly at the end of the rest period, but slowly stop at the start of the rest period. Therefore, the period before the stationary period in which the coronary artery moves at a low speed may be determined as the stationary period. That is, the resting period of the coronary artery may be detected earlier than the actual resting period.

  Therefore, the length of the resting period of the coronary artery may be measured in advance for a large number of subjects P, and the possible range of the length of the resting period may be determined. In that case, if the length of the resting period of the coronary artery detected from the graph representing the movement distance of the template exceeds the statistically determined threshold value, the length of the detected resting period of the coronary artery is determined A process of correcting to the length of Specifically, by delaying the start phase without changing the end phase of the detected rest period of the coronary artery, it is possible to make a correction to shorten the length of the rest period to an allowable length.

  Next, the operation and action of the magnetic resonance imaging apparatus 20 will be described.

  FIG. 8 is a flow chart showing an example of the operation of the magnetic resonance imaging apparatus 20 shown in FIG.

  First, the subject P is set on the bed 37 in advance, and a static magnetic field is formed in the imaging region of the static magnetic field magnet 21 (superconductive magnet) excited by the static magnetic field power supply 26. In addition, a current is supplied from the shim coil power supply 28 to the shim coil 22 to make the static magnetic field formed in the imaging region uniform.

  Furthermore, in the imaging condition setting unit 40, imaging conditions including at least an axial cross-sectional image data, coronal cross-sectional image data and sagittal cross-sectional image data depicting the heart of the subject P drawn are set as scout image data Be done. The set imaging condition is output to the sequence controller 31.

  Then, the sequence controller 31 causes the gradient magnetic field to be formed in the imaging region in which the subject P is set by driving the gradient magnetic field power source 27, the transmitter 29, and the receiver 30 according to the imaging conditions, and an RF signal from the RF coil 24. Generate For this reason, an MR signal generated by nuclear magnetic resonance inside the subject P is received by the RF coil 24 and given to the receiver 30. The receiver 30 receives an MR signal from the RF coil 24 and supplies the MR signal to the sequence controller 31. The sequence controller 31 outputs an MR signal to the data processing unit 41.

  The data processing unit 41 arranges the MR signal in the k space formed in the k space data storage unit 42 as k space data. Next, the image generation unit 41A acquires k-space data from the k-space data storage unit 42 and performs an image reconstruction process. Thereby, axial cross-sectional image data, coronal cross-sectional image data, and sagittal cross-sectional image data in which the heart of the subject P is depicted are generated as scout image data. The generated scout image data is displayed on the display device 34 for setting the imaging area.

  Then, in step S1, MR image data covering the heart in the middle diastole of the heart under ECG synchronization is acquired. As an example, axial cross-sectional image data of multiple frames covering the entire heart are collected.

  For that purpose, the imaging condition setting unit 40 sets a plurality of slice sections in the axial direction covering the entire heart as an imaging section with reference to the scout image data. Further, as exemplified in FIG. 3B, the delay time from the R wave is set so that the MR signal is acquired in the cardiac phase of about 75% of the inter-R wave (RR) of the heart.

  Then, a plurality of axial cross-sectional image data are collected in the same flow as the collection of scout image data. However, collection of a plurality of axial cross-sectional image data is performed under ECG synchronization. Also, desirably, multi-slice imaging of the axial cross section is performed during breath holding. A plurality of frames of axial cross-sectional image data covering the entire collected heart are stored in the image data storage unit 43.

  Next, in step S2, the feature point detection unit 41B automatically detects a feature point and a reference cross section of the heart from axial cross-sectional image data of a plurality of frames. That is, the positions of the apex, the mitral valve and the tricuspid valve are automatically detected by an arbitrary feature point detection process. At the same time, the position of the reference cross section of the heart such as the four-chamber cross section is automatically calculated by the feature point detection unit 41B.

  The position of the automatically calculated four-chamber cross section is displayed on the display device 34. Then, the user can finely adjust the position of the four-chamber cross section by the operation of the input device 33, and set it as the imaging cross section of the cine image data.

