Real-time dynamic MRI tumor imaging method and system based on compressed sensing

Abstract

The application provides a real-time dynamic MRI tumor imaging method and system based on compressed sensing, and relates to the technical field of medical images. The application collects dynamic signal data of two types of weighted images generated by structural transformation of a detection device at different time points through compressed sensing technology. A multi-echo water-fat separation sequence is used for dynamic scanning, signal separation is performed according to water-fat chemical shift difference, an initial water map and a non-water phase map are generated, time resolution analysis is performed in combination with dynamic signal data and the two types of images, a signal intensity curve of the detection device is generated, and specific enhanced signals are separated. Finally, in combination with time dimension information, an image is reconstructed by using a compressed sensing algorithm, and a dynamic enhancement image representing a target tissue is obtained. The application can realize precise real-time dynamic MRI tumor imaging by suppressing artifacts through the time evolution mode of the detection device.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of medical imaging technology, in particular to a real-time dynamic MRI tumor imaging method and system based on compressed sensing. BACKGROUND

[0002] In scenarios such as accurately distinguishing tumor tissue from surrounding normal tissue and early detecting small tumor lesions, imaging technology needs to have high spatial resolution, high temporal dynamics and strong specificity to clearly present the tumor boundary and capture the signal of small lesions, so as to provide accurate basis for clinical treatment plan and evaluation of efficacy.

[0003] Currently, the commonly used scheme in clinic is dynamic contrast-enhanced magnetic resonance imaging technology. This technology uses the perfusion difference of contrast agents between tissues to dynamically acquire MRI signals at different time points, and obtains the blood perfusion information of tumor tissue after image reconstruction, so as to assist in judging the position and range of tumor.

[0004] However, the existing scheme has obvious defects: the contrast agent lacks tumor targeting, the enhancement difference between normal tissue and tumor tissue is small, the specificity is insufficient, and it is easy to cause the tumor boundary to be blurred; the detection ability of small tumor lesions is limited, which is limited by spatial resolution and signal-to-noise ratio, and it is difficult to capture the signal change of small lesions; the scanning time is long, the temporal resolution is low, the instantaneous signal change of tumor tissue cannot be monitored in real time, and it is easy to be disturbed by motion artifacts, affecting the imaging accuracy. SUMMARY

[0005] The present application aims to provide a real-time dynamic MRI tumor imaging method and system based on compressed sensing to solve the problem of low temporal resolution in the prior art.

[0006] To solve the above technical problems, in a first aspect, the present application provides a real-time dynamic MRI imaging method based on compressed sensing, comprising:

[0007] Using compressed sensing technology to collect dynamic signal data of the detection device on the first type of weighted image and the second type of weighted image caused by structural transformation at different time points;

[0008] Using a multi-echo water-fat separation sequence to dynamically scan the target region, synchronously acquiring multi-echo data during the dynamic scanning process, separating the multi-echo data according to the chemical shift difference between water and fat to obtain water signal and fat signal, and generating an initial water map based on the separated water signal through MRI image reconstruction, and generating a non-water phase map based on the separated fat signal through MRI image reconstruction;

[0009] based on the dynamic signal data, the initial water map and the non-water map, performing time-resolved analysis on the multi-echo data to generate a signal intensity curve of the detection device over time, and separating specific first type and second type enhanced signals according to the enhancement characteristics of the first type and second type weighted image signals at different time points in the signal intensity curve;

[0010] According to the specific first type and second type enhanced signals, and in combination with the time dimension information in the dynamic signal data, a compressed sensing algorithm is used to perform MRI image reconstruction to obtain a dynamic enhancement image of the target tissue, and a specific evolution pattern of the detection device in the time dimension is used to identify and suppress artifacts in the dynamic enhancement image.

