A method and system for detecting fatty acid chemical components by nuclear magnetic resonance imaging

Abstract

The present invention relates to the technical field of computer-aided diagnosis in medical imaging, and provides a method for detecting fatty acid chemical components by magnetic resonance imaging, including: S1: setting the sequence parameters of magnetic resonance imaging (MRI), and performing magnetic resonance imaging (MRI) scanning based on the sequence parameters; S2: segmenting the original magnitude image to form a bi-clustering segmentation mask, applying the bi-clustering segmentation mask to remove the background signal in the original phase image, correcting the zero-order phase and the first-order phase of the multi-echo original phase image, and combining the original magnitude image and the corrected original phase image to generate a final tissue image; S3: modeling the real part of the signal, separating the fat and water proton densities, calculating the proton density fat fraction, repeatedly fitting the parameters in the model, and finally calculating the values of the fatty acid chemical components of saturated fatty acids, monounsaturated fatty acids and polyunsaturated fatty acids. It realizes in-vivo, real-time and non-invasive detection of the fatty acid chemical components of human adipose tissue.

Description

Technical Field

[0001] The present invention relates to the technical field of computer-aided diagnosis in medical imaging, and particularly to a method and system for detecting fatty acid chemical components by nuclear magnetic resonance imaging. Background Art

[0002] As a newly discovered endocrine organ that has been seriously underestimated, adipose tissue plays an important role in the endocrine system, immune system, and the occurrence and development of tumors. The main component of adipose tissue is triglyceride molecules, which can be classified into saturated fatty acids (SFA), monounsaturated fatty acids (MUFA), and polyunsaturated fatty acids (PUFA) according to the number of unsaturated double bonds. Fatty acids in the human body are important energy sources and are also involved in activities such as cell signal transduction, regulation of immune responses, and maintenance of internal environmental homeostasis. Therefore, the fatty acid chemical composition (FAC) of triglycerides in human adipose tissue and its changes have an impact on the occurrence and development of various diseases. The fatty acid chemical composition of triglycerides in adipose tissue and its changes are closely related to the occurrence and development of various diseases, such as:

[0003] Atherosclerosis: Unsaturated fatty acids, especially PUFA, help reduce the level of low-density lipoprotein cholesterol (LDL-C), thereby reducing the risk of atherosclerosis.

[0004] Acute coronary syndrome: Changes in fatty acid composition may affect the risk of cardiovascular diseases.

[0005] Heart failure: Disorders of fatty acid metabolism may lead to impaired cardiac function.

[0006] Diabetes: Fatty acid metabolism plays an important role in insulin resistance and the occurrence of diabetes.

[0007] Tumor: Changes in the metabolic and secretory functions of adipose tissue can affect the tumor microenvironment and promote or inhibit tumor growth.

[0008] Currently, the detection methods for fatty acids at home and abroad are mainly for in vitro detection of foods (vegetable oils, animal fats, milk powders, etc.) and biological samples. The most widely used methods are traditional titration and colorimetry. Colorimetry has become the main method for fatty acid detection, with high sensitivity, wide application range, simple operation, and easy large-scale popularization of operation technology far exceeding that of traditional titration. Clinically, the correlation between the occurrence and development of diseases such as atherosclerosis, acute coronary syndrome, and heart failure is mainly indirectly analyzed by measuring free fatty acids in blood using liquid chromatography-mass spectrometry (LC-MS) and biochemical methods.

[0009] At present, there is still a lack of in vivo, real-time, accurate, and repeatable non-invasive methods for detecting the chemical components of fatty acids in adipose tissue in clinical practice. Even though there is currently a method for detecting unsaturated fatty acids using proton nuclear magnetic resonance spectroscopy in existing technologies, this method is affected by the field strength, and the measurement accuracy on currently available clinical nuclear magnetic resonance machines still needs to be improved. Summary of the Invention

[0010] In view of the above problems, the purpose of the present invention is to provide a method and system for detecting the chemical components of fatty acids by nuclear magnetic resonance imaging. Aiming at the problem that there is currently a lack of in vivo, real-time, accurate, and repeatable non-invasive methods for detecting the chemical components of fatty acids in adipose tissue in clinical practice, the present invention intends to establish a method for detecting the chemical components of fatty acids by nuclear magnetic resonance imaging based on multi-echo gradient echo and chemical shift encoding, to achieve in vivo, real-time, and non-invasive detection of the chemical components of fatty acids in adipose tissue in various parts of the human body, and to evaluate the role of adipose tissue in the occurrence and development of the endocrine system, immune system, and tumors.

[0011] The above object of the present invention is achieved through the following technical solutions:

[0012] A method for detecting the chemical components of fatty acids by nuclear magnetic resonance imaging, comprising the following steps:

[0013] S1: Set the sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding, and perform the MRI scan based on the sequence parameters to obtain multiple groups of original magnitude images and original phase images at different echo times (TE).

[0014] S2: Segment the original magnitude images to form a double-cluster segmentation mask, apply the double-cluster segmentation mask to remove the background signal in the original phase images, and only retain the tissue signals of interest. Correct the zero-order phase and first-order phase of the multi-echo original phase images, and combine the original magnitude images and the corrected original phase images to generate the final tissue images.

[0015] S3: Model the real part of the signals of the final tissue images, separate the proton densities of fat and water, calculate the proton density fat fraction, and repeatedly fit the parameters in the model to finally calculate the numerical values of the chemical components of fatty acids including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids.

