Magnetic resonance imaging apparatus and method for adjusting the center frequency

Abstract

To improve the accuracy of setting the center frequency. [Solution] The MRI apparatus according to this embodiment has a processing circuit that acquires a first dataset collected based on a first sequence condition that suppresses at least one chemical species to a first suppression state, and a second dataset collected based on a second sequence condition that suppresses the chemical species to a second suppression state, performs fitting on the first dataset using a first model kernel corresponding to the first signal shape of the first dataset to obtain a first fitting result, performs fitting on the second dataset using a second model kernel corresponding to the second signal shape of the second dataset to obtain a second fitting result with the same frequency shift value, and corrects the center frequency based on the first and second fitting results. At least one chemical species is suppressed in different ways in the first and second suppression states.

Description

[Technical Field]

[0001] Embodiments disclosed herein and in the drawings relate to magnetic resonance imaging apparatuses and methods for adjusting the center frequency. For example, embodiments disclosed herein generally relate to methods, systems, processing circuits, and computer program products for controlling the magnetic resonance imaging (MRI) apparatus described herein. In one embodiment, the embodiments also relate to methods, systems, processing circuits, and computer program products for controlling the center frequency used for acquiring MRI image data. [Background technology]

[0002] In known MRI systems, a pre-scan process is generally performed on the clinical MR scanner to calibrate the center frequency (CF) for the specific subject being scanned. Because each patient has a unique B0 magnetic field distortion, the center frequency needs to be calibrated for each patient and each RF coil. Furthermore, the calibration is specific to the biological structure being scanned. Inaccurate CF calibration can result in reduced water signal, insufficient fat suppression, and / or loss of image quality (IQ) due to geometric inaccuracies.

[0003] As part of MRI processing, different chemical species (e.g., water, aliphatic fats, olefin fats, silicon, etc.) have different resonance frequencies depending on their chemical structure. Interspecies frequency separation is known a priori based on physical chemical models and laboratory experiments. Generally, since most MR imaging focuses on water content in tissues, MR scanners aim to select the water frequency as the focal frequency (CF).

[0004] In a known CF calibration method, the CF is simply assigned to the frequency of the peak signal in the prescan spectrum. However, this method may incorrectly identify the CF if the scanned anatomically specific prescan area contains more fat than water, such as in fatty tissue.

[0005] Another known CF calibration method applies model fitting that constructs a model kernel of an ideal peak. The kernel can include fat and water (or fat, water, and silicon) each constituted by a predetermined chemical separation. In such a configuration, the kernel is convolved with the measurement spectrum. The measurement spectrum is obtained by shifting the spectrum in frequency steps and calculating the cross-correlation value with the fat suppression mode by the inversion recovery method (often referred to as the STIR mode) turned off (STIR off ). The shift value corresponding to the highest cross-correlation value is selected as the adjustment value to be applied to CF. As shown in FIG. 1A, in the above-described model-based method, since the model utilizes the elliptical shape of the predicted spectrum, accurate CF prediction is possible even when there is a certain amount of noisy data. Alternatively, as shown in FIG. 1B, when the fat signal is significantly higher than the water signal, even if the predicted CF is actually within the fat region, the cross-correlation becomes unintentionally the highest, and thus the process ends in failure.

[0006] As used herein, "STIR on " means applying an RF pulse of a non-selective fat suppression method (short tau inversion recovery: STIR) to the sample. By utilizing the difference in the longitudinal relaxation times of fat and water (the longitudinal relaxation time of fat is shorter), an inversion recovery time (TI) is selected such that most of the fat signal is suppressed and most of the water signal remains as it is. In the case of STIR off , there is no inversion recovery RF pulse. Therefore, in STIR off data, neither the fat signal nor the water signal is suppressed.

