Flexible echo time water-fat separation method, system, and apparatus based on a main magnetic field map

Abstract

The application discloses a flexible echo time water-fat separation method and system based on a main magnetic field map (B0map) and equipment, and belongs to the technical field of magnetic resonance imaging. The application combines a B0map which needs to be collected in chemical exchange saturation transfer (CEST) imaging, and proposes a new seed point selection and seed point error phase determination method. The seed point error phase determined by the method can grow a relatively accurate error vector map in space, thereby ensuring the final water-fat accurate separation. In the embodiment of the application, the proposed method is verified on a water model and pork, and ideal water-fat separation results are obtained.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of magnetic resonance imaging, and particularly relates to a flexible echo time water-fat separation method, system and equipment based on a main magnetic field map (B0 map). BACKGROUND

[0002] Magnetic resonance imaging (MRI) has become a common diagnostic technique in modern medicine. Due to the characteristics of short T1 and long T2 of human body fat tissue, fat signals are high-brightness signals in MRI, which often mask the diagnosis of edema, tumors, inflammation and other lesions. In addition, the quantification of tissue fat is often used as a clinical indicator for the diagnosis of diseases such as liver, pancreas, adrenal gland and bone marrow. Water-fat separation in magnetic resonance imaging, which can obtain tissue water and fat images simultaneously, has great clinical significance. Compared with the fat suppression techniques such as inversion recovery technique (STIR) and frequency selective saturation technique, water-fat separation technique is less sensitive to the main magnetic field inhomogeneity, and the signal-to-noise ratio of the image result is higher, so it has great application prospect in clinic. With the development of technology, water-fat separation technology has derived single-point, two-point and multi-point Dixon methods. Among them, two-point Dixon technology has been widely used in MRI, especially in dynamic imaging or breath-hold imaging, due to its short acquisition time and flexible echo time adjustment. Two-point Dixon technology is based on the chemical shift between the hydrogen protons of water and fat in human tissue, and can obtain water and fat images respectively. The main problem is how to find the correct error phase caused by field inhomogeneity. Under normal circumstances, the analysis and calculation of the complex images acquired by two echoes can obtain two possible solutions of the error phase of each pixel. The commonly used method to select the two candidate error phases is to use the region growing algorithm. This method considers that the change of the main magnetic field in space is slow, so the error phase caused by the inhomogeneity of the main magnetic field is also smooth in space. The error phase of the "seed point" is known as the starting point of growth, and the error phase of each pixel in the whole image is determined by the region growing algorithm. Therefore, the judgment of the error phase of the "seed point" is crucial, otherwise the wrong error phase will be propagated and accumulated in the process of region growing, resulting in the inversion or residual of the final water and fat images. For the selection of the initial "seed point" of the two-point Dixon method and the determination of the error phase at this point, the commonly used method in the existing technology is the global vector graph gradient method, the basic principle of which can be described as follows: for each pixel in the whole image, two candidate error phases can be solved by mathematical formula, which can be constructed into two vectors a and b respectively. Then a vector graph A and a vector graph B can be constructed from all pixel points. A point with the smallest global gradient in the vector graph A is selected as the initial seed point of region growing, and the phase error of this point is considered as a. However, this method mainly relies on a certain similarity measurement criterion, and too much depends on the assumption that the phase change caused by the inhomogeneity of the main magnetic field is smooth in space.Thus, it is prone to some problems in actual operation, for example, in the case of serious spatial noise or motion artifact, the error vector of the noise point is prone to be selected as interference to make the seed point error phase wrong, in addition, the seed point with the minimum gradient of the global error vector diagram does not cover all non-connected effective image regions, and the growth process needs to cross the background noise region, spread and accumulate errors in space. The misselection of the seed point error phase will cause the error to be transmitted in space during the growth process, resulting in large-area water and fat separation error.

[0003] Therefore, for the two-point water-fat separation method of region growing type, a more accurate method for selecting seed points and confirming seed point phase errors is needed. CEST imaging usually needs to perform B0 field inhomogeneity correction, and multiple methods (such as GRE B0, WASSR and WASABI) are usually used to obtain a B0 map, and it is expected to use the obtained B0 map to optimize the selection of seed points in the region growing method of two-point water-fat separation and the determination process of the whole error phase diagram. SUMMARY

[0004] The purpose of the present application is to solve the above-mentioned problems existing in the prior art, and to provide a flexible echo time water-fat separation method based on a B0 map.