[0016] Further, in step S1, the sequence parameters include: repetition time (TR), echo time (TE), field of view (FOV), acquisition matrix, slice thickness, slice gap, acquisition time, and readout bandwidth;

[0017] Among them, the repetition time TR is the interval between two radiofrequency pulses during the magnetic resonance imaging (MRI) scan. The echo time TE is the time interval from the emission of the radiofrequency pulse to signal acquisition. Generally, there are multiple echo times TE. The field of view (FOV) is the size of the physical area covered by the obtained original magnitude image. The acquisition matrix is the number of pixels in the original magnitude image, usually expressed as the number of rows and columns. The slice thickness defines the thickness of each slice, and the slice gap is the distance between adjacent slices. The acquisition time is the total time required to complete one MRI scan. The readout bandwidth is the frequency range per pixel during signal acquisition;

[0018] In step S1, the original magnitude image is the image information extracted from the amplitude part of the MRI signal, reflecting the signal intensity at each pixel position. The original phase image is the image information extracted from the phase part of the MRI signal, reflecting the phase angle of the signal at each pixel position.

[0019] Further, in step S2, the original magnitude image is segmented to form the biclustering segmentation mask, and the biclustering segmentation mask is applied to remove the background signal in the original phase image, only retaining the tissue signal of interest. Specifically:

[0020] The biclustering method is used to segment the tissue and background in the original magnitude image, suppress the background, only retain the tissue, form the biclustering segmentation mask, and apply the biclustering segmentation mask to the original phase image to remove the background signal, only retaining the tissue signal of interest;

[0021] Among them, the biclustering method uses k-means clustering, selects k = 2, segments the original magnitude image into two clusters, one representing the tissue signal and the other representing the background signal, suppresses the background cluster, only retains the tissue signal cluster, and forms the biclustering segmentation mask.

[0022] Further, in step S2, the zero-order phase and first-order phase of the multi-echo original phase image are corrected, and the original magnitude image and the corrected original phase image are combined to generate the final tissue image. Specifically:

[0023] The minimum cost network flow is used to unroll the multi-echo original phase image echo by echo. From the unrolled original phase image, the zero-order phase and the first-order phase are extracted pixel by pixel. This process uses the phase noise ratio difference related to the echo time TE through the gradient echo (GRE) sequence model and fits the phase of the MRI signal using weighted linear least squares.

[0024]

[0025] Among them, represents the zero-order phase, represents the first-order phase, T E is the echo time TE;

[0026] For the heterogeneity ΔB0 of the main magnetic field B0 field in the magnetic resonance imaging (MRI) device, by using it is deduced that, by using the zero-order phase and the first-order phase of the original phase image of the corrected multi-echo, the phase error caused by the heterogeneity of the main magnetic field B0 field is eliminated;

[0027] The original amplitude image and the corrected original phase image are combined to generate the final tissue image, accurately reflecting the internal structure and composition of the tissue.

[0028] Furthermore, in step S3, the real part of the signal of the final tissue image is modeled to separate the fat and water proton densities, calculate the proton density fat fraction, and repeatedly fit the parameters in the model. Finally, the numerical values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids are calculated, specifically as follows:

[0029] S31: Use a three-parameter biexponential model to integrate nine fat resonators for the real part S real of the signal of the final tissue image for modeling, separating the fat proton density PD f and the water proton density PD w . Under the steady-state condition of ignoring the contribution of the longitudinal relaxation time T1 and correcting the heterogeneity of the main magnetic field B0 field, the equation of the real part S real of the signal of the final tissue image with respect to the echo time TE is:

[0030]

[0031] where S real (T E ) represents the real part signal varying with the echo time TE, T2* is the transverse decay, C k is a coefficient, equal to the ratio of the signal of the kth fat resonator to the total fat signal, f k corresponds to the frequency shift between water and the kth fat resonance peak, and i is the imaginary unit, used to represent the phase information and frequency components of the signal;

[0032] According to the data of the real part of the signal of the final tissue image combined with the values of C k and f k , given PDf , PD w , the initial estimate of the initial estimate of T2*, use a non-linear optimization algorithm to adjust the model parameters to minimize the error between the fitted signal value and the actual measured value, and estimate PD f and PD w ;

[0033] Using the estimated PD f and PD w , calculate the proton density fat fraction PDFF, the formula is: PDFF = [PD f / (PD f +PD w )]×100;

[0034] S32: Modify the fat spectrum model to express it according to the number of protons and indicators including the number of unsaturated double bonds ndb in each fatty acid chain of each triglyceride molecule, the number of double bonds nmidb separating methylene groups in each fatty acid molecule, and the average carbon chain length CL of the fatty acid chains in the triglyceride molecule. Under the steady-state condition of ignoring the contribution of the longitudinal relaxation time T1 and correcting the inhomogeneity of the main magnetic field B0 field, the real part S of the signal of the final tissue image real The equation about the echo time TE is expressed as follows:

[0035]

[0036] where w and f represent the number of water and triglyceride molecules respectively, n k (ndb,CL,nmidb) represents the number of protons of the fat spectrum component k, f k represents the frequency shift between water and the k-th fat spectrum component, n water represents the number of protons in water molecules, T2* represents the transverse decay, and i is the imaginary unit, which is used to represent the phase information and frequency components of the signal;

[0037] CL and nmidb are expressed as functions of ndb, which are CL = 16.8 + 0.25×ndb and nmidb = 0.093×ndb 2 );

[0038] S33: Repeat the fitting process to fit the parameters w, f, and nmidb. The values of ndb and nmidb are converted into the numerical values of fatty acid chemical components including saturated fatty acid f SFA , monounsaturated fatty acid f MUFA and polyunsaturated fatty acid f PUFA as follows:

[0039]

[0040] where the proportion f of the tri-unsaturated fatty acid TriUFA is approximately represented by a fixed value of 2%.