[0011] Optionally, according to the specific first type and second type enhanced signals, and in combination with the time dimension information in the dynamic signal data, a compressed sensing algorithm is used to perform MRI image reconstruction to obtain a dynamic enhancement image of the target tissue, and a specific evolution pattern of the detection device in the time dimension is used to identify and suppress artifacts in the dynamic enhancement image, including:

[0012] The specific first type and second type enhanced signals are combined into a joint enhanced signal, and the spatial position of the joint enhanced signal is consistent with the dynamic signal data;

[0013] Based on the time dimension information of the dynamic signal data, a time evolution constraint function is constructed, which describes the prior pattern of the signal intensity of the detection device over time;

[0014] The joint enhanced signal is taken as the initial input, and the time evolution constraint function is taken as the sparsity constraint to construct a target optimization function, and the signal estimation value of each spatial position at consecutive time nodes is updated by iteratively solving the target optimization function;

[0015] The updated signal estimation value is compared with the dynamic signal data for residual error, and if the residual error does not converge, the iteration is repeated, and when the residual error converges, the signal estimation value is Fourier transformed to reconstruct a dynamic enhancement image of the target tissue;

[0016] During the reconstruction process, abnormal fluctuation points in the detection signal intensity curve that do not conform to the time evolution constraint function are detected, the spatial positions corresponding to the abnormal fluctuation points are marked as artifact interference regions, and the signal intensity of the artifact interference regions in the dynamic enhancement image is set to zero.

[0017] Optionally, the joint enhancement signal is taken as an initial input, and the time evolution constraint function is taken as a sparsity constraint to construct an objective optimization function, and the signal estimation value of each spatial position at a continuous time node is updated by iteratively solving the objective optimization function, including:

[0018] The joint enhancement signal is assigned to the signal estimation value as an initial value, and the time evolution constraint function is taken as a basis for sparsity evaluation to construct the objective optimization function;

[0019] The objective optimization function includes a data matching part between the signal estimation value and the dynamic signal data and a pattern matching part of the signal estimation value under the time evolution constraint function;

[0020] The signal estimation value is updated through a loop calculation process, and in each loop, the objective optimization function value corresponding to the current signal estimation value is calculated, and the signal estimation value is adjusted based on the objective optimization function value;

[0021] The loop calculation process is repeated until the change of the objective optimization function value is lower than a preset stop condition, and the updated signal estimation value is output.

[0022] Optionally, based on the dynamic signal data, the initial water map and the non-water phase map, the multi-echo data is subjected to time resolution analysis to generate a signal intensity curve of the detection device changing with time, and specific first and second enhancement signals are separated according to the enhancement characteristics of the first and second weighted image signals at different time points in the signal intensity curve, including:

[0023] The initial water map is taken as a spatial mask to eliminate the signal values of the background region in the dynamic signal data, and the fat region is marked in the non-water phase map to filter out the signal values of the fat region in the dynamic signal data;

[0024] Based on the signal values obtained after eliminating the background region and filtering out the fat region, a two-dimensional matrix about the signal intensity of the time node and the spatial position is constructed, wherein the rows of the two-dimensional matrix correspond to the spatial positions, and the columns correspond to the time nodes;

[0025] The data of the signal intensity changing with time in each row of the two-dimensional matrix is subjected to curve fitting to generate a first weighted image signal intensity curve and a second weighted image signal intensity curve of each spatial position, respectively;

[0026] According to the rising slope and peak position of the first weighted image signal intensity curve, the region meeting the preset threshold condition is marked as a specific first enhancement signal;

[0027] According to the falling delay characteristic of the second type of weighted image signal intensity curve, mark the region satisfying the preset time delay condition as a specific second type of enhanced signal.