[0005] The specific technical solutions adopted by the present application are as follows:

[0006] In a first aspect, the present application provides a flexible echo time water-fat separation method based on a main magnetic field map (B0 map), which comprises:

[0007] S1, acquiring a main magnetic field map as a first signal image and two echo times TE1 and TE2 corresponding to a second signal image and a third signal image for a magnetic resonance imaging target, and the three types of signal images acquired have the same image pixel domain;

[0008] S2, calculating a first possible solution b1 and a second possible solution b2 of the error phase b of each pixel point in the image pixel domain based on the second signal image and the third signal image; performing phase angle operation on the B0 field offset mapping value of each pixel in the image pixel domain obtained from the main magnetic field map, the first possible solution b1 and the second possible solution b2, respectively, and then aligning the operation results of the three through the echo time difference ΔTE of the second signal image and the third signal image, so as to obtain a first phase angle, a second phase angle and a third phase angle, respectively;

[0009] S3, for each pixel in the image pixel domain, calculate the difference between the second phase angle and the first phase angle as a first deviation angle, and calculate the difference between the third phase angle and the first phase angle as a second deviation angle, each pixel takes the absolute value of the smaller one of the first deviation angle and the second deviation angle as the error phase angle value of the pixel; select one or more pixel points with the smallest error phase angle value from the entire image pixel domain as seed points, and the error phase of the seed point is the smaller one of the absolute value of the first deviation angle and the second deviation angle corresponding to the possible solution;

[0010] S4, for the rest of the pixels in the image pixel domain except the seed points, iteratively calculate the error phase of the neighborhood pixel points of each seed point by region growing algorithm, until the error phase of all pixels in the image pixel domain is obtained, and then the amplitude of water and fat signals is solved to realize water and fat separation.

[0011] As a preferred embodiment of the first aspect, the first signal image is a B0 field inhomogeneity correction map, which is acquired by a double echo gradient sequence in CEST imaging, WASSR (Water Saturation Shift Referencing) technology or WASABI (Simultaneous mapping of water shift and B1) technology.

[0012] As a preferred embodiment of the first aspect, in S2, the phase angle taking operation results of the B0 field offset mapping value corresponding to each pixel, the first possible solution b1 and the second possible solution b2 are aligned as follows:

[0013] The phase angle taking operation result of the B0 field offset mapping value is multiplied by the echo time difference ΔTE of the second signal image and the third signal image to obtain the first phase angle; the phase angle taking operation results of the first possible solution b1 and the second possible solution b2 are respectively divided by the echo time difference ΔTE of the second signal image and the third signal image to obtain the second phase angle and the third phase angle respectively.

[0014] As a preferred embodiment of the first aspect, in S3, the number of seed points selected from the entire image pixel domain is a preset adjustable value, and the minimum is 1.

[0015] As a preferred embodiment of the first aspect, in S4, the specific implementation of the region growing algorithm is as follows:

[0016] S41, for each neighborhood pixel of each seed point in the current image pixel domain which has not yet determined the error phase, select the possible solution with smaller angle between the error phase of the seed point from the first possible solution b1 and the second possible solution b2 of the neighborhood pixel as the error phase of the neighborhood pixel;

[0017] S42, all pixel points of the newly determined error phase are taken as new seed points, and the iteration of S41 is continuously performed until all pixel points in the image pixel domain are determined error phase.

[0018] As a preferred embodiment of the first aspect, in S41, for a two-dimensional image pixel domain obtained by two-dimensional magnetic resonance imaging, each seed point needs to traverse four neighboring pixels when performing the region growing algorithm, and for a three-dimensional image pixel domain obtained by three-dimensional magnetic resonance imaging, each seed point needs to traverse six neighboring pixels when performing the region growing algorithm.

[0019] As a preferred embodiment of the first aspect, in S4, the specific method for solving the amplitudes of the water and fat signals is:

[0020] The error phase of each pixel in the image pixel domain is substituted into the relationship model of the water and fat components and the composite signal, and the amplitude W of the water signal and the amplitude F of the fat signal are solved by using the least square method;

[0021] The relationship model of the water and fat components and the composite signal is:

[0022] S1=(W1F)b0

[0023] 2=(W2F)b0b

[0024] Wherein, S1 and S2 represent the pixel values of the second signal image and the third signal image respectively; 0 is the phase vector of the water signal collected when the echo time is TE1, b represents the error phase of each pixel; and 1 and 2 represent the coefficients caused by the chemical shift of water and fat under the fat multi-peak model.

[0025] Secondly, the present application provides a data processing system, characterized by comprising a memory and a processor;

[0026] The memory is used for storing a computer program;

[0027] The processor is used for realizing the flexible echo time water-fat separation method based on the main magnetic field map according to any one of the first aspect when the computer program is executed.