[0007] Another known method used by the applicant's known MRI system (disclosed in U.S. Patent No. 9,662,037) utilizes information from two acquired spectra. One of the spectra is acquired with STIR turned off (STIR off ), and the other is acquired with fat suppression by inversion recovery turned on (STIR on) is obtained. Next, each STIR on spectrum and STIR off frequency F0 in the spectrum on and F0 off , that is, peak signal S1 on and S1 off are respectively identified to determine the estimated value of the water frequency of each spectrum. This process usually starts from the initial estimated values (F0 on , F0 off ), analyzes the spectrum from -3 to -4 ppm downfield, and based on the fact that the chemical shift of fat relative to the position of water in the spectrum is -3.5 ppm, identifies the peak signals (S2 on , S2 off ) in that region. That is, the ideal positions of the peak signals are S2 on @F0 on -3.5 ppm and S2 off @F0 off -3.5 ppm. Next, this process facilitates the determination of the corrected value of CF using the values of S1 on , S1 off , S2 off , and S2 on . For example, when the fat signal is appropriately suppressed in the STIR on spectrum as expected, (S2 off / S2 on ) > (S1 off / S1 on ), so it is confirmed that CF is F0 on .

[0008] However, by using the peak signals in the spectral region that is estimated to correspond to water and fat, this process may mispredict the center frequency. For example, as shown in Figure 1C, for a more appropriate prediction in reality, although the center frequency is at the center of the broad water peak, this process may detect the peak near the edge of that broad peak as the center frequency. In the case of Figure 1D, due to the presence of a large fat signal, the process makes an incorrect prediction with the fat frequency (left peak) instead of the water frequency (right peak) as the center frequency. [Prior art documents] [Patent Documents]

[0009] [Patent Document 1] Patent No. 5858716 [Patent Document 2] U.S. Patent No. 9662037 [Overview of the project] [Problems that the invention aims to solve]

[0010] One of the problems that the embodiments disclosed herein and in the drawings aim to solve is to improve the accuracy of setting the center frequency. However, the problems that the embodiments disclosed herein and in the drawings aim to solve are 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]

[0011] The magnetic resonance imaging apparatus according to this embodiment includes a processing circuit. The processing circuit acquires a first dataset collected based on a first sequence condition that suppresses at least one chemical species to a first suppression state, and a second dataset collected based on a second sequence condition that suppresses the chemical species to a second suppression state. The processing circuit performs fitting on the first dataset using a first model kernel corresponding to the first signal shape of the first dataset and acquires a first fitting result. The processing circuit performs fitting on the second dataset using a second model kernel corresponding to the second signal shape of the second dataset and acquires a second fitting result with the same frequency shift value. The processing circuit corrects the center frequency based on the first fitting result and the second fitting result. The at least one chemical species is suppressed in different ways in the first suppression state of the first sequence and the second suppression state of the second sequence. [Brief explanation of the drawing]

[0012] [Figure 1A] Figure 1A shows an overview of the model-based cross-correlation process used to identify the water frequency to be used as the center frequency. [Figure 1B] Figure 1B shows an overview of a model-based cross-correlation process that misidentifies water frequencies due to the presence of peak signals originating from adipose tissue. [Figure 1C] Figure 1C shows an overview of the algorithmic process that incorrectly identifies the water frequency (used as the center frequency) because it considers the peak beyond the noisy spectrum, which includes a broad water peak. [Figure 1D] Figure 1D shows an overview of the algorithmic process that incorrectly identifies the water frequency (used as the center frequency) because it considers the high fat peak relative to the water peak. [Figure 2] Figure 2 is a schematic diagram of an MRI system. [Figure 3] Figure 3 is a flowchart illustrating the overview of the center frequency correction process described herein. [Figure 4] Figure 4 shows an example of a pseudocodebase for the center frequency correction value determination process. [Figure 5A] Figure 5A is a table showing two example kernels to be used with spectra obtained under different suppression conditions. [Figure 5B] Figure 5B is a graph showing two kernel examples from Figure 5A, used with spectra obtained under different suppression conditions. [Modes for carrying out the invention]

[0013] In this specification, “one” is defined as one or more; “multiple” is defined as two or more; and “other” is defined as at least the second and subsequent. The expressions “including” and / or “having” are defined as “equipped with” (i.e., non-restrictive terms). Throughout this specification, “one embodiment,” “multiple embodiments,” “embodiments,” “examples,” “examples,” or other similar expressions mean that certain features, structures, or characteristics described in relation to the applicable embodiment are included in at least one embodiment of this disclosure. That is, such expressions found in many places in this specification do not necessarily mean the same embodiment. Furthermore, certain features, structures, or characteristics can be combined in any way as appropriate in one or more embodiments, without limitation.