  Next, in step S3, an initial spatial search area for detecting the positions of the left and right coronary arteries from the cine image data of the four-chamber cross section by the coronary artery detection unit 41C is located at the position of the feature point detected by the feature point detection unit 41B. Set based on. Specifically, as shown in FIG. 4, the position of the left coronary artery and the position of the right coronary artery are estimated based on the positions of the apex, mitral valve and tricuspid valve, respectively.

  Then, a predetermined range centered on the estimated position of the left coronary artery is set in the initial spatial search region of the left coronary artery. Similarly, a predetermined range centered on the estimated position of the right coronary artery is set in the initial spatial search region of the right coronary artery. The size of the initial spatial search area can be about 50 pixels square if collecting 320-pixel square four-chamber cross-sectional image data.

  Next, in step S4, cine imaging of the four-chamber cross section in which the left and right coronary arteries and the heart are depicted is started. That is, four-chamber cross-sectional image data of a plurality of frames in time series are sequentially acquired under ECG synchronization.

  Next, in step S5, the coronary artery detection unit 41C selects cine image data of a frame in the middle diastolic phase as a target of template matching, and sets the templates of left and right coronary arteries prepared in advance as an initial template. That is, four-chamber cross-sectional image data in the cardiac phase of about 75% of the R-wave interval (RR) at which axial cross-sectional image data covering the entire heart can be considered identical to the acquired cardiac phase is selected as the first target of template matching. Be done. The four-chamber cross-sectional image data in the cardiac phase of about 75% of the R-wave interval (RR) is the four-chamber at the 45th frame if the 60-frame four-chamber cross-sectional image data is acquired in the R-wave interval (RR). It becomes cross-sectional image data.

  As described above, when axial cross-sectional image data is selected from four-chamber cross-sectional image data in the same or near cardiac phase as the acquired cardiac phase, it is estimated based on the positions of apical part, mitral valve and tricuspid valve The positions of the left and right coronary arteries can be regarded as the estimated positions of the left and right coronary arteries on the four-chamber cross-sectional image data.

  However, since there are individual differences in the size and shape of the heart, the estimated positions of the left and right coronary arteries do not necessarily become the correct positions of the left and right coronary arteries. Therefore, the selected mid-dilatation four-chamber cross-sectional image data is used as an initial search target for the left and right coronary arteries by template matching. Templates for the left and right coronary arteries prepared in advance are used as templates for that purpose. It is desirable from the viewpoint of reliably searching the left and right coronary arteries that the initial template be a template selected from a plurality of types of characteristic templates of left and right coronary arteries as illustrated in FIG. 5.

  Next, in step S6, the coronary artery detection unit 41C performs template matching for the initial spatial search area of the left and right coronary arteries set in the four-chamber cross-sectional image data in the middle-diastole phase. Thereby, the representative positions of the left and right coronary arteries can be detected as the positions of the left and right coronary artery templates. When template matching is performed using multiple types of templates as at least one of the left coronary artery template and the right coronary artery template, the position of the template at which the cross correlation coefficient is maximized should be the representative position of the coronary artery. Can.

  The target region of template matching with respect to the four-chamber cross-sectional image data in the middle-diastole phase is limited within the initial spatial search region set based on the position of the feature point of the heart in step S3. Therefore, the representative position of the coronary artery can be detected with high accuracy by simple processing.

  Next, in step S7, the coronary artery detection unit 41C records the representative positions of the left and right coronary arteries identified as the positions of the left and right coronary artery templates.

  Next, in step S8, the coronary artery detection unit 41C determines whether there is four-chamber cross-sectional image data of the next frame to be detected for the position of the coronary artery. If it is determined that the next frame is present, the region of the four-chamber cross-sectional image data of the current frame including the detection positions of the left and right coronary arteries is cut out as a template of the next frame in step S9.