[0028] Optionally, the data of signal intensity change with time of each row in the two-dimensional matrix is curve-fitted to generate the first type of weighted image signal intensity curve and the second type of weighted image signal intensity curve of each spatial position respectively, including:

[0029] Extract the time-series signal data corresponding to each spatial position in the two-dimensional matrix, establish the corresponding relationship between the time node and the signal intensity value, and form a discrete data point set;

[0030] According to the physical characteristics of the detection device, the signal data is divided into a first signal channel and a second signal channel, wherein the first signal channel corresponds to a fast response feature, and the second signal channel corresponds to a delayed response feature;

[0031] For the first signal channel, a polynomial function approximation method is used to determine the fitting parameters by least squares method to obtain a continuous function expression of the first signal channel;

[0032] For the second signal channel, an exponential decay function model is used for fitting, and the decay constant and initial amplitude parameters are solved by nonlinear regression method to obtain a continuous function expression of the second signal channel;

[0033] According to the reliability of signal intensity, different weights are assigned to the two continuous function expressions, and the continuous function expressions processed by the weight factor are smoothed to eliminate high-frequency noise, to generate the first type of weighted image signal intensity curve and the second type of weighted image signal intensity curve.

[0034] Optionally, a compressed sensing technology is used to collect dynamic signal data of the detection device on the first type of weighted image and the second type of weighted image caused by structural transformation at different time points, including:

[0035] At each time node, a first type of excitation pulse sequence is transmitted using the radio frequency coil of the magnetic resonance device to excite the first type of signal response of the detection device at the frequency parameter corresponding to the first type of weighted image, and the first type of frequency domain data is collected based on a random frequency sampling template;

[0036] At the same time node, switch to the frequency parameter corresponding to the second type of weighted image, transmit a second type of excitation pulse sequence to excite the second type of signal response of the detection device, and collect the second type of frequency domain data based on a random frequency sampling template;

[0037] Integrate the first type of frequency domain data and the second type of frequency domain data of each time node in chronological order to form dynamic signal data containing spatial coordinates, time nodes and signal type identification.

[0038] Optionally, a multi-echo water-fat separation sequence is used to perform dynamic scanning on the target region, multi-echo data is synchronously acquired during the dynamic scanning, the multi-echo data is separated according to the chemical shift difference between water and fat to obtain water signals and fat signals, an initial water map is generated based on the separated water signals through MRI image reconstruction, and a non-water phase map is generated based on the separated fat signals through MRI image reconstruction, comprising:

[0039] A multi-echo gradient echo sequence is used to perform continuous scanning on the target region, and frequency domain raw data is synchronously acquired at each echo time point. The frequency domain raw data at each echo time point is combined into a multi-channel multi-echo data block.

[0040] Based on the preset chemical shift difference between water and fat, a phase difference evolution model containing phase accumulation functions at different echo time points is constructed. The multi-echo data block is input into the phase difference evolution model, and the water signals and fat signals of each voxel are separated by solving linear equations.

[0041] The separated water signals are inverse Fourier transformed and reconstructed into a two-dimensional initial water map, and the separated fat signals are inverse Fourier transformed and reconstructed into a two-dimensional non-water phase map.

[0042] In a second aspect, the present application provides a real-time dynamic MRI tumor imaging system based on compressed sensing, comprising:

[0043] The acquisition module uses compressed sensing technology to acquire dynamic signal data of the detection device on the first and second weighted images caused by structural changes at different time points.

[0044] The separation module is used to perform dynamic scanning on the target region using a multi-echo water-fat separation sequence, synchronously acquire multi-echo data during the dynamic scanning, separate the multi-echo data according to the chemical shift difference between water and fat to obtain water signals and fat signals, generate an initial water map based on the separated water signals through MRI image reconstruction, and generate a non-water phase map based on the separated fat signals through MRI image reconstruction.

[0045] The generation module is used to perform time-resolved analysis on the multi-echo data based on the dynamic signal data, the initial water map and the non-water phase map, generate a signal intensity curve of the detection device changing with time, and separate specific first and second enhancement signals according to the enhancement characteristics of the first and second weighted image signals at different time points in the signal intensity curve.