[0041] In step S1, set the echo intervals and the number of echoes of different said echo times TE, and respectively use the method of nuclear magnetic resonance imaging for detecting fatty acid chemical components in steps S1 - S3 to calculate the values of fatty acid chemical components including the saturated fatty acid, the mono-unsaturated fatty acid, and the poly-unsaturated fatty acid at different said echo intervals and said number of echoes;

[0042] For the values of fatty acid chemical components calculated for different said echo intervals and said number of echoes, use the mean absolute error MAE to evaluate the influence of the echo interval and the number of echoes on the calculation accuracy of the fatty acid chemical component FAC. The formula is:

[0043]

[0044] where y i represents the reference value, and represents the calculated measured value.

[0045] Perform a linear correlation analysis and comparison on the measured values of fatty acid chemical components including the saturated fatty acid, the mono-unsaturated fatty acid, and the poly-unsaturated fatty acid calculated by using the method of nuclear magnetic resonance imaging for detecting fatty acid chemical components in steps S1 - S3 with the reference value to evaluate the accuracy of the quantitative analysis MR-FAC quantification of the magnetic resonance imaging MRI;

[0046] Use the Pearson linear correlation coefficient r and its 95% confidence interval to test the linearity, and perform a linear regression analysis including the intercept, slope, and the corresponding 95% confidence range;

[0047] In addition, the p-value for testing the null hypothesis that there is no linear relationship between the measured value and the reference value is also calculated. A significance level P < 0.05 is considered to be statistically significant.

[0048] Evaluate the consistency:

[0049] Two measurers measure respectively to obtain two sets of measured values M1 and M2 of the CSE-MRI sequence of the magnetic resonance imaging MRI based on the multi-echo gradient echo and the chemical shift encoding for ICC consistency analysis;

[0050] The evaluation criteria for the ICC consistency analysis are: ICC < 0.4 indicates poor correlation, ICC between 0.40 and 0.59 indicates medium correlation, ICC between 0.60 and 0.74 indicates good correlation, and ICC > 0.74 indicates excellent correlation;

[0051] Evaluation of repeatability:

[0052] Two sets of measurement values, scan1 and scan2, of the CSE-MRI sequence of the magnetic resonance imaging (MRI) based on the multi-echo gradient echo and the chemical shift encoding were obtained by a single measurer through two measurements before and after respectively, and a statistical Bland-Altman analysis was performed on them.

[0053] The evaluation criterion for the Bland-Altman analysis is that at least 95% of the two differences are within the limits of agreement, that is, more than 95% of the points in the graph are within the limits of agreement.

[0054] A magnetic resonance imaging system for detecting fatty acid chemical components by using the method of magnetic resonance imaging for detecting fatty acid chemical components as described above, comprising:

[0055] An original image acquisition module, configured to set sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding, perform scanning of the magnetic resonance imaging (MRI) based on the sequence parameters, and obtain multiple sets of original amplitude images and original phase images at different echo times (TE).

[0056] A tissue image generation module, configured to segment the original amplitude image to form a bi-clustering segmentation mask, apply the bi-clustering segmentation mask to remove background signals in the original phase image, only retain tissue signals of interest, correct the zero-order phase and the first-order phase of the original phase image with multiple echoes, and combine the original amplitude image and the corrected original phase image to generate a final tissue image.

[0057] A chemical component calculation module, configured to model the real part of the signal of the final tissue image, separate the fat and water proton densities, calculate the proton density fat fraction, repeatedly fit the parameters in the model, and finally calculate the numerical values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids.

[0058] A computer-readable storage medium stores computer code, and when the computer code is executed, the method as described above is executed.

[0059] Compared with the prior art, the beneficial effects of the present invention are:

[0060] By providing a method for detecting fatty acid chemical components by nuclear magnetic resonance imaging, including: S1: Set the sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding, and perform the MRI scan based on the sequence parameters to obtain multiple sets of original magnitude images and original phase images at different echo times (TE). S2: Segment the original magnitude images to form a bi-clustering segmentation mask, apply the bi-clustering segmentation mask to remove the background signals in the original phase images, only retain the tissue signals of interest, correct the zero-order phase and first-order phase of the original phase images with multiple echoes, and combine the original magnitude images and the corrected original phase images to generate the final tissue images. S3: Model the real part of the signals of the final tissue images, separate the fat and water proton densities, calculate the proton density fat fraction, repeatedly fit the parameters in the model, and finally calculate the values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids. The above technical solution realizes the functions of in-vivo, real-time, and non-invasive detection of fatty acid chemical components in adipose tissues of various parts of the human body, and can be used to evaluate the role of adipose tissues in the occurrence and development of the endocrine system, immune system, and tumors. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 FIG. is the overall flowchart of the method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to the present invention;

[0062] Figure 2 FIG. is the reference diagram of fatty acid components of the vegetable oil model according to the present invention;

[0063] Figure 3 FIG. is a schematic diagram of the linear correlation analysis between the measured values and reference values of the fatty acid components according to the present invention, where Figure 3 (a) is the linear correlation result between the measured values and reference values of monounsaturated fatty acids, Figure 3 (b) is the linear correlation result between the measured values and reference values of polyunsaturated fatty acids, Figure 3 (c) is the linear correlation result between the measured values and reference values of saturated fatty acids;

[0064] Figure 4 FIG. is the fatty acid component fraction map of human adipose tissue according to the present invention;

[0065] Figure 5 FIG. is the overall structure diagram of the system for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0067] Those skilled in the art of this technology can understand that unless specifically stated otherwise, the singular forms "a", "an", "the", and "said" used herein may also include the plural forms. It should be further understood that the term "comprising" used in the specification of the present invention means the presence of the described features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or their groups.

[0068] First Embodiment

[0069] As Figure 1 shown, this embodiment provides a method for detecting fatty acid chemical components by nuclear magnetic resonance imaging, including the following steps:

[0070] S1: Set the sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding, and perform the MRI scan based on the sequence parameters to obtain multiple sets of original magnitude images and original phase images at different echo times (TE).