  Specifically, as shown in FIG. 6, a portion of four-chamber cross-sectional image data corresponding to the position of the template specified in the search area of the current frame by template matching is cut out as a template for the next frame. Be Thereby, the templates for the left and right coronary arteries are respectively updated according to the luminance values of the left and right coronary arteries in the current frame.

  The size of the template cut out for the next frame should be around 25 pixels square so that not only the coronary artery but also the tissue around the coronary artery is included if collecting four cavity cross-sectional image data of 320 pixels square. Is appropriate. This allows more accurate tracking of the coronary arteries.

  Next, in step S10, the search region to be a target of template matching is also updated according to the detection position of the left and right coronary arteries as needed. For example, a predetermined range centered on the center position of the template specified by template matching can be set as the search area of the next frame.

  Alternatively, the search area may be updated only between the initial frame and two or more frames. For example, in the case of collecting 320-pixel square four-chamber cross-sectional image data, the size of the search area for two frames or later should be about 70 pixels square so that the left and right coronary arteries will surely be within the search area. It can be set.

  When the template for the next frame and the search area are determined, template matching in step S6 and recording of representative positions of the left and right coronary arteries in step S7 are performed again on the four-chamber cross-sectional image data of the next frame. Then, the update of the template in step S9, the update of the search area in step S10, the template matching in step S6, and the recording of the representative positions of the left and right coronary arteries in step S7 are determined in the determination of step S8 that the next frame does not exist. Repeat until. That is, while updating the template, template matching and recording of the position of the template are sequentially performed on the four-chamber cross-sectional image data of all the target frames.

  When the template matching and the recording of the position of the template on the four-chamber cross-sectional image data of all the frames are completed, it is determined in the determination of step S8 that the next frame does not exist. Then, in step S11, detection of the resting period of the coronary artery based on the positional change of the left and right coronary arteries is performed by the resting period detection unit 41D.

  If the movement distance of the center position of the template between adjacent frames is measured and plotted, a graph as shown in FIG. 7 is obtained. Then, the graph shown in FIG. 7 can be regarded as a change in the movement distance of the representative position of the coronary artery. Therefore, it is possible to extract the section of the frame in which the movement distance of the center position of the template is equal to or less than the threshold as the stationary period of the coronary artery. In the example shown in FIG. 7, the section of the frame in which the movement distance of the center position of the template is zero is extracted as the stationary period of the coronary artery.

  In practice, the static period of the coronary artery may be detected only by numerical calculation including threshold processing without creating a graph as shown in FIG.

  Once the resting period of the coronary artery is detected, it is possible to perform the desired imaging utilizing the resting period of the coronary artery. Magnetic resonance angiography (MRA), which is performed frequently, includes WHCA (Whole-heart Coronary MRA) and delayed enhancement (DE) imaging. When performing both DE imaging and WHCA involving administration of a contrast agent, 10 to 15 minutes from when the contrast agent is administered to the subject P until imaging of the DE image is possible, for shortening the imaging time. In many cases, WHCA is executed and then DE imaging is performed.

  In that case, WHCA using the stationary period of the coronary artery is performed in step S12. Next, in step S13, DE-MRI using the stationary period of the coronary artery is performed.

  In addition to DE imaging and WHCA, there is also a need to perform imaging by FBI (Fresh Blood Imaging) method during a stationary period of a coronary artery. The FBI (Fresh Blood Imaging) method is a non-contrast MRA that repeatedly collects echo data for each of a plurality of heartbeats by delaying a trigger signal synchronized with a reference wave representing the cardiac phase of the subject P such as R wave for a predetermined time. is there.

  In other words, the magnetic resonance imaging apparatus 20 as described above narrows down the estimated positions of the left and right coronary arteries based on the detected positions of the three feature points of the mitral valve, tricuspid valve and apex, and template matching limited to the search area Position change and coronary artery position change are required to obtain a small stationary phase. Furthermore, the magnetic resonance imaging apparatus 20 tracks the coronary artery depicted in the cine image while updating the template used for template matching to the image data immediately before the coronary artery is depicted.