[0046] The reconstruction module is used to reconstruct MRI images using a compressed sensing algorithm based on the specific first-type enhancement signal and the second-type enhancement signal, combined with the time dimension information in the dynamic signal data, to obtain a dynamic enhanced image of the target tissue, and to identify and suppress artifact interference in the dynamic enhanced image by utilizing the specific evolution pattern of the detection device in the time dimension.

[0047] Thirdly, this application provides an electronic device, comprising:

[0048] Memory, used to store computer programs;

[0049] A processor, configured to execute the computer program to implement the steps of a real-time dynamic MRI tumor imaging method based on compressed sensing as described in the first aspect above.

[0050] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed by a processor, can implement the steps of a real-time dynamic MRI tumor imaging method based on compressed sensing as described in the first aspect above.

[0051] This application provides a real-time dynamic MRI tumor imaging method based on compressed sensing. By employing compressed sensing technology to acquire dynamic signal data of two types of weighted images generated by structural changes at different time points, it can capture specific dynamic signals related to the target tissue by leveraging the targeting and structural change characteristics of the detection device's microenvironment. Simultaneously, compressed sensing technology can efficiently acquire signals, laying the foundation for accurate target tissue identification. By using a multi-echo water-lipid separation sequence to dynamically scan and separate water and fat signals, generating initial water and non-water phase images, interference between water and fat signals can be eliminated, resulting in pure basic signals and images, providing high-quality data for subsequent analysis. By combining dynamic signal data with the two types of images for time-resolved analysis, signal intensity curves are generated and specific enhanced signals are separated. This accurately captures the signal change patterns of the detection device over time, and signals only related to the target tissue are selected based on enhancement features, effectively eliminating interference from normal tissue and improving the specificity of target tissue identification. Based on the specific enhanced signals and time dimension information, the image is reconstructed using a compressed sensing algorithm, and artifacts are suppressed by utilizing the temporal evolution mode of the detection device, ultimately obtaining a highly accurate dynamic enhanced image of the target tissue. Attached Figure Description

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the accompanying drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced. Obviously, the accompanying drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0053] Figure 1 A flowchart of a real-time dynamic MRI tumor imaging method based on compressed sensing provided by an embodiment of the present application;

[0054] Figure 2 A specific implementation flowchart of a real-time dynamic MRI tumor imaging method based on compressed sensing provided by an embodiment of the present application;

[0055] Figure 3 A scene diagram of a real-time dynamic MRI tumor imaging method based on compressed sensing provided by an embodiment of the present application;

[0056] Figure 4 A structural schematic diagram of a real-time dynamic MRI tumor imaging system based on compressed sensing provided by an embodiment of the present application. DETAILED DESCRIPTION

[0057] In the process of tumor image imaging, the existing commonly used dynamic contrast-enhanced magnetic resonance imaging technology has obvious deficiencies and is difficult to meet the needs of clinical precise detection. The gadolinium-based contrast agent used by this technology has no tumor specificity, and the imaging enhancement difference between the tumor tissue and the surrounding normal tissue is small, so that it is difficult for doctors to clearly distinguish the tumor boundary; for small early-stage tumor lesions, due to insufficient imaging clarity and signal sensitivity, it is also difficult to accurately detect; at the same time, its scanning time is long, and it cannot capture the dynamic changes of the tumor in real time, and it is also easy to produce image interference due to slight body movement, affecting the accuracy of tumor imaging.

[0058] In view of the above problems, the present application provides a real-time dynamic MRI tumor imaging method based on compressed sensing. The method first uses compressed sensing technology to capture the imaging signals of the detection device at different times; then separates the signals of water and fat in the image through a special scanning sequence to generate a clearer base image; then combines these signals and images to analyze the signal change rule of the detection device and filter out the specific signals belonging to the target tissue; finally, the image is reconstructed using the compressed sensing algorithm to obtain a dynamic enhancement image showing only the target tissue, and the image interference can be eliminated by using the signal change rule of the detection device. The method improves the distinction between target tissue and normal tissue by targeted detection device, improves the detection ability of small lesions by signal separation and accurate algorithm, and at the same time shortens the scanning time and reduces the image interference.