[0071] Specifically, in this embodiment, it is first necessary to set the sequence parameters for scanning the multi-echo gradient echo and chemical shift encoding MRI, and then perform the scan based on the set sequence parameters. The sequence parameters include: repetition time (TR), the echo time (TE), field of view (FOV), acquisition matrix, slice thickness, slice gap, acquisition time, readout bandwidth, and number of echoes;

[0072] The repetition time (TR) is the interval between two radiofrequency pulses during the MRI scan. Its function is that TR affects the recovery time of longitudinal magnetization and thus affects the image contrast. A shorter TR will result in T1-weighted imaging, emphasizing the T1 relaxation characteristics of tissues; a longer TR is helpful for T2-weighted imaging, emphasizing the T2 relaxation characteristics of tissues.

[0073] The echo time (TE) is the time interval from the emission of the radiofrequency pulse to the signal acquisition. Generally, there are multiple echo times (TE). Its function is that TE affects the decay time of transverse magnetization and thus affects the image contrast. A shorter TE will produce T1-weighted imaging, and a longer TE is helpful for T2-weighted imaging. In a multi-echo sequence, different TEs are helpful for separating different chemical components.

[0074] The field of view (FOV) is the size of the physical area covered by the acquired original magnitude image, and its function is to determine the spatial coverage of the image. A larger FOV can cover a larger area, but the resolution may be lower; a smaller FOV can provide higher resolution.

[0075] The acquisition matrix is the number of pixels in the original magnitude image, usually expressed as the number of rows and columns. Its function is that the size of the acquisition matrix determines the resolution of the image. A larger acquisition matrix provides higher resolution, but increases the scan time and data volume.

[0076] The slice thickness defines the thickness of each slice, and its function is that the slice thickness determines the longitudinal resolution and signal-to-noise ratio of the image. A thinner slice thickness provides higher resolution, but the signal-to-noise ratio may be lower; a thicker slice thickness provides a better signal-to-noise ratio.

[0077] The slice gap is the distance between adjacent slices. Its function is that if the slice gap is 0 mm, it means no-gap acquisition and provides continuous image data. If the slice gap is greater than 0 mm, it means there is a gap between slices, which helps to reduce the scan time and reduce motion artifacts.

[0078] The acquisition time is the total time required to complete one magnetic resonance imaging (MRI) scan. Its function is to affect the patient's comfort and motion artifacts. A shorter acquisition time can reduce the artifacts caused by patient movement, but may require sacrificing the resolution or signal-to-noise ratio of the image.

[0079] The readout bandwidth is the frequency range per pixel during signal acquisition. Its function is that a higher readout bandwidth can reduce chemical shift artifacts and distortion, but will reduce the signal-to-noise ratio; a lower readout bandwidth can increase the signal-to-noise ratio, but may increase artifacts and distortion.

[0080] The echo train length is the number of echoes acquired in one repetition time (TR). The more echoes, the more signal change information can be captured, which helps in the reconstruction of multi-echo images and chemical shift analysis.

[0081] For example, after detection, a preferred embodiment is a repetition time (TR) of 181.00 ms, echo times (TE) of 1.23, 2.07, 2.91, 3.75, 4.59, 5.43, 6.27, 7.11, 7.95, 8.79, 9.63, 10.47, 11.31, 12.15, 12.99, 13.83 ms, a field of view (FOV) of 380 mm × 310 mm, an acquisition matrix of 128 × 128, a slice thickness of 10 mm, a slice gap of 0 mm, an acquisition time of 20 seconds (image acquisition completed in one breath-hold). A high readout bandwidth of 1950 Hz / px was selected, and the echo train length was 16.

[0082] After the sequence parameters are set, the magnetic resonance imaging (MRI) scan is performed based on the sequence parameters to obtain multiple sets of original magnitude images and original phase images at different echo times (TE). The scanned MRI signals include different information. The original magnitude image is the image information extracted from the amplitude part of the MRI signal, reflecting the signal intensity at each pixel position. Specifically, it reflects the signal intensity at each pixel position. Magnitude images are typically used to display anatomical structures because they provide tissue contrast and can clearly show the distribution of different tissue types. The original phase image is the image information extracted from the phase part of the MRI signal, reflecting the phase angle of the signal at each pixel position. The phase image provides the phase angle of the signal at each pixel position and is typically used for more detailed quantitative analysis, such as correcting B0 field inhomogeneity (magnetic field inhomogeneity) and distinguishing signals in different chemical environments, such as separating fat and water.

[0083] S2: Segment the original magnitude image to form a bi-clustering segmentation mask. Apply the bi-clustering segmentation mask to remove the background signal in the original phase image and only retain the tissue signal of interest. Correct the zero-order phase and first-order phase of the multi-echo original phase image, and combine the original magnitude image and the corrected original phase image to generate the final tissue image.

[0084] Specifically, in step S2, segment the original magnitude image to form the bi-clustering segmentation mask, and apply the bi-clustering segmentation mask to remove the background signal in the original phase image and only retain the tissue signal of interest, specifically:

[0085] Use the bi-clustering method to segment the tissue and background in the original magnitude image, suppress the background, only retain the tissue, form the bi-clustering segmentation mask, and apply the bi-clustering segmentation mask to the original phase image to remove the background signal and only retain the tissue signal of interest;

[0086] Among them, the bi-clustering method uses k-means clustering. Select k = 2 to segment the original magnitude image into two clusters, one representing the tissue signal and the other representing the background signal. Suppress the background cluster and only retain the tissue signal cluster to form the bi-clustering segmentation mask.

[0087] In step S2, correct the zero-order phase and first-order phase of the multi-echo original phase image, and combine the original magnitude image and the corrected original phase image to generate the final tissue image, specifically:

[0088] Unwrap the original phase image of multiple echoes one by one using the minimum-cost network flow. From the unwrapped original phase image, extract the zero-order phase and the first-order phase pixel by pixel. This process uses the phase noise ratio difference related to the echo time TE through the gradient echo GRE sequence model and fits the phase of the magnetic resonance imaging (MRI) signal using the weighted linear least squares method.