[0028] Thirdly, the present application provides a magnetic resonance imaging device, comprising a magnetic resonance scanner and a control unit;

[0029] The magnetic resonance scanner is used for collecting the first signal image, the second signal image and the third signal image for a magnetic resonance imaging target;

[0030] The control unit stores a computer program which, when executed, implements the flexible echo time water-fat separation method based on a main magnetic field map according to any one of the first aspect.

[0031] Compared with the prior art, the present application has the following beneficial effects:

[0032] (1) The present application establishes a two-point water-fat signal model, considers a fat multi-peak model, and an error vector b represents an error phase accumulated by field inhomogeneity within a Delta TE.

[0033] (2) The present application uses a commonly used B0 map in CEST imaging to provide a new method for finding seed points for the region growing process of the error phase, because both represent the accumulation effect of the error phase and have high similarity.

[0034] (3) In the present application, the B0 map and the error phase map are connected through the normalized vector angle, and both more closely depict the offset mapping of the B0 field in terms of angle change.

[0035] In summary, the present application combines the commonly needed acquisition of B0 map in CEST imaging, proposes a new method for selecting seed points and determining the error phase of the seed points, and uses the seed point error phase determined by the method to grow a relatively accurate error vector map in space. The method proposed in the embodiments of the present application is verified on a water phantom and pork, and ideal water-fat separation results are obtained. BRIEF DESCRIPTION OF DRAWINGS

[0036] Figure 1 A flowchart of the flexible echo time water-fat separation method based on a B0 map;

[0037] Figure 2 A two-point water-fat separation result diagram of a sunflower seed oil-PBS solution water phantom;

[0038] Figure 3 A two-point water-fat separation result diagram of a fat and lean pork;

[0039] Figure 4 A three-point water-fat separation result diagram of a fat and lean pork. DETAILED DESCRIPTION

[0040] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described in detail below with reference to the drawings. In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without departing from the spirit of the present application, so the present application is not limited to the specific embodiments disclosed below. The technical features in each embodiment of the present application can be combined accordingly without conflict.

[0041] In the description of the present application, it should be understood that the terms "first", "second" are only used for distinguishing description purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the technical features indicated. Therefore, the features defined with "first", "second" can be explicitly or implicitly included at least one of the features.

[0042] The present application proposes a two-point water-fat separation method based on B0 map, which has no limitation on echo time to allow flexible shortening of acquisition time or improving spatial resolution. The model of the signal takes into account the multi-peak model of fat to achieve accurate water-fat separation. The estimation of the phase error vector diagram uses a region growing method, and the B0 map commonly used in chemical exchange saturation transfer (CEST) imaging is innovatively introduced to determine the seed point and the error phase of the seed point.

[0043] First, the basic principles of the two-point water-fat separation method based on B0 map of the present application are described in detail below.

[0044] Due to the difference in the molecular environment of hydrogen protons in human tissue water and fat, the resonance frequency of water and fat is different under the same magnetic field, i.e. chemical shift. The two-point water-fat separation technology is to adjust the echo time by using the chemical shift between water and fat, and water and fat form two magnetization vector angles under two echo times, and by solving two vector triangles composed of water, fat components and complex signals, water and fat are separated. For magnetic resonance imaging targets, two complex signal images S1 and S2 corresponding to two echo times TE1 and TE2 are obtained by acquisition, in each voxel, the relationship model of water, fat components and complex signal can be represented by the following formula (1):

[0045]

[0046] where S1 and S2 represent the two signal images collected, i.e. the vector sum of water and fat signals; W and F represent the amplitude of water and fat signals, respectively. b0 is the phase vector of water signal collected at TE1, b represents the error phase accumulated due to magnetic field inhomogeneity within ΔTE, ΔTE represents the echo time difference of S1 and S2. δ1 and 2 represent the chemical shifts of water and fat under the fat multi-peak model, which are defined according to the following formula (2):

[0047]

[0048] where i is the imaginary unit, γ is the gyromagnetic ratio of hydrogen proton, usually 42.6 MHz / T, B1 is the main magnetic field strength, and δ is the chemical shift of fat (≈3.4 ppm, relative to the water resonance frequency). The parameter μ represents the inverse of the apparent of fat due to spectral broadening, which can be assumed to be known a priori and obtained by fitting the fat multi-peak model.