  Therefore, according to the magnetic resonance imaging apparatus 20, the resting period of the coronary artery can be easily and automatically detected. That is, it is possible to omit the troublesome operation conventionally required for the engineer to manually designate the ROI corresponding to the coronary artery while viewing the image. This leads to shortening of the inspection time.

  In addition, even if the shape of the coronary artery changes with time phase, the template can be updated following the shape of the coronary artery. Therefore, the coronary artery can be reliably tracked by template matching. As a result, the resting period of the coronary artery can be determined with the same accuracy as that of the technician.

  Although the specific embodiments have been described above, the described embodiments are merely examples and do not limit the scope of the invention. The novel methods and apparatus described herein may be embodied in various other ways. Also, various omissions, substitutions and changes in the form of the methods and apparatus described herein may be made without departing from the spirit of the invention. The appended claims and their equivalents are intended to cover such different forms and modifications as fall within the scope and spirit of the invention.

  For example, in the above-described embodiment, after the stationary period of the coronary artery is specified, another imaging is performed in the stationary period of the coronary artery as an example. However, after collecting diagnostic image data, the stationary period of the coronary artery is detected as post-processing In such a case, the stationary period can be detected in the same manner. Therefore, the medical image processing apparatus connected to the magnetic resonance imaging apparatus 20 via the network may be provided with a function of specifying the resting period of the coronary artery based on the position of the feature point of the heart and the MR cine image data.

  In the embodiment described above, the case of using the correlation coefficient as an index of the degree of coincidence in template matching has been described as an example, but other indices of coincidence such as sum of squared residuals (SSR) may be used. You may use.

  In the embodiment described above, volume image data covering the heart is acquired in the systolic or diastolic resting phase, but the heart is covered in the systolic and diastolic resting phases. Volume image data may be collected. Then, it becomes possible to automatically detect the position of the heart feature point corresponding to the systolic stationary phase and the position of the heart feature point corresponding to the diastolic stationary phase. For this reason, the position of the feature point automatically detected with higher accuracy can be used for narrowing the initial spatial search region of template matching. In other words, even if the feature points of the heart in either the systolic or diastolic stationary phase are not automatically detected with good accuracy, re-collection of volume image data covering the heart is avoided. can do.

Reference Signs List 20 magnetic resonance imaging apparatus 21 magnet for static magnetic field 22 shim coil 23 gradient magnetic field coil 24 RF coil 25 control system 26 static magnetic field power source 27 gradient magnetic field power source 28 shim coil power source 29 transmitter 30 receiver 31 sequence controller 32 computer 33 input device 34 display device 35 arithmetic unit 36 storage unit 37 bed 38 ECG unit 40 imaging condition setting unit 41 data processing unit 42 k space data storage unit 43 image data storage unit 44 dictionary file storage unit P object

Claims (10)