[0014] This disclosure relates to a method, system, and a non-temporary computer-readable storage medium for storing computer-readable instructions for correcting the center frequency used by an MRI system when acquiring an MRI image by jointly applying multiple models to multiple spectra acquired under first and second sequence conditions based on different suppression conditions.

[0015] In one embodiment, this disclosure can be considered as a system. An MRI apparatus is used as an example of an embodiment, but other system configurations may use other medical imaging devices (e.g., CT systems and integrated MRI and CT systems).

[0016] Next, refer to the drawings. Figure 2 is a block diagram showing the overall configuration of MRI apparatus 1. MRI apparatus 1 comprises a gantry 100, a control cabinet 30, a console 40, a patient table 50, and a radio frequency (RF) coil 20. The gantry 100, control cabinet 30, and patient table 50 constitute the scanner, or imaging unit.

[0017] The gantry 100 comprises a static magnetic field magnet 10, a gradient magnetic field coil 11, and a whole body (WB) coil 12, and these components are housed in a cylindrical casing. The bed 50 includes a bed body 52 and a table 51.

[0018] The control cabinet 30 includes three gradient coil power supplies 31 (31x for the X-axis, 31y for the Y-axis, and 31z for the Z-axis), a coil selection circuit 36, an RF receiver 32, an RF transmitter 33, and a sequence controller 34.

[0019] The console 40 includes a processing circuit 45, memory 41, display 42, and input interface 43. The console 40 functions as a host computer.

[0020] The static magnetic field magnet 10 of the gantry 100 has a roughly cylindrical shape and generates a static magnetic field inside the bore through which objects such as patients are moved. The bore is the space within the cylindrical structure of the gantry 100. The static magnetic field magnet 10 has a superconducting coil inside. The superconducting coil is cooled to extremely low temperatures by liquid helium. In excitation mode, the static magnetic field magnet 10 generates a static magnetic field by supplying current to the superconducting coil from a static magnetic field power source (not shown). Subsequently, the static magnetic field magnet 10 transitions to persistent current mode, and the static magnetic field power source is disconnected. Once in persistent current mode, the static magnetic field magnet 10 continues to generate a strong static magnetic field for a long period of time, for example, more than one year.

[0021] Furthermore, the gradient magnetic field coil 11 also has a roughly cylindrical shape and is fixed inside the static magnetic field magnet 10. The gradient magnetic field coil 11 uses current supplied from the gradient magnetic field coil power supplies 31x, 31y, and 31z to apply gradient magnetic fields (e.g., gradient pulses) to an object in the X, Y, and Z axis directions, respectively.

[0022] The bed body 52 of the bed 50 is capable of moving the table 51 vertically and horizontally. Before imaging, the bed body 52 moves the table 51 on which the object is placed to a predetermined height. Then, when imaging the object, the bed body 52 moves the table 51 horizontally to move the object into the bore.

[0023] The WB coil 12 has a roughly cylindrical shape that surrounds the object and is fixed inside the gradient coil 11. The WB coil 12 applies RF pulses transmitted from the RF transmitter 33 to the object. The WB coil 12 also receives magnetic resonance (MR) signals emitted from the object due to the excitation of hydrogen nuclei.

[0024] As shown in Figure 2, the MRI apparatus 1 may include RF coils 20 in addition to the WB coil 12. Each RF coil 20 is a coil placed near the surface of the object. There are various types of RF coils 20. For example, as shown in Figure 2, types of RF coils 20 include body coils attached to the chest, abdomen, and legs of the object, and spinal coils attached to the back of the object. Another type of RF coil 20 is a head coil for imaging the head of the object. Most RF coils 20 are for receiving only, but some RF coils 20, such as head coils, perform both transmission and reception. The RF coils 20 can be attached to and detached from the table 51 via cables.

[0025] The RF transmitter 33 generates each RF pulse based on instructions from the sequence controller 34. The generated RF pulses are transmitted to the WB coil 12 and applied to the object. The application of one or more RF pulses causes the object to emit an MR signal. Each MR signal is received by the RF coil 20 or the WB coil 12.

[0026] The MR signal received by the RF coil 20 is transmitted to the coil selection circuit 36 ​​via a cable installed on the table 51 and the bed body 52. ​​The MR signal received by the WB coil 12 is also transmitted to the coil selection circuit 36.

[0027] The coil selection circuit 36 ​​selects the MR signal output from each RF coil 20 or the MR signal output from the WB coil 12 according to the control signal output from the sequence controller 34 or console 40.