[0049] In order to solve the correct W and F, the error phase b is required to be obtained first. By defining a fat fraction Q, Q=F / (W+F), and using formula (1), the following formula (2) can be obtained:

[0050]

[0051] where:

[0052]

[0053] From formula (1), the following formula (3) can be obtained:

[0054]

[0055] In combination with the two possible solutions Q1 and Q2 of Q in formula (3), for each pixel position on the image, two possible solutions b1 and b2 of the error phase vector can be obtained, and the process of determining the error phase map is to select a correct vector from (b1, b2) for each pixel point in the image pixel domain, so as to form a correct phase vector map. The correct phase vector map obtained is substituted into formula (1), and the amplitude solutions W and F of water and fat are solved by using the least square method.

[0056] The key to water-fat separation is the confirmation of the error phase map. The commonly used region growing algorithm is based on the smoothness of the phase change between adjacent pixel points within ΔTE, and the seed point is usually selected as the point with the minimum global gradient in the two candidate error phase vector maps, and the error phase corresponding to the vector map is assigned to the seed point. Such a simple similarity measurement criterion depending on the smoothness between pixels is easily disturbed by noise. The present application proposes a method for seed point selection and error phase selection by using the commonly collected B0 map in CEST imaging.

[0057] CEST imaging usually requires B0 field inhomogeneity correction. A common method is to acquire B0 map using double echo gradient sequence (e.g. GRE B0 sequence), which divides the phase difference between B0 phase images obtained at two different echo times by the echo time difference to obtain B0 map, which quantifies the B0 field offset map of each voxel. Another common method for B0 correction is WASSR (Water Saturation Shift Referencing) technique, which applies a RF pulse with small enough power and short enough duration to make the direct water saturation (DS) effect play a major role while the magnetization transfer (MT) and CEST effects are ignored, and then corrects the frequency shift of the free water signal to 0 ppm by observing the frequency shift of the water signal after direct water saturation, thus providing the B0 map. In the two-point water-fat model, b in equation (1) represents the error phase accumulated within ΔTE due to field inhomogeneity, b characterizes the accumulation effect of field inhomogeneity over time, and the B0 map acquired before CEST imaging quantifies the offset of the magnetic field strength of each voxel in the B0 field. Considering the continuous accumulation of error phase over time due to field inhomogeneity, within a certain range, b / ΔTE and B0 map and both of these two parameters characterize the phenomenon of error phase accumulation, and should be approximately equal.

[0058] The application determines the seed point of error phase b before CEST imaging by using B0 map, and the specific method is to normalize the calculated B0 map value and take the angle to obtain angle(B0 map), which represents the phase shift direction angle of the pixel points at each position accumulated in unit time. The candidate error phase b1 and b2 are normalized and written in vector form to avoid phase wrapping, and the angle is taken to obtain angle(b1) and angle(b2), which represent the shift direction angles of the two alternative error phase vectors. In this way, the correct angle(b) / ΔTE has a corresponding relationship with angle(B0 map), and under ideal conditions, they should be approximately equal. In this way, the calculated B0 map can be processed in a series of ways, and the principle of shift direction approximation in the same time is used to find a more accurate seed point. For the two candidate angles(b1) and(b2) of each pixel point, the b that makes angle(b) / ΔTE closer to angle(B0 map) is considered to be the correct b. The closeness of the two can more accurately select the seed point of regional growth and accurately find the seed point phase error. In actual operation, in order to reduce the influence of background noise, a "amplitude weight image" is usually introduced, and a noise threshold method is introduced to ensure that the seed point falls on a non-background noise point with high amplitude, so as to prevent the spread of false phase.

[0059] The method for selecting seed points is that, first, points satisfying noise threshold conditions in the "amplitude weight image" are taken as candidate points, then one or more points in the candidate points that make angle(b) / ATE closest to angle(B0 map) in two candidate angle(b1), angle(b2) maps are taken as seed points for region growing, and the number of seed points can be adjusted by a specified number or a set threshold. With B0 map in CEST imaging as a more accurate basis, the application provides a new seed point selection method, which no longer simply relies on the global gradient of the phase error vector map, and is more conducive to seed point selection and correct phase identification. Meanwhile, the application can adjust the number of seed points, so that the selection of initial seed points is more flexible, and multiple non-connected regions are covered as much as possible under the premise of ensuring the correct error phase of the seed points. After the positions of the seed points and the error phases of each position are determined, the region growing algorithm uses the characteristic that the phase change between adjacent pixels is relatively smooth, calculates the vector angle value between b1, b2 and the seed point b for each pixel in the four-neighborhood (2D) or six-neighborhood (3D) of the seed point, takes the smaller one as the error phase value of each neighborhood pixel, and updates the label of the seed point that has been visited. The judgment process of the seed point neighborhood phase is repeated until the b of all pixel points in the image data is determined, the region growing is stopped, and the final error phase b map is obtained. The error phase b map is substituted back into formula (1), and the amplitudes W and F of the water and fat signals are solved by using the least square method, so that water-fat separation is realized.