An image acquisition unit for acquiring magnetic resonance image data covering a heart of a subject, and magnetic resonance cine image data depicting the coronary artery of the subject;
A feature point detection unit that automatically detects the position of a feature point of the heart based on magnetic resonance image data covering the heart;
At least an initial spatial search area in the magnetic resonance cine image data is automatically set based on the position of the feature point, and the magnetic field is subjected to template matching for a plurality of initial search areas corresponding to a plurality of frames. A coronary artery detection unit for detecting a change in a frame direction of a representative position of the coronary artery from resonance cine image data;
A stationary period detection unit that detects a period during which the coronary artery can be considered to be stationary based on a change in the frame direction of the representative position of the coronary artery;
Equipped with
The coronary artery detection unit is configured to select a template having the largest matching score as a result of the template matching for the initial spatial search area among the plurality of templates simulating the shape of the coronary artery as the template for the initial spatial search area. A magnetic resonance imaging apparatus configured to be employed as a template for matching .   The magnetic resonance imaging apparatus according to claim 1, wherein the coronary artery detection unit is configured to use a part of the magnetic resonance cine image data as at least one template for the template matching.   The coronary artery detection unit determines a template for the template matching, which targets a search area after the initial spatial search area, identified by the past template matching, and the magnetic resonance of the area where the coronary artery is drawn The magnetic resonance imaging apparatus according to claim 1, wherein the cine image data is configured to be updated at a predetermined frame interval.   The coronary artery detection unit is configured to update a search area after the initial spatial search area at a predetermined frame interval based on a representative position of the coronary artery detected in the past. The magnetic resonance imaging apparatus of any one of 3.   The magnetic resonance imaging apparatus according to any one of claims 1 to 4, wherein the coronary artery detection unit is configured to use a template corresponding to a region including a tissue around a coronary artery for the template matching.   The coronary artery detection unit detects at least an initial spatial search area of the right coronary artery and an initial spatial search area of the left coronary artery based on the positions of the apex of the heart, the tricuspid valve, and the mitral valve detected as the positions of the feature points. The magnetic resonance imaging according to any one of claims 1 to 5, wherein the magnetic resonance imaging is configured to set and detect a change in the frame direction of the representative position of the right coronary artery and a change in the frame direction of the representative position of the left coronary artery. apparatus.   The coronary artery detection unit is a first line segment connecting the apex and the tricuspid valve, a second line segment connecting the tricuspid valve and the estimated position of the right coronary artery, and Assuming that a line connecting the apex of the heart and the mitral valve is a third line, and a line connecting the mitral valve and the estimated position of the left coronary artery is a fourth line, the first line A region based on a point specified by an empirical value of a ratio of a length of the second line segment to a length of the second line segment and an empirical value of an angle formed by the first line segment and the second line segment The experiential value of the ratio of the length of the third line segment to the length of the fourth line segment and the third line segment and the fourth line segment while setting the initial spatial search region of the right coronary artery 7. The magnetic joint device according to claim 6, wherein the initial spatial search region of the left coronary artery is set as a region based on a point specified by an experience value of an angle between the two. Imaging apparatus.   The coronary artery detection unit is a first line segment connecting the apex and the tricuspid valve, a second line segment connecting the tricuspid valve and the estimated position of the right coronary artery, and Assuming that a line connecting the apex of the heart and the mitral valve is a third line, and a line connecting the mitral valve and the estimated position of the left coronary artery is a fourth line, the first line The right coronary artery specified corresponding to the possible range of the ratio of the length of the second line segment to the length of the second line segment and the possible range of the angle between the first line segment and the second line segment The possible range of the position of is set in the initial spatial search region of the right coronary artery, while the possible range of the ratio of the length of the third line segment to the length of the fourth line segment and the third The possible range of the position of the left coronary artery identified corresponding to the possible range of the angle formed by the line segment and the fourth line segment is the initial spatial search of the left coronary artery The magnetic resonance imaging apparatus configured according to claim 6, wherein to set the area. The image acquisition unit is configured to acquire magnetic resonance imaging data covering the heart during the middle diastole of the heart,
The magnetic resonance imaging apparatus according to any one of claims 1 to 8, wherein the coronary artery detection unit is configured to set the initial spatial search region in magnetic resonance cine image data in the middle diastole of the heart. An image acquisition unit for acquiring image data covering a heart of a subject, and cine image data in which a coronary artery of the subject is depicted;
A feature point detection unit that automatically detects the position of a feature point of the heart based on image data covering the heart;
At least an initial spatial search area in the cine image data is automatically set based on the positions of the feature points, and the cine image data is generated by template matching targeting the initial and subsequent search areas corresponding to a plurality of frames. A coronary artery detection unit that detects a change in the frame direction of the representative position of the coronary artery from the
A stationary period detection unit that detects a period during which the coronary artery can be considered to be stationary based on a change in the frame direction of the representative position of the coronary artery;
Equipped with
The coronary artery detection unit is configured to select a template having the largest matching score as a result of the template matching for the initial spatial search area among the plurality of templates simulating the shape of the coronary artery as the template for the initial spatial search area. An image processing apparatus configured to be employed as a template for matching.

Images