[0059] The present application mainly takes the target tissue including active tumor tissue as an example to illustrate the following embodiments. In order to make the person skilled in the art better understand the present application scheme, the present application will be further described in detail below in combination with the drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by the person skilled in the art without creative labor are within the scope of protection of the present application.

[0060] The core of the present application is to provide a real-time dynamic MRI tumor imaging method based on compressed sensing. The flowchart of a specific embodiment is shown in Figure 1 The method comprises the following steps.

[0061] S101, using compressed sensing technology to collect dynamic signal data of the detection device on the first type of weighted image and the second type of weighted image caused by structural transformation at different time points;

[0062] Optionally, step S101 can specifically include the following steps:

[0063] S1011, at each time node, using the radio frequency coil of the magnetic resonance device to emit the first type of excitation pulse sequence to excite the first type of signal response of the detection device with the first type of weighted image corresponding frequency parameter, and collecting the first type of frequency domain data based on the random frequency sampling template;

[0064] S1012, at the same time node, switch to the frequency parameter corresponding to the second type of weighted image, emit the second type of excitation pulse sequence to excite the second type of signal response of the detection device, and collect the second type of frequency domain data based on the random frequency sampling template;

[0065] S1013, integrate the first type of frequency domain data and the second type of frequency domain data of each time node in time sequence to form dynamic signal data containing spatial coordinates, time nodes and signal type identification.

[0066] In the above steps, the detection device in actual application can be a pH-responsive bimodal nanoprobes, which is a nano-scale detection material sensitive to environmental pH value, which can change its conformation under acidic conditions such as tumor microenvironment, and can produce two different signal responses. The target area refers to the biological tissue sample or preclinical model in the magnetic resonance scanning environment. The set time node is a plurality of specific time points selected for signal acquisition, used to capture the detected signal characteristics over time; the random frequency sampling template is generated for each time node, which is used to guide the magnetic resonance device to collect frequency domain data, and the sampling position is randomly generated and meets the preset compression rate condition. The compression rate condition is the ratio requirement of the amount of sampling data to the amount of complete data, which can reduce the amount of sampling under the premise of ensuring signal quality; the radio frequency coil of the magnetic resonance device is a component of the magnetic resonance device for transmitting excitation pulses and receiving signals; the first type of excitation pulse sequence and the second type of excitation pulse sequence are two different parameter radio frequency pulse sequences, which are used to excite the first type of weighted image and the second type of weighted image corresponding to the signal response of the detection device; the first type of weighted image and the second type of weighted image are two different imaging weight MRI images, which can reflect the signal characteristics of different tissues; the first type of frequency domain data and the second type of frequency domain data are the original signal data in the frequency domain collected by the corresponding excitation pulse sequence; the dynamic signal data is a comprehensive signal data set integrating the two types of frequency domain data of each time node, and containing spatial coordinates, time nodes and signal type identification.

[0067] In the embodiment of the present application, first, the radio frequency coil of the magnetic resonance device is used to transmit the first type of excitation pulse sequence in step S1011 to excite the first type of signal response of the detection device with the frequency parameter corresponding to the first type of weighted image. The first type of frequency domain data is collected based on the random frequency sampling template. For example, at the first time node, that is, 0.5 hours after starting detection, the medical staff helps the target area D lie in the scanning cabin of the magnetic resonance device, adjusts the body position to a comfortable and stable state, and avoids affecting data acquisition during scanning due to body position changes. Subsequently, the magnetic resonance device is started, and the radio frequency coil of the device is controlled to transmit the first type of excitation pulse sequence, and the frequency parameter of the sequence is set to a specific resonance frequency matched with the first type of weighted image. At the same time, the magnetic resonance device only collects the signal data of the frequency position marked in the template according to the random frequency sampling template corresponding to the time node. The collection process lasts about 5 minutes, and finally the first type of frequency domain data of the time node is obtained. The data is saved through the storage module of the device, and can reflect the first type of signal characteristics generated by the detection device at different positions at this time.