[0028] The selected MR signal is output to the RF receiver 32. The RF receiver 32 performs analog-to-digital (AD) conversion of the MR signal and outputs the converted signal to the sequence controller 34. The digital MR signal is sometimes called raw data. AD conversion is performed within each RF coil 20 or within the coil selection circuit 36.

[0029] The sequence controller 34 scans for objects by driving the inclined coil power supply 31, RF transmitter 33, and RF receiver 32 under the control of the console 40. The sequence controller 34 receives raw data from the RF receiver 32 as the scan is performed and transmits the received raw data to the console 40.

[0030] The sequence controller 34 includes a processing circuit (not shown). The processing circuit is configured as, for example, a processor that executes a predetermined program, or as hardware such as a field programmable gate array (FPGA) or an application-specific integrated circuit (ASIC).

[0031] As described above, the console 40 includes a memory 41, a display 42, an input interface 43, and a processing circuit 45.

[0032] Memory 41 is a recording medium that includes ROM (read-only memory) and RAM (random access memory) in addition to external memory devices such as hard disk drives (HDD) and optical disc drives. Memory 41 stores various programs executed by the processor of the processing circuit 45, as well as various data and information.

[0033] The input interface 43 includes various devices for the operator to input various information and data, and consists of, for example, a mouse, keyboard, trackball, and / or touch panel.

[0034] The display 42 is a display device such as a liquid crystal display panel, a plasma display panel, or an organic EL panel.

[0035] The processing circuit 45 is, for example, a circuit equipped with a central processing unit (CPU) and / or a purpose-specific or general-purpose processor. The processor implements the following functions by executing a program stored in memory 41.

[0036] The processing circuit 45 can be configured as hardware such as an FPGA and an ASIC. The various functions shown below can also be realized by such hardware. Furthermore, the processing circuit 45 can realize various functions by combining hardware processing and software processing based on its own processor and program.

[0037] Figure 3 is a flowchart outlining the process described herein. The center frequency adjustment method 300 is initiated in step 310 and, in prescan, follows different first and second suppression conditions (e.g., STIR on vs STIR off ) Collect the first dataset and the second dataset based on the first and second sequence conditions. In step 320, fit the first dataset using the first model corresponding to the first signal shape of the first dataset and obtain the first fitting result. In step 330, fit the second dataset using the second model corresponding to the second signal shape of the second dataset and obtain the second fitting result. In steps 320 and 330, both datasets (STIR on and STIR offFor the above, the frequency shift (frequency shift value) applied to the model is the same. In step 340, the center frequency used in MRI acquisition (i.e., after prescan) is corrected based on the first fitting result and the second fitting result. This completes the center frequency adjustment method 300, and the system can start the MRI acquisition process using the corrected center frequency.

[0038] More specifically, in step 310, STIR on Spectrum and STIR off The spectrum is the original estimated value F0 of the center frequency. original It is obtained using [this method]. Next, a Fast Fourier Transform (FFT) is performed on the spectrum to obtain the absolute value of the result and create an amplitude spectrum. Each spectrum is individually normalized to match an equivalent absolute value model (e.g., kernel). Estimated value F0 original This is pre-configured and stored in memory 41.

[0039] In other words, in step 310, the processing circuit 45 acquires a first dataset collected based on a first sequence condition that suppresses at least one chemical species to a first suppression state. The first sequence condition is, for example, a sequence condition corresponding to STIR on. In this case, at least one chemical species corresponds to fat, and the first suppression state corresponds to fat suppression (STIR on). In these cases, the first dataset is STIR on It corresponds to the spectrum.

[0040] The processing circuit 45 obtains a first dataset by, for example, normalizing the amplitude spectrum corresponding to STIR ON by the maximum value of the amplitude spectrum. Normalized STIR corresponding to the first dataset. on The spectrum corresponds to the first signal shape. In this case, since the maximum value of the first signal shape is 1, the maximum value of the first model kernel, which is implemented as a kernel or double Gaussian, is set to 1, for example.

[0041] Furthermore, in step 310, the processing circuit 45 acquires a second dataset collected based on a second sequence condition that suppresses at least one chemical species to a second suppression state. If the first sequence condition is a sequence condition corresponding to STIR on, the second sequence condition is a sequence condition corresponding to STIR off. In this case, at least one chemical species corresponds to fat, and the second suppression state corresponds to fat non-suppression (STIR off). In these cases, the second dataset is STIR off It corresponds to the spectrum.