[0060] Based on the above theoretical introduction, in a preferred embodiment of the application, a specific process of a flexible echo time water-fat separation method based on B0 map is further given, as shown in Figure 1 The specific steps of the method are as follows:

[0061] S1, acquiring a first signal image composed of B0 field offset mapping values collected for a magnetic resonance imaging target, and a second signal image and a third signal image corresponding to two echo times TE1 and TE2 respectively.

[0062] It should be noted that the three types of signal images collected in the above S1 step, i.e., the first signal image, the second signal image and the third signal image, need to maintain the same image size, have the same image pixel domain, i.e., the pixels in the three types of images need to be one-to-one corresponding.

[0063] S2. Based on the second and third signal images, calculate the first possible solution b1 and the second possible solution b2 of the error phase b of each pixel in the image pixel domain. The solution method is the same as the formulas (1) to (5) mentioned above, and will not be repeated here. Then, perform phase angle calculation on the B0 field offset mapping value, the first possible solution b1 and the second possible solution b2 corresponding to each pixel in the image pixel domain, and then align the calculation results of the three through the echo time difference ΔTE of S1 and S2 to obtain the first phase angle A1, the second phase angle A2 and the third phase angle A3 respectively.

[0064] It should be noted that since each pixel receives a B0 field offset mapping value representing the frequency in Hz when generating the B0 map, when comparing the first possible solution b1 and the second possible solution b2 with the B0 field offset mapping value in the B0 map, it is necessary to re-align the phase angle calculation results of the three using ΔTE.

[0065] The alignment is achieved by multiplying the B0 field offset mapping value by ΔTE in the B0 map, thereby aligning it to b1 and b2. This means that the B0 field offset mapping value of a single pixel in the B0 map is the B0 map value. i,j Multiplying the phase angle calculation result by the echo time difference ΔTE, we obtain the first phase angle A1 as angle(B0 map). i,j )*ΔTE; The phase angle calculation results of the first possible solution b2 and the second possible solution b2, angle(b1) and angle(b2), are respectively used as the second phase angle A2 and the third phase angle A3.

[0066] S3. For each pixel in the image pixel domain, calculate the difference between the second phase angle A2 and the first phase angle A1 as the first deviation angle ΔA1, and calculate the difference between the third phase angle A3 and the first phase angle A1 as the second deviation angle ΔA2. For each pixel, the smaller of the absolute values ​​of the first deviation angle ΔA1 and the second deviation angle ΔA2 is taken as the error phase angle value A0 = min(ΔA1, ΔA2). Select one or more pixels with the smallest error phase angle value A0 from the entire image pixel domain as seed points. The error phase of the seed point is the possible solution corresponding to the smaller of the absolute values ​​of the first deviation angle ΔA1 and the second deviation angle ΔA2. That is, if the seed point A0 = ΔA1, then the error phase of the seed point is the first possible solution b1; if the seed point A0 = ΔA2, then the error phase of the seed point is the second possible solution b2.

[0067] It's important to note that the number of seed points selected from the entire image pixel domain is a preset, adjustable value. The minimum is one, but multiple points can be selected as needed. Different initial seed point numbers will lead to different algorithm efficiency. In practical applications, the initial number of seed points can be changed by setting a preset value, or a maximum threshold for the error phase angle value A0 can be set, using all pixels with A0 not exceeding this maximum threshold as seed points, thus altering the initial seed point number.

[0068] S4. For the remaining pixels in the image pixel domain except for the seed point, the error phase of the neighboring pixels of each seed point is calculated iteratively using the region growing algorithm until the error phase of all pixels in the image pixel domain is obtained. Then, the amplitude of the water and fat signals is solved to achieve water-fat separation.

[0069] In the embodiments of the present invention, the specific implementation of the above-mentioned region growth algorithm is as follows:

[0070] S41. For each neighboring pixel of each seed point in the current image pixel domain whose error phase has not yet been determined, select the possible solution with the smaller angle between the error phase of the seed point and the first possible solution b1 and the second possible solution b2 of the neighboring pixel as the error phase of the neighboring pixel.

[0071] S42. Take all the newly determined error phase pixels as new seed points and continue to iterate through S41 until all pixels in the image pixel domain have determined the error phase.