[0068] Then, at the same time node, the second type of excitation pulse sequence corresponding to the second type of weighted image is switched to by step S1012 to excite the second type of signal response of the detection device, the second type of frequency domain data is collected based on the random frequency sampling template, for example, after the first time node is completed, that is, the detection starts at 0.5 hours and the first type of frequency domain data is collected, the position of the target region D is not moved, the control terminal of the magnetic resonance device automatically switches the frequency parameter to the value matched with the second type of weighted image, and then controls the radio frequency coil to emit the second type of excitation pulse sequence. The sequence generates a signal response corresponding to the second type of weighted image. According to the random frequency sampling template of the time node, the device collects signal data of the template marked frequency position, and the collection time is the same as that of the first type of frequency domain data, about 5 minutes, and finally the second type of frequency domain data of the time node is obtained. The data is also saved to the storage module, which can reflect the second type of signal characteristics generated at different positions at the same time, realize continuous collection of two types of signals at the same time node, and avoid signal deviation caused by body position change of the target region.

[0069] Finally, the first type of frequency domain data and the second type of frequency domain data of each time node are integrated in time sequence by step S1013 to form dynamic signal data containing spatial coordinates, time nodes and signal type identifiers, for example, after the two types of frequency domain data of four time nodes are collected, the computer retrieves all data from the storage module of the magnetic resonance device, and arranges them in time from early to late, that is, the two types of data of 0.5 hours are arranged first, and then the two types of data of 1 hour, 1.5 hours and 2 hours are arranged in turn. Then, the computer obtains the spatial coordinates corresponding to each data through the positioning function of the magnetic resonance device, that is, the specific position information of each region in the body, and labels each group of data with the corresponding time node and signal type identifier. After labeling, the computer integrates these data into a complete dynamic signal data file, which can be opened and viewed through professional software, clearly presenting the changes of the two types of signals generated by the detection device at different times and different positions, providing basic data for subsequent image analysis and reconstruction.

[0070] In the overall scheme of the above step S101, by setting multiple time nodes and generating a random frequency sampling template, signal changes over time can be tracked, and the amount of sampling data can be reduced based on the compressed sensing technology to improve signal collection efficiency; by collecting two types of frequency domain data at the same time node, double signal information of the detection device can be obtained, and the signal dimension is enriched; by integrating the data to form dynamic signal data containing multi-dimensional information, comprehensive and continuous signal support can be provided for subsequent image analysis and reconstruction, and the overall efficient and accurate collection of related dynamic signals of the target tissue is realized, which lays a data foundation for subsequent generation of specific images.

[0071] S102, performing dynamic scanning on the target region by using a multi-echo water-fat separation sequence, synchronously acquiring multi-echo data in the dynamic scanning process, separating the multi-echo data according to the chemical shift difference between water and fat to obtain water signals and fat signals, and generating an initial water map based on the separated water signals through MRI image reconstruction, and generating a non-water phase map based on the separated fat signals through MRI image reconstruction;

[0072] Optionally, step S102 can specifically include the following steps:

[0073] S1021, performing continuous scanning on the target region by using a multi-echo gradient echo sequence, synchronously acquiring frequency domain raw data at each echo time point, and combining the frequency domain raw data at each echo time point into a multi-channel multi-echo data block;

[0074] S1022, constructing a phase difference evolution model containing phase accumulation functions at different echo time points based on a preset chemical shift difference between water and fat, inputting the multi-echo data block into the phase difference evolution model, and separating water signals and fat signals of each voxel by solving linear equations;

[0075] S1023, performing inverse Fourier transform on the separated water signals to reconstruct a two-dimensional initial water map, and performing inverse Fourier transform on the separated fat signals to reconstruct a two-dimensional non-water phase map.