[0042] The processing circuit 45 obtains a second dataset by, for example, normalizing the amplitude spectrum corresponding to STIR off by the maximum value of that amplitude spectrum. The normalized STIR corresponding to the second dataset. off The spectrum corresponds to the second signal shape. In this case, since the maximum value of the second signal shape is 1, the maximum value of the second model implemented as a kernel or double Gaussian is set to 1, for example. These then give rise to the estimated center frequency F0, which is the difference in signal values ​​caused by STIR on and STIR off. original This can reduce the impact on the correction.

[0043] In steps 320 and 330, two kernels (kernels) corresponding to the two models (first model kernel and second model kernel) on and kernel off ) are the corresponding amplitude spectra STIR on and STIR off Shifted to (F0 original (compared to the previous example). For each kernel × spectrum after the shift, the dot product of each point i is calculated as shown in the following equation.

[0044] a(i) = kernel on (i + shift) × STIR on (i) b(i) = kernel off(i + shift) × STIR off (i)

[0045] Next, each point ((sign)a 2 (i) and (sign)b 2 (i)) is squared, and a polarity is assigned to each point i based on the shifted kernel polarity (signed L2 norm) of that point. Here, a 2 The sign assigned to (i) is the first model at each point i (e.g., the first kernel) on This corresponds to the sign (positive or negative) of ). Also, b 2 The sign assigned to (i) is the second model at each point i (e.g., the second kernel) off This corresponds to the sign (positive or negative) of ).

[0046] In step 340, all points in each convolution dataset are calculated using i(A=Σ(sign)a 2 B = Σ(sign)b 2 The values ​​are summed up, and the summed values ​​are combined with weighting coefficients k1 and k2 so that the weighted sum Q is given by the following equation. The weighting coefficients k1 and k2 are pre-set and stored in memory 41. The convolution dataset shows, for example, the convolution operation between the first dataset and the first model kernel, and the convolution operation between the second dataset and the second model kernel.

[0047] For example, the processing circuit 45 performs fitting on the first dataset using a first model kernel corresponding to the first signal shape of the first dataset, and obtains a first fitting result. Specifically, the processing circuit 45 obtains a series of first fitting results by performing a convolution operation between the first dataset and the first model kernel. The processing circuit 45 also performs fitting on the second dataset using a second model kernel corresponding to the second signal shape of the second dataset, and obtains a second fitting result with the same frequency shift value. Specifically, the processing circuit 45 obtains a series of second fitting results by performing a convolution operation between the second dataset and the second model kernel.

[0048] Q = k1A + k2B

[0049] The shift value that results in the highest Q value corresponds to the frequency correction coefficient ΔF, where CF = F0 original The equation +ΔF holds true. That is, the processing circuit 45 corrects the center frequency CF used to acquire MRI image data based on the first fitting result and the second fitting result. For example, the processing circuit 45 obtains a weighted sum of the series of first fitting results and the series of second fitting results. Specifically, the processing circuit 45 obtains (calculates) the weighted sum by using a larger weight for the fitting result corresponding to the model kernel in the higher suppression state among the first and second model kernels. Then, the processing circuit 45 corrects the center frequency CF to correspond to the frequency with the largest weighted sum among the weighted sums obtained for the series of first and second fitting results.

[0050] Figure 4 shows an example of a pseudocodebase for the center frequency correction value determination process. As shown in the figure, the spectrum is stored in the arrays stirON and stirOFF. The spectrum is normalized using its respective maximum values ​​maxValON and maxValOFF. The process shifts the data step by step and calculates the dot products ccOn and ccOFF for each shift using the normalized spectra (stirON, stirOFF) and the corresponding kernels (kernelON, kernelOFF). A weighted sum is calculated for each shift and added to position ii (corresponding to the current shift) of the weighted sum array data (data). By classifying the values ​​in the array data (data), the process finds the shift value corresponding to the highest weighted sum and sets the calculated shift value back to the original center (F0 original It is used as a correction factor for ).