[0072] It should be noted that in S41 above, it is assumed that the error phase of seed point i is b. i The first possible solution for the neighboring pixel j is b. j 1. The second possible solution is b. j 2. Then the angle ΔA between the error phase of the two possible solutions and the seed point is... ij This can be achieved by first performing a phase angle calculation and then subtracting the results, i.e., the first possible solution b for the neighboring pixels. j The angle ΔA between 1 and the error phase of the seed point ij1 =|angle(b j 1)-angle(b i The second possible solution b for the neighboring pixel is... j 2. The angle ΔA between the error phase of the seed point and the error phase of the seed point ij2 =|angle(b j 2)-angle(b i If ΔA ij1 ≤ΔA ij2 Then the error phase of neighboring pixel j is taken as b. j 1. If ΔAij2 <ΔA ij1 Then the error phase of neighboring pixel j is taken as b. j 2; Additionally, the angle ΔA between the error phase of the two possible solutions and the seed point. ij Alternatively, it can be achieved directly by dividing the two phases and then performing a phase angle operation, which yields the first possible solution b for the neighboring pixel. j The angle ΔA between 1 and the error phase of the seed point ij1 =|angle(b j 1 / i The second possible solution b for the neighboring pixel is... j 2. The angle ΔA between the error phase of the seed point and the error phase of the seed point ij2 =|angle(b j 2 / i If ΔA ij1 ≤ΔA ij2 Then the error phase of neighboring pixel j is taken as b. j 1. If ΔA ij2 <ΔA ij1 Then the error phase of neighboring pixel j is taken as b. j 2.

[0073] The above step S41 needs to be executed iteratively. That is, in each iteration, it is necessary to traverse every neighboring pixel of each seed point in the current image pixel domain that has not yet had its error phase determined. An error phase is generated for these traversed neighboring pixels, and they are then used as seed points for the next iteration. It is important to note that for the 2D image pixel domain obtained from 2D magnetic resonance imaging, each seed point needs to traverse its four neighboring pixels when executing the region growing algorithm. For the 3D image pixel domain obtained from 3D magnetic resonance imaging, each seed point needs to traverse its six neighboring pixels when executing the region growing algorithm.

[0074] In step S4 of this invention, the specific method for solving the amplitudes of the water and fat signals is to substitute the error phase of each pixel in the image pixel domain into the relational model shown in equation (1), and then use the least squares method to solve for the amplitude W of the water signal and the amplitude F of the fat signal. The specific solution algorithm belongs to the prior art and will not be described in detail here.

[0075] Similarly, based on the same inventive concept, another preferred embodiment of the present invention also provides a data processing system corresponding to the flexible echo time-based water-lipid separation method based on B0 map provided in the above embodiment, the system including a memory and a processor;

[0076] The memory is used to store computer programs;

[0077] The processor is configured to implement the B0 map-based flexible echo time water-fat separation method according to any one of the above embodiments when executing the computer program.

[0078] It can be understood that the storage medium can include a random access memory (RAM) and can also include a non-volatile memory (NVM), such as at least one disk memory. Meanwhile, the storage medium can also be a U disk, a mobile hard disk, a magnetic disk or an optical disk, and various media that can store program codes. Of course, with the wide application of cloud servers, the above software program can also be loaded on a cloud platform to provide corresponding services, and therefore the computer-readable storage medium is not limited to the form of local hardware.

[0079] It can be understood that the processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; and can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0080] In addition, it should be noted that the skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device can refer to the corresponding process in the foregoing method embodiments, which will not be described here. In the embodiments provided in the present application, the division of steps or modules in the device and method is only a logical functional division, and actual implementation can have another division mode, for example, a plurality of modules or steps can be combined or integrated together, or a module or step can be split.

[0081] Similarly, based on the same inventive concept, another preferred embodiment of the present application also provides a magnetic resonance imaging device corresponding to the B0 map-based flexible echo time water-fat separation method provided in the above embodiments, which comprises a magnetic resonance scanner and a control unit.

[0082] The magnetic resonance scanner is configured to acquire the first signal image, the second signal image and the third signal image for a magnetic resonance imaging target;

[0083] The control unit stores a computer program which, when executed, implements the B0 map-based flexible echo time water-fat separation method described above.

[0084] It should be noted that the magnetic resonance imaging device can be any magnetic resonance scanner capable of implementing the parallel imaging method, the structure of which belongs to the prior art and can adopt mature commercial products, and the specific model is not limited. In addition, in addition to storing the above computer program, the control unit of the magnetic resonance imaging device should also have the necessary imaging sequences and other software programs to implement CEST imaging.

[0085] Of course, the above control unit can be a separate control unit, or it can be the control unit of the magnetic resonance scanner, that is, the above B0 map-based flexible echo time water-fat separation method can be integrated in the control unit of the magnetic resonance imaging device in the form of a data processing program, so that the reconstructed results can be directly output by the magnetic resonance scanner without the need for additional new control units.