[0089]

[0090] Wherein, represents the zero-order phase, represents the first-order phase, T E is the echo time TE;

[0091] For the heterogeneity ΔB0 of the main magnetic field B0 field in the magnetic resonance imaging (MRI) device, by using it is deduced that by using the calculated zero-order phase and the first-order phase of the corrected original phase image of multiple echoes, the phase error caused by the heterogeneity of the main magnetic field B0 field is eliminated;

[0092] Combine the original magnitude image and the corrected original phase image to generate the final tissue image, accurately reflecting the internal structure and composition of the tissue.

[0093] S3: Model the real part of the signal of the final tissue image, separate the fat and water proton densities, calculate the proton density fat fraction, and repeatedly fit the parameters in the model. Finally, calculate the values of the fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids.

[0094] Specifically, calculating the values of the fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids specifically includes the following three steps:

[0095] S31: Use a three-parameter biexponential model to integrate nine fat resonators to model the real part S real of the signal of the final tissue image, separate the fat proton density PD f and the water proton density PD w Under the steady-state condition of ignoring the contribution of the longitudinal relaxation time T1 and correcting the heterogeneity of the main magnetic field B0 field, the real part S real of the signal of the final tissue image with respect to the echo time TE has the following equation:

[0096]

[0097] Wherein, S real (T Erepresents the real - part signal varying with the echo time TE, T2* is the transverse relaxation, C k is a coefficient, equal to the ratio of the signal of the k - th fat resonator to the total fat signal, f k corresponds to the frequency shift between water and the k - th fat resonance peak, i is the imaginary unit, used to represent the phase information and frequency components of the signal;

[0098] Given C k and f k values, initial estimates of PD f , PD w , and initial estimates of the initial estimates of T2* are given, and a non - linear optimization algorithm is used to adjust the model parameters to minimize the error between the fitted signal value and the actual measured value, and PD f and PD w are estimated;

[0099] Using the estimated PD f and PD w , the proton density fat fraction PDFF is calculated, and the formula is: PDFF = [PD f / (PD f +PD w )]×100;

[0100] As shown in Table 1, the component model parameter information of nine fat resonators

[0101] Table 1

[0102]

[0103]

[0104] S32: Modify the fat spectrum model to be expressed according to the number of protons and indicators including the number of unsaturated double bonds ndb in the fatty acid chain of each triglyceride molecule, the number of double bonds nmidb separating methylene groups in each fatty acid molecule, and the average carbon chain length CL of the fatty acid chain in the triglyceride molecule. Under the steady - state condition of ignoring the contribution of the longitudinal relaxation time T1 and correcting the inhomogeneity of the main magnetic field B0, the real - part S real of the signal of the final tissue image with respect to the echo time TE is expressed by the following equation:

[0105]

[0106] where w and f represent the number of water and triglyceride molecules respectively, n k (ndb, CL, nmidb) represents the number of protons of the fat spectrum component k, f k represents the frequency shift between water and the k - th fat spectrum component, nwater represents the number of protons in a water molecule, T2* represents transverse relaxation, and i is the imaginary unit, which is used to represent the phase information and frequency components of the signal;

[0107] In this step, CL and nmidb are expressed as functions of ndb, specifically CL = 16.8 + 0.25×ndb and nmidb = 0.093×ndb 2 ), and the T2* value from the previous step is used. Therefore, the fitting parameters are w, f, and ndb.

[0108] S33: Repeat the fitting process for the fitting parameters w, f, and nmidb. The values of ndb and nmidb are converted to saturated fatty acid f SFA , monounsaturated fatty acid f MUFA and polyunsaturated fatty acid f PUFA for the numerical values of fatty acid chemical components including:

[0109]

[0110] where the proportion f of triunsaturated fatty acid TriUFA is approximately represented by a fixed value of 2%.

[0111] This embodiment also includes the technical optimization of multi-echo gradient echo and chemical shift encoding MRI for detecting fatty acid components, specifically:

[0112] In step S1, set the echo intervals and the number of echoes for different echo times TE, and calculate the numerical values of fatty acid chemical components including saturated fatty acid, monounsaturated fatty acid, and polyunsaturated fatty acid at different echo intervals and the number of echoes respectively by using the method of nuclear magnetic resonance imaging for detecting fatty acid chemical components in steps S1 - S3;

[0113] For the numerical values of fatty acid chemical components calculated for different echo intervals and the number of echoes, use the mean absolute error MAE to evaluate the influence of the echo intervals and the number of echoes on the calculation accuracy of fatty acid chemical component FAC. The formula is:

[0114]

[0115] where, y i represents the reference value, represents the calculated measured value.

[0116] For example, this embodiment adopts as Figure 2Eight vegetable oil models shown were verified. The echo spacing ΔTE = 0.84 ms and 1.23 ms were selected, and the number of echoes was 8 and 16. Half an hour before the measurement, the vegetable oil model was placed in the scanning room to ensure that the oil temperature was constant during all experiments. It was placed in the scanner with the container perpendicular to the main magnetic field and perpendicular to the measured coronal section. All the collected data sets were processed using the proposed method. The average fatty acid values of 8 oils and oil mixtures were calculated within the region of interest in the fatty acid maps of all the conducted experiments. By calculating the percentage of the mean absolute error (MAE) of saturated fatty acids SFA, monounsaturated fatty acids MUFA, and polyunsaturated fatty acids PUFA relative to the manufacturer's reference values for each measurement protocol, the influence of the echo spacing time and the number of echoes on the accuracy of the fatty acid chemical component FAC parameter map was evaluated in %.