[0051] In one embodiment, the kernel is generated (configured) to easily identify a single peak. Figure 5A shows two example kernels in tabular form with spectra acquired under different suppression states (e.g., stirON, stirOFF). Figure 5B shows the two example kernels from Figure 5A in graphical form with spectra acquired under different suppression states. Both figures show that the kernel is designed to highlight the difference between water and fat separated by known frequencies. In the illustrated embodiment, a negative amplitude centered on the frequency representing fat is used for STIR on By utilizing the kernel (dashed line in Figure 5B), the first model leverages a first kernel having a first shape corresponding to the relative predicted signal positions (e.g., 3.5 ppm) of at least two chemical species (e.g., water and fat) in a first suppression state of the first sequence condition. For example, the first model kernel corresponds to a first kernel having a first shape (dashed line in Figure 5B) corresponding to the relative predicted signal positions of at least two chemical species in a first suppression state (STIR on) of the first sequence condition. As shown by the dashed line in Figure 5B, this negative fat amplitude prevents incorrectly assigning F0 to the fat peak.

[0052] STIROFF The kernel (solid line in Figure 5B) is configured to identify both fat and water by using a second model that includes a second kernel of a second shape corresponding to the relative predicted signal positions of at least two chemical species in a second suppression state of the second sequence condition. The second model kernel corresponds to a second kernel having a second shape corresponding to the relative predicted signal positions of at least two chemical species in a second suppression state (STIR off) of the second sequence condition. OFF The kernel, as shown by the solid line in Figure 5B, uses the same amplitude for fat and water to prevent weighting of solutions that would incorrectly assign CF to strong fat peaks. It also prevents errors for broadband fat or water peaks by having similar bandwidths (BW). During the generation of the weighted sum, STIR is used. off STIR on the data on Weighting the data more than it actually is useful for detecting water peaks. For example, empirically determined values ​​of k1=1 and k2=2 are helpful for detecting water peaks.

[0053] Alternatively, other forms of weighting functions are also available. For example, ((sign)a 2 (i) and (sign)b 2 (i)) or ((sign)a N (i) and (sign)b N (i)) is available. N is any non-zero number.

[0054] Generally, STIR off The kernel is intended to find two peaks (fat and water), so it needs to have a positive water amplitude and a positive fat amplitude. On the other hand, STIR onThe kernels need to have positive water amplitudes and negative fat amplitudes to facilitate single peak identification. For example, the first and second kernels include at least two kernels with different amplitude polarities, as shown in Figures 5A and 5B. The negative fat amplitude acts as a penalty term for fat. This makes it easier to obtain a total value for identifying water. The first and second kernels may also include at least two kernels with the same amplitude polarity.

[0055] The method described above combines the advantages of model fitting with physical information obtained through STIRon vs. STIRoff. In a set of 40 test samples from subjects (head, pelvis, cervical spine, thoracic spine, shoulder, wrist), the two kernel-based models were successful in all 40 cases, whereas the applicant's previous algorithm was successful in only 39 of the 40 cases.

[0056] Besides using STIRon and STIRoff as mechanisms to emphasize the difference between fat and water, at least three other possible species identification mechanisms include the following: First, use on / off of the saturation recovery pulse instead of on / off of the inversion recovery pulse. For example, the first and second suppression states may correspond to the on and off of the saturation recovery pulse. The fact that fat recovery is faster than water leads to the suppression of water. Second, use a long time between the excitation pulse and the readout (referred to as the time of echo (TE)) for a short TE. For example, the first and second suppression states may correspond to a long and short echo time (TE). It is known that the T2 of fat is shorter than that of water. Therefore, a dataset with a long TE will contain a suppressed fat signal and a strong water signal, while a dataset with a short TE will contain a strong fat signal and a strong water signal. Third, the TE time can be selected to be in-phase or out-of-phase based on the frequency characteristics of the chemical species. For example, the first and second suppression states may correspond to in-phase echo times (TE) and out-of-phase echo times. The water signal and fat signal are additive; in the case of in-phase data (e.g., TE = 2.2 ms at 3T), the fat peak is stronger. In the case of in-phase data (e.g., 3.4 ms at 3T), the water signal and fat signal partially cancel each other out, partially suppressing the fat peak.