[0086] In summary, in the above embodiments of the present application, the seed points of b map region growth are determined in combination with the B0 map used in CEST imaging, which, compared with the traditional method, not only depends on the similarity measurement criterion of the smoothing between pixels, but also realizes accurate water-fat separation. In order to prove the technical effect of the present application, the above S1-S4 B0 map-based flexible echo time water-fat separation method will be applied to a specific example to specifically show its water-fat separation results.

[0087] Embodiment

[0088] In this embodiment, the MRI experimental scheme is as follows: a 3-tesla Siemens Prisma scanner (Erlangen, Germany) with a 64-channel head coil was used to scan a water phantom and two pieces of fat and lean separated pork. The water phantom was composed of two solutions, the upper layer was sunflower oil, and the lower layer was 50 mmol / L phosphate buffer solution (PBS solution). The water phantom and pork were both subjected to fl3D VIBE (Volumetric interpolated breath-hold examination) sequence, which is a kind of radio frequency phase-encoding gradient echo sequence, slice thickness = 5.0 mm, field of view (FOV) = 212 x 212 mm 3, repetition time (TR) = 7.40 ms, echo time adjustable, flip angle = 10°, acquisition matrix 256 x 256. B0 map is a 3D GRE sequence with the same field of view, direction and resolution as the fl3DVIBE sequence, TR = 50 ms, run in double echo mode, TE is 4.92 ms and 9.84 ms respectively, the echo time is consistent with the water and fat spin rotation phase.

[0089] The water-fat separation algorithm is as follows, taking the water phantom water-fat separation as an example. The echo times are TE1 = 2.2 ms and TE2 = 3.2 ms respectively, and after obtaining the scanning acquisition image, two candidate error vector angle maps angle(b1), angle(b2) are obtained by processing according to the above method Figure 2 Figure 2 A、 Figure 2 B) are the correct area growth keys, and the correct water-fat separation result Figure 2 C) is obtained by using the growth result Figure 2 D、 Figure 2 E、 Figure 2 F respectively represent the seed points of phase error b1, the seed points of phase error b2, and all seed points. By using the condition that the phase change is relatively smooth, the error phase of the neighborhood pixel points of the seed points is judged one by one from the seed points, until the error phase positions of all pixel points are determined Figure 2 G、 Figure 2 H are the growth results of the points whose error phases are b1 and b2 respectively, thus the error phase b of all pixel points is determined. Substitute b into equation (1), and the amplitude solutions W and F of water and fat are obtained by using the least square method, to realize water-fat separation Figure 2 I、 Figure 2 J).

[0090] By using the same principle, two pieces of pork (lean meat on the left and fat on the right, not connected in space) are scanned in this embodiment, and the echo times are TE1 = 2.5 ms and TE2 = 4.0 ms respectively. The difference from the Figure 2 water phantom is that the pixel points of the acquired image signals are not all connected, to ensure that there are seed points in each connected region Figure 3 A、 Figure 3 B) are the correct area growth keys, and the correct water-fat separation result Figure 3 C、 Figure 3 D) is obtained by using the growth result Figure 3 E、 Figure 3 E). ​

[0091] To further demonstrate the generality of the present application, this embodiment also conducts experiments on three-point water-fat separation technique, and the results show that the B0 map can also provide information for finding the region growing seed point. The signal model of three points is usually established as:

[0092]

[0093] wherein,

[0094] a0=exp(i2πΔfTE1), a1=exp(i2πΔf(TE2-TE1)), a2=exp(i2πΔf(TE3-TE2)), b1=exp(i2πγΔB0(TE2-TE1), b2=exp(i2πγΔB0(TE3-TE2).

[0095] The relationship between the B0 map and ΔB0 in equation (7) is established, the point closest to each other in the phase direction angle is found as the seed point of region growing, and finally the ideal water-fat separation result can also be obtained, see Figure 4 , TE1=2.2 ms, TE2=2.8 ms, TE3=3.5 ms.

[0096] The above-described embodiments are only a preferred scheme of the present application, and are not intended to limit the present application. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present application. Therefore, all technical solutions obtained by equivalent replacement or equivalent transformation fall within the protection scope of the present application.