[0117] The echo spacing ΔTE = 0.84 ms and 1.23 ms were selected, and the number of echoes was 8 and 16. The results of the mean absolute error MAE of the fatty acid chemical components of the vegetable oil model are as follows:

[0118] Influence of echo spacing time and number of echoes on the accuracy of the FAC parameter map (MAE, %)

[0119]

[0120] It can be seen that the scheme with ΔTE = 0.84 ms, 16 echo times, and a total readout duration of the echo sequence of 13.83 ms achieved the minimum MAES value for the three fatty acid components. Shorter (7.11 ms) and longer (19.68 ms) echo sequence durations led to reduced accuracy, especially for the saturated fatty acid component.

[0121] This embodiment also includes an in vitro accuracy analysis of multi - echo gradient echo and chemical shift - encoded MRI for detecting fatty acid components, specifically:

[0122] The measured values of the fatty acid chemical components including the saturated fatty acids, the monounsaturated fatty acids, and the polyunsaturated fatty acids calculated by the method for detecting fatty acid chemical components using nuclear magnetic resonance imaging in steps S1 - S3 were linearly correlated and analyzed and compared with the reference values to evaluate the accuracy of the quantitative analysis MR - FAC quantification of the magnetic resonance imaging MRI;

[0123] The Pearson linear correlation coefficient r and its 95% confidence interval were used to test the linearity, and a linear regression analysis including the intercept, slope, and the corresponding 95% confidence range was performed;

[0124] In addition, the p-value for the null hypothesis that there is no linear relationship between the test measurement values and the reference values was calculated, and a significance level of P < 0.05 was considered statistically significant.

[0125] For example, as Figure 3 shown, the measurement results of the optimized scanning parameter optimization scheme (ΔTE = 0.84 ms, number of echoes 16) selected by optimization were compared with the reference values for linear correlation analysis to evaluate the accuracy of MR-FAC quantification. The Pearson linear correlation coefficient r and its 95% confidence interval were used to test linearity, and linear regression analysis (intercept, slope, and corresponding 95% confidence ranges) was performed. In addition, the p-value for the null hypothesis that there is no linear relationship between the test measurement values and the reference values was calculated, and a significance level of P < 0.05 was considered statistically significant. All statistical analyses were performed using MedCalc software version 20.100 (MedCalc Software Ltd., Ostend, Belgium).

[0126] In as Figure 3 shown, the results of the linear correlation analysis of the chemical shift-encoded magnetic resonance imaging quantitative analysis of FAC and the reference values showed that there was a strong correlation (Pearson linear correlation coefficient r = 0.79) and a significant correlation (p = 0.018) between the measured values of the saturated fat component (fSFA) and the reference fatty acid values. The regression analysis showed a slope of 0.583 (95% CI: 0.1382, 1.0288). There were very strong (Pearson linear correlation coefficients r were 0.98 and 1.00 respectively) and significant (P < 0.001) correlations between the measured values of the monounsaturated fat component (fMUFA) and the polyunsaturated fat component (fPUFA) and the reference values. The regression analysis showed slopes of 1.052 (95% CI: 0.8576, 1.2458) and 1.114 (95% CI: 1.0339, 1.1937) respectively.

[0127] This embodiment also included the consistency and repeatability of multi-echo gradient echo and chemical shift-encoded MRI in detecting fatty acid components in human adipose tissue, specifically:

[0128] Evaluating consistency:

[0129] Two measurers measured separately to obtain two sets of measurement values M1 and M2 of the CSE-MRI sequence of the magnetic resonance imaging MRI based on the multi-echo gradient echo and the chemical shift encoding for ICC consistency analysis;

[0130] The evaluation criteria for ICC consistency analysis are as follows: ICC < 0.4 indicates poor correlation, ICC between 0.40 and 0.59 indicates moderate correlation, ICC between 0.60 and 0.74 indicates good correlation, and ICC > 0.74 indicates excellent correlation;

[0131] Evaluate repeatability:

[0132] Two sets of measurement values scan1 and scan2 of the CSE-MRI sequence of the magnetic resonance imaging MRI based on the multi-echo gradient echo and the chemical shift encoding are obtained by a measurer through two measurements before and after respectively, and a statistical Bland-Altman analysis is performed;

[0133] The evaluation criteria for the Bland-Altman analysis are that at least 95% of the two differences are within the limits of agreement, that is, more than 95% of the points in the graph are within the limits of agreement.

[0134] As Figure 4 shown, when imaging human abdominal adipose tissue using the above method, in the detection of fatty acid components in human adipose tissue, the ICC values measured twice by the same observer are 0.855 - 0.939. The intra-observer measurement values of fatty acids in subcutaneous / visceral adipose tissue have good repeatability. According to the evaluation criteria, the repeatability grades are all excellent. The inter-observer ICC values are 0.808 - 0.944. The inter-observer measurement values of fatty acids in subcutaneous / visceral adipose tissue have good repeatability. According to the evaluation criteria, the repeatability grades are all excellent. The repeatability ICC values for short-term repeated scan measurements are 0.75 - 0.985. The short-term repeated detection measurement values of fatty acids in subcutaneous / visceral adipose tissue have good repeatability. According to the evaluation criteria, the repeatability grades are all excellent. The Bland-Altman analysis shows that the ratios of the fatty acid fraction values of the two scans before and after within the 95% limits of agreement are respectively subcutaneous adipose tissue - fSFA: 100.0% (20 / 20), subcutaneous adipose tissue - fMUFA: 100% (20 / 20), subcutaneous adipose tissue - fPUFA: 95% (19 / 20), visceral adipose tissue - fSFA: 95% (19 / 20), visceral adipose tissue - fMUFA: 100% (20 / 20), and visceral adipose tissue - fPUFA: 95% (19 / 20).