[0057] The methods and systems described herein are implementable in many arts but generally relate to imaging devices and processing circuits that perform the processing described herein. In one embodiment, the processing circuit (e.g., an image processing circuit and a control circuit) is implemented as one or a combination of the following: an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a generic array of logic (GAL), a programmable array of logic (PAL), a circuit that allows a logic gate (e.g., using a fuse) or a reprogrammable logic gate to be programmed once. The processing circuit may also include a computer processor and embedded and / or external non-volatile computer-readable memory (e.g., RAM, SRAM, FRAM®, PROM, EPROM, and / or EEPROM). The memory stores computer instructions (binary execution instructions and / or interpreted computer instructions) that control the computer processor to perform the processing described herein. Computer processor circuits enable the creation of a single processor or a multiprocessor, each with one or more cores, each corresponding to one or more threads.

[0058] Furthermore, embodiments of this disclosure may be provided for in the following supplementary matters.

[0059] (1) The center frequency correction method for a magnetic resonance imaging apparatus includes, but is not limited to, the following operations: (a) acquiring a first dataset collected based on a first sequence condition in which at least one chemical species is suppressed according to a first suppression state, and (b) acquiring a second dataset collected based on a second sequence condition, wherein at least one chemical species is suppressed in different ways in the first suppression state of the first sequence and the second suppression state of the second sequence. The first fitting result is obtained by performing a fitting on the first dataset using a first model corresponding to the first signal shape of the first dataset, When fitting the first dataset and the second dataset, the second dataset is fitted using a second model corresponding to the second signal shape of the second dataset, and a second fitting result is obtained with the same frequency shift value, and Correcting the center frequency based on the first fitting result and the second fitting result.

[0060] (2) In the method described in (1), the first model comprises a first model kernel having a first shape corresponding to the relative predicted signal positions of at least two chemical species in the first suppression state of the first sequence condition.

[0061] (3) In the method described in (2), the second model comprises a second model kernel having a second shape corresponding to the relative predicted signal positions of the at least two chemical species in the second suppression state of the second sequence condition.

[0062] (4) In any one of the methods of (1)-(3), obtaining the first fitting result by fitting the first dataset with a first model corresponding to the first signal shape of the first dataset includes obtaining a series of first fitting results by convolving the first dataset into the first model.

[0063] In the method described in (5)(4), obtaining the second fitting result by fitting the first dataset with a second model corresponding to the second signal shape of the second dataset includes obtaining a series of second fitting results by convolving the second dataset into the second model.

[0064] (6)(5) The method of correcting the center frequency based on the first fitting result and the second fitting result includes, but is not limited to, obtaining a weighted sum of the series of first fitting results and the second series of fitting results, and correcting the center frequency to correspond to the frequency having the largest weighted sum among the weighted sums obtained for the first and second series of fitting results.

[0065] (7) In the method of (6), obtaining a weighted sum of the first set of fitting results and the second set of fitting results includes, in no limitation, using greater weights for the fitting results of the first and second fitting results that correspond to the model of the first model and the model of the second model that is in a higher state of suppression.

[0066] (8) In the method described in any one of (1)-(7), the first suppression state and the second suppression state are STIR on and STIR off.

[0067] (9) In the method of any one of (1)-(7), the first suppression state and the second suppression state are the saturation recovery pulse being on and the saturation recovery pulse being off.

[0068] (10) In the method described in any one of (1)-(7), the first suppression state and the second suppression state are a long echo time (TE) and a short echo time.

[0069] (11) In the method described in any one of (1)-(7), the first suppression state and the second suppression state are in-phase echo time (TE) and out-of-phase echo time.

[0070] In the method described in any one of (12)(3)-(7), the first model kernel and the second model kernel include at least two model species having the same amplitude polarity.

[0071] In the method described in any one of (13)(3)-(7), the first model kernel and the second model kernel include at least two model species having different amplitude polarities.

[0072] (14) A device for correcting the center frequency of a magnetic resonance imaging apparatus, which non-limitingly includes a processing circuit that performs any one of the steps described in (1) to (13).

[0073] (15) A non-temporary computer-readable storage medium that stores computer-readable instructions causing a computer to perform any one of the image processing methods described in (1)-(13).

[0074] According to the embodiments described above, the accuracy of setting the center frequency can be improved.