Claims

1. A flexible echo time water-fat separation method based on a main magnetic field map, characterized by, The method comprises the following steps: S1, acquiring a main magnetic field map as a first signal image and two echo time TE1 and TE2 corresponding second signal image and third signal image for a magnetic resonance imaging target, and the three types of signal images have the same image pixel domain; S2, calculating a first possible solution of the error phase b of each pixel point in the image pixel domain based on the second signal image and the third signal image and a second possible solution ; performing a phase angle operation on the B0 field offset mapping value corresponding to each pixel in the image pixel domain obtained from the main magnetic field map, the first possible solution and the second possible solution respectively, and then performing a phase angle operation on the echo time difference between the second signal image and the third signal image aligning the operation results of the three, so as to obtain a first phase angle, a second phase angle and a third phase angle respectively; S3, for each pixel in the image pixel domain, the difference between the second phase angle and the first phase angle is calculated as the first deviation angle, and the difference between the third phase angle and the first phase angle is calculated as the second deviation angle, and each pixel takes the absolute value of the smaller one of the first deviation angle and the second deviation angle as the error phase angle value of the pixel; select one or more pixel points with the smallest error phase angle value from the entire image pixel domain as seed points, and the error phase of the seed point is the smaller one of the first deviation angle and the second deviation angle corresponding to the possible solution; S4, for the rest of the pixels in the image pixel domain except the seed points, the error phase of the neighborhood pixel points of each seed point is iteratively calculated by the region growing algorithm, and after the error phase of all pixels in the image pixel domain is obtained, the amplitude of water and fat signals is solved to realize water-fat separation; The specific implementation of the region growing algorithm is as follows: S41. For each seed point in the current image pixel domain and each neighboring pixel whose error phase has not yet been determined, from the first possible solution of that neighboring pixel... Second possible solution Select the possible solution with the smaller angle between the error phase and the seed point as the error phase of the neighboring pixel; S42, all the pixels with newly determined error phase are taken as new seed points, and S41 is continuously iterated until the error phase of all pixels in the image pixel domain is determined.

2. The flexible echo time water-fat separation method based on a main magnetic field map of claim 1, wherein, The first signal image is a B0 field inhomogeneity correction map, which is acquired by a double echo gradient sequence in CEST imaging, WASSR (Water Saturation Shift Referencing) technology or WASABI (Simultaneous mapping of water shift and B1) technology.

3. The flexible echo time water-fat separation method based on a main magnetic field map of claim 1, wherein, In the S2, the phase angle operation results of the B0 field offset mapping value corresponding to each pixel, the first possible solution and the second possible solution are aligned as follows: multiplying the result of the phase angle taking operation of the B0 field offset map value by an echo time difference of the second signal image and the third signal image , to obtain a first phase angle; dividing the result of the phase angle taking operation of the first possible solution and the result of the phase angle taking operation of the second possible solution by the echo time difference of the second signal image and the third signal image, respectively , to obtain a second phase angle and a third phase angle, respectively.

4. The flexible echo time water-fat separation method based on a main magnetic field map of claim 1, wherein, In S3, the number of seed points selected from the entire image pixel domain is a preset adjustable value, and the minimum is one.

5. The flexible echo time water-fat separation method based on a main magnetic field map of claim 1, wherein, In S41, for a two-dimensional image pixel domain obtained by two-dimensional magnetic resonance imaging, each seed point needs to traverse its four neighborhood pixels when the region growing algorithm is executed, and for a three-dimensional image pixel domain obtained by three-dimensional magnetic resonance imaging, each seed point needs to traverse its six neighborhood pixels when the region growing algorithm is executed.

6. The flexible echo time water-fat separation method based on a main magnetic field map of claim 1, wherein, In S4, the specific method for solving the amplitude of water and fat signals is: Substitute the error phase of each pixel in the image pixel domain into the relationship model of water and fat components and composite signals, and solve the amplitude W of water signal and the amplitude F of fat signal by using the least square method; The relationship model of water and fat components and composite signals is: wherein: S1 and S2 represent the pixel value of the second signal image and the pixel value of the third signal image, respectively; the phase vector of the water signal acquired for an echo time TE1, denotes the error phase of each pixel; and represent the coefficients due to the chemical shift of water and fat under a fat multi-peak model.

7. A data processing system, characterized by It comprises a memory and a processor; The memory is used to store a computer program; The processor is used to realize the flexible echo time water-fat separation method based on the main magnetic field map as claimed in any one of claims 1-6 when the computer program is executed.

8. A magnetic resonance imaging apparatus, characterized by It comprises a magnetic resonance scanner and a control unit; The magnetic resonance scanner is used to acquire the first signal image, the second signal image and the third signal image for a magnetic resonance imaging target; The control unit stores a computer program which, when executed, implements the flexible echo time water-fat separation method based on a main magnetic field map according to any one of claims 1-6.

Images