[0135] Second Embodiment

[0136] As Figure 5 shown, this embodiment provides a system for detecting fatty acid chemical components by nuclear magnetic resonance imaging using the method for detecting fatty acid chemical components by nuclear magnetic resonance imaging as in the first embodiment, including:

[0137] The original image acquisition module 1 is used to set the sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding, perform the scan of the magnetic resonance imaging (MRI) based on the sequence parameters, and obtain multiple groups of original magnitude images and original phase images at different echo times (TE);

[0138] The tissue image generation module 2 is used to segment the original magnitude images to form a bi-clustering segmentation mask, apply the bi-clustering segmentation mask to remove the background signal in the original phase images, only retain the tissue signals of interest, correct the zero-order phase and the first-order phase of the multi-echo original phase images, and combine the original magnitude images and the corrected original phase images to generate the final tissue images;

[0139] The chemical component calculation module 3 is used to model the real part of the signals of the final tissue images, separate the fat and water proton densities, calculate the proton density fat fraction, repeatedly fit the parameters in the model, and finally calculate the numerical values of the fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids.

[0140] A computer-readable storage medium stores computer code, and when the computer code is executed, the above method is executed. Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program, and this program can be stored in a computer-readable storage medium. The storage medium can include: read-only memory (ROM, Read Only Memory), random access memory (RAM, Random Access Memory), magnetic disks, or optical discs, etc.

[0141] The above is only the preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions falling within the idea of the present invention belong to the protection scope of the present invention. It should be pointed out that for those of ordinary skill in the art in this technical field, several improvements and refinements made without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

[0142] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0143] It should be noted that the above embodiments can be freely combined as needed. The above are only the preferred embodiments of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A method for detecting fatty acid chemical components by nuclear magnetic resonance imaging, characterized in that Including the following steps: S1: Set the sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding. Perform the MRI scan based on the sequence parameters to obtain multiple sets of original magnitude images and original phase images at different echo times (TE). S2: Segment the original magnitude image to form a bi-clustering segmentation mask. Apply the bi-clustering segmentation mask to remove the background signal in the original phase image, and only retain the tissue signal of interest. Correct the zero-order phase and first-order phase of the multi-echo original phase image. Combine the original magnitude image and the corrected original phase image to generate the final tissue image. S3: Model the real part of the signal of the final tissue image, separate the fat and water proton densities, calculate the proton density fat fraction, repeatedly fit the parameters in the model, and finally calculate the numerical values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids. In step S3, model the real part of the signal of the final tissue image, separate the fat and water proton densities, calculate the proton density fat fraction, repeatedly fit the parameters in the model, and finally calculate the numerical values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids, specifically as follows: S31: Integrate the real part S of the signals of nine fat resonators in the final tissue image using a three-parameter biexponential model real for modeling, separate the fat proton density PD f and the water proton density PD w Under the steady-state condition of ignoring the contribution of the longitudinal relaxation time T1 and correcting the inhomogeneity of the main magnetic field B0 field, the real part S of the signal of the final tissue image real The equation regarding the echo time TE is as follows: Among them, represents the real part signal varying with the echo time TE, where TE is the echo time, T2* is the transverse relaxation, and C k is a coefficient equal to the ratio of the signal of the k-th fat resonator to the total fat signal, and f k corresponds to the frequency shift between water and the k-th fat resonance peak, and i is the imaginary unit used to represent the phase information and frequency components of the signal; Combining data of the real part of the signal of the final tissue image with C k and f k values, given initial estimates of PD f and PD w and T2*, the model parameters are adjusted using a non-linear optimization algorithm to minimize the error between the fitted signal value and the actual measured value, and PD f and PD w are estimated; Using the estimated PD f and PD w , calculate the proton density fat fraction PDFF, with the formula: PDFF = [PD f / (PD f + PD w )] × 100; S32: Modify the fat spectrum model to be expressed according to the number of protons and indicators including the number of unsaturated double bonds ndb in the fatty acid chain of each triglyceride molecule, the number of double bonds nmidb separating methylene groups in each fatty acid molecule, and the average carbon chain length CL of the fatty acid chain in the triglyceride molecule. Under the steady-state condition of ignoring the contribution of the longitudinal relaxation time T1 and correcting the inhomogeneity of the main magnetic field B0 field, the real part S of the signal of the final tissue image real The equation regarding the echo time TE is expressed as follows: where w and f represent the number of water and triglyceride molecules, respectively, represents the number of protons in the fat spectral component k, f k represents the frequency shift between water and the k-th fat resonance peak, n water represents the number of protons in a water molecule, T2* represents transverse relaxation, and i is the imaginary unit used to represent the phase information and frequency components of the signal; CL and nmidb are expressed as functions of ndb, specifically CL = 16.8 + 0.25×ndb and nmidb = 0.093×ndb 2 ; S33: Repeat the fitting process to fit the parameters w, f, and nmidb. The values of ndb and nmidb are converted to the values of saturated fatty acid f SFA , monounsaturated fatty acid f MUFA and polyunsaturated fatty acid f PUFA in the numerical values of fatty acid chemical components, specifically: where the proportion f of tri-unsaturated fatty acids TriUFA is represented by a fixed value of 2%.

2. The method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to claim 1, wherein, In step S1, the sequence parameters include: repetition time (TR), echo time (TE), field of view (FOV), acquisition matrix, slice thickness, slice gap, acquisition time, and readout bandwidth. Among them, the repetition time (TR) is the interval between two radiofrequency pulses during the MRI scan. The echo time (TE) is the time interval from the emission of the radiofrequency pulse to signal acquisition. There are multiple echo times (TE). The field of view (FOV) is the size of the physical area covered by the obtained original magnitude image. The acquisition matrix is the number of pixels in the original magnitude image, usually expressed as the number of rows and columns. The slice thickness defines the thickness of each slice. The slice gap is the distance between adjacent slices. The acquisition time is the total time required to complete one MRI scan. The readout bandwidth is the frequency range per pixel during signal acquisition. In step S1, the original magnitude image is the image information extracted from the amplitude part of the MRI signal, reflecting the signal intensity at each pixel position. The original phase image is the image information extracted from the phase part of the MRI signal, reflecting the phase angle of the signal at each pixel position.