[0075] While several embodiments have been described, these embodiments are presented as examples only and are not intended to limit the scope of the invention. The novel methods, apparatus, and systems can be implemented 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]

[0076] 1 MRI machine 10 Static magnetic field magnet 11. Gradient field coils 12 Whole Body (WB) Coils 20 Radio Frequency (RF) Coils 30 Control Cabinets 31. Power supply with three gradient magnetic field coils 32 RF receivers 33 RF transmitter 34 Sequence Controller 36 Coil Selection Circuit 40 Console 41 memory 42 displays 43 Input Interfaces 45 Processing Circuit 50 berths 51 Tables 52 Bed frame 100 Gantry

Claims

1. A first dataset is obtained based on a first sequence condition that suppresses at least one chemical species to a first suppression state, and a second dataset is obtained based on a second sequence condition that suppresses the chemical species to a second suppression state. The first dataset is fitted using a first model kernel corresponding to the first signal shape of the first dataset, and the first fitting result is obtained. Using a second model kernel corresponding to the second signal shape of the second dataset, fitting is performed on the second dataset to obtain a second fitting result with the same frequency shift value. The system includes a processing circuit that corrects the center frequency based on the first fitting result and the second fitting result. The at least one chemical species is suppressed in different ways in the first suppression state of the first sequence and the second suppression state of the second sequence. Magnetic resonance imaging device.

2. The first model kernel corresponds to a first kernel having a first shape corresponding to the relative predicted signal positions of at least two chemical species in the first suppression state of the first sequence condition, The magnetic resonance imaging apparatus according to claim 1.

3. The second model kernel corresponds to a second kernel having a second shape corresponding to the relative predicted signal positions of the at least two chemical species in the second suppression state of the second sequence condition, The magnetic resonance imaging apparatus according to claim 2.

4. Performing a fitting on the first dataset using a first model kernel corresponding to the first signal shape of the first dataset to obtain the first fitting result includes obtaining a series of the first fitting results by performing a convolution operation between the first dataset and the first model kernel. The magnetic resonance imaging apparatus according to claim 1.

5. Performing a fitting on the second dataset using a second model kernel corresponding to the second signal shape of the second dataset to obtain the second fitting result includes obtaining a series of the second fitting results by performing a convolution operation between the second dataset and the second model kernel. The magnetic resonance imaging apparatus according to claim 4.

6. Correcting the center frequency based on the first fitting result and the second fitting result is: Obtaining a weighted sum of the series of first fitting results and the series of second fitting results, This includes correcting the center frequency to correspond to the frequency having the largest weighted sum among the weighted sums obtained for the series of first fitting results and the series of second fitting results, The magnetic resonance imaging apparatus according to claim 5.

7. Obtaining a weighted sum of the series of first fitting results and the series of second fitting results includes using greater weights for the fitting results corresponding to the model kernels in a higher suppression state among the first and second model kernels. The magnetic resonance imaging apparatus according to claim 6.

8. The first suppression state and the second suppression state are STIR on and STIR off. A magnetic resonance imaging apparatus according to any one of claims 1 to 7.

9. The first suppression state and the second suppression state are the saturation recovery pulse being on and the saturation recovery pulse being off. The magnetic resonance imaging apparatus according to claim 1.

10. The first and second suppression states are long echo times (TE) and short echo times, respectively. The magnetic resonance imaging apparatus according to claim 1.

11. The first and second suppression states are in-phase echo times (TE) and out-of-phase echo times. The magnetic resonance imaging apparatus according to claim 1.

12. The first model kernel and the second model kernel each include at least two kernels having the same amplitude polarity. The magnetic resonance imaging apparatus according to claim 3.

13. The first model kernel and the second model kernel each include at least two kernels having different amplitude polarities. The magnetic resonance imaging apparatus according to claim 3.

14. A first dataset is obtained based on a first sequence condition that suppresses at least one chemical species to a first suppression state, and a second dataset is obtained based on a second sequence condition that suppresses the chemical species to a second suppression state. The first dataset is fitted using a first model kernel corresponding to the first signal shape of the first dataset, and the first fitting result is obtained. During the fitting of the first and second datasets, the second dataset is fitted using a second model kernel corresponding to the second signal shape of the second dataset, and the second fitting result is obtained with the same frequency shift value. Based on the first fitting result and the second fitting result, the center frequency is corrected. The at least one chemical species is suppressed in different ways in the first suppression state of the first sequence and the second suppression state of the second sequence. Center frequency adjustment method.

Images