3. The method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to claim 1, wherein In step S2, segment the original magnitude image to form the bi-clustering segmentation mask. Apply the bi-clustering segmentation mask to remove the background signal in the original phase image, and only retain the tissue signal of interest, specifically as follows: Use the biclustering method to segment the tissue and background in the original magnitude image, suppress the background, retain only the tissue, form the biclustering segmentation mask, and apply the biclustering segmentation mask to the original phase image to remove the background signal and retain only the tissue signal of interest; Among them, the biclustering method uses k-means clustering, selects k = 2, segments the original magnitude image into two clusters, one representing the tissue signal and the other representing the background signal, suppresses the background cluster, and retains only the tissue signal cluster to form the biclustering segmentation mask.

4. The method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to claim 1, wherein, In step S2, correct the zero-order phase and first-order phase of the original phase image with multiple echoes, and combine the original magnitude image and the corrected original phase image to generate the final tissue image. Specifically: Unwrap the original phase image of the multi-echo by using the minimum cost network flow for each echo, and extract the zero-order phase and the first-order phase pixel by pixel from the unwrapped original phase image. This process uses the phase noise ratio difference related to the echo time TE of the gradient echo GRE sequence model, and fits the phase of the magnetic resonance imaging MRI signal using the weighted linear least squares method : Among them, represents the zero-order phase, represents the first-order phase, is the echo time TE; For the heterogeneity ΔB0 of the main magnetic field B0 field in the magnetic resonance imaging (MRI) device, it is derived by using ΔB0 = / 2π. By using the zero-order phase and the first-order phase of the original phase image of the calculated corrected multi-echo, the phase error caused by the heterogeneity of the main magnetic field B0 field is eliminated; Combine the original magnitude image and the corrected original phase image to generate the final tissue image, accurately reflecting the internal structure and composition of the tissue.

5. The method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to claim 1, wherein It also includes: In step S1, set the echo intervals and the number of echoes of different echo times TE, and calculate the values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids at different echo intervals and the number of echoes respectively by using the method of nuclear magnetic resonance imaging for detecting fatty acid chemical components in steps S1-S3; For the values of fatty acid chemical components calculated for different echo intervals and the number of echoes, use the mean absolute error MAE to evaluate the influence of the echo interval and the number of echoes on the calculation accuracy of fatty acid chemical components FAC. The formula is: Among them, represents the reference value, represents the calculated measured value.

6. The method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to claim 1, characterized in that, It also includes: Perform a linear correlation analysis and comparison on the measured values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids calculated by using the method of nuclear magnetic resonance imaging for detecting fatty acid chemical components in steps S1-S3 with the reference values to evaluate the accuracy of quantitative analysis MR-FAC quantification of magnetic resonance imaging MRI; Use the Pearson linear correlation coefficient r and its 95% confidence interval to test the linearity, and perform a linear regression analysis including the intercept, slope, and the corresponding 95% confidence range; In addition, the p-value for testing the null hypothesis that there is no linear relationship between the measured values and the reference values is also calculated, and a significance level P < 0.05 is considered to be statistically significant.

7. The method for detecting fatty acid chemical components by nuclear magnetic resonance imaging according to claim 1, characterized in that, It also includes: Evaluate consistency: Two measurers measure respectively to obtain two groups of measured values M1 and M2 of the CSE-MRI sequence of magnetic resonance imaging MRI based on the multi-echo gradient echo and chemical shift encoding for ICC consistency analysis; The evaluation criteria for ICC consistency analysis are: ICC < 0.4 indicates poor correlation, ICC between 0.40 and 0.59 indicates moderate correlation, ICC between 0.60 and 0.74 indicates good correlation, and ICC > 0.74 indicates excellent correlation; Evaluate repeatability: Two sets of measurement values, scan1 and scan2, of the CSE-MRI sequence of the magnetic resonance imaging (MRI) based on the multi-echo gradient echo and the chemical shift encoding are obtained by a single measurer through two measurements before and after, and a statistical Bland-Altman analysis is performed on them. The evaluation criterion for the Bland-Altman analysis is that at least 95% of the differences between the two measurements are within the limits of agreement, that is, more than 95% of the points in the graph are within the limits of agreement.

8. A nuclear magnetic resonance imaging system for detecting fatty acid chemical components by using the method for detecting fatty acid chemical components in nuclear magnetic resonance imaging according to any one of claims 1-7, characterized in that, It includes: An original image acquisition module, configured to set sequence parameters of magnetic resonance imaging (MRI) based on multi-echo gradient echo and chemical shift encoding, perform scanning of the magnetic resonance imaging (MRI) based on the sequence parameters, and obtain multiple sets of original magnitude images and original phase images at different echo times (TE). A tissue image generation module, configured to segment the original magnitude image to form a bi-clustering segmentation mask, apply the bi-clustering segmentation mask to remove background signals in the original phase image, only retain tissue signals of interest, correct the zero-order phase and the first-order phase of the original phase image with multiple echoes, and combine the original magnitude image and the corrected original phase image to generate a final tissue image. A chemical component calculation module, configured to model the real part of the signal of the final tissue image, separate the proton densities of fat and water, calculate the proton density fat fraction, repeatedly fit the parameters in the model, and finally calculate the numerical values of fatty acid chemical components including saturated fatty acids, monounsaturated fatty acids, and polyunsaturated fatty acids.

9. A computer-readable storage medium storing computer code, which when executed, performs the method according to any one of claims 1 to 7.

Images