[0076] In the above steps, the multi-echo water-fat separation sequence is a magnetic resonance scanning sequence that can acquire multiple echo signals in one scan and separate the signals based on the chemical difference between water and fat; dynamic scanning refers to continuous and uninterrupted scanning of the target region to capture the dynamic changes of signals in the region; multi-echo data is all signal data acquired at multiple echo time points during dynamic scanning; the chemical shift difference between water and fat refers to the difference in resonance frequency of water molecules and fat molecules in a magnetic resonance field due to different chemical structures, which is the core basis for water-fat signal separation; water signal is a magnetic resonance signal only from water molecules separated from multi-echo data, and fat signal is a magnetic resonance signal only from fat molecules; MRI image reconstruction is the process of converting acquired frequency domain signal data into visual images; the initial water map is an MRI image reconstructed based on water signal, mainly showing the distribution of water molecules in the body; the non-water map is an MRI image reconstructed based on fat signal, mainly showing the distribution of fat tissue in the body; the multi-echo gradient echo sequence is a specific implementation form of the multi-echo water-fat separation sequence, which can acquire multiple echo signals in a short time through gradient magnetic field control; the target region refers to the region to be scanned in the target region selected according to the detection requirements, such as the suspected target tissue site; the echo time point refers to a specific time node for acquiring echo signals during scanning, and signals at different echo time points carry different information; the frequency domain raw data is unprocessed signal data in the frequency domain directly acquired from a magnetic resonance device; the multi-channel multi-echo data block is a structured data set formed by integrating frequency domain raw data at multiple echo time points according to channel and time order; the preset chemical shift difference value is a value preset based on the known resonance frequency difference between water and fat, used to construct a signal separation model; the phase difference evolution model is a mathematical model describing the phase change law of water and fat signals at different echo time points, including phase accumulation functions at different echo time points; a voxel is the smallest pixel unit in an MRI image, and each voxel corresponds to the signal of a small region in the body; inverse Fourier transform is a signal processing technique that can convert frequency domain signals to spatial domain signals, and is a key technology for MRI image reconstruction.

[0077] In the embodiments of the present application, first, the target region is continuously scanned by using a multi-echo gradient echo sequence in step S1021, and the frequency domain raw data is synchronously acquired at each echo time point. The frequency domain raw data at each echo time point is combined into a multi-channel multi-echo data block. For example, when the target region D of the suspected target tissue is detected, the region where the abdominal suspected tumor of the target region D is located is determined as the target region, the magnetic resonance device is started, and the multi-echo gradient echo sequence is selected. The scanning parameters are set so that the device generates 8 echo time points in one continuous scanning, such as from 1 ms to 8 ms, and a time point is set every 1 ms. During the scanning process, the multi-channel radio frequency coil of the device synchronously acquires the frequency domain raw data of the target region at each echo time point, and each channel corresponds to a group of independent raw data. After the scanning is completed, the frequency domain raw data of each channel at 8 echo time points is integrated in the order of “channel-echo time point” to form a multi-channel multi-echo data block containing channel information, time information and signal data, which provides complete raw data support for subsequent water-fat signal separation.

[0078] Secondly, a phase difference evolution model containing phase accumulation functions at different echo time points is constructed based on the preset chemical shift difference of water and fat in step S1022, the multi-echo data block is input into the phase difference evolution model, and the water signal and the fat signal of each voxel are separated by solving the linear equation set. For example, first, the preset chemical shift difference of water and fat is set according to the known resonance frequency difference between water and fat. Then, based on the difference, a phase difference evolution model is constructed. The model contains the phase accumulation function corresponding to each echo time point, which can describe the change rule of the phase difference between water and fat signals with the passage of echo time. Then, the multi-channel multi-echo data block obtained in step S1021 is input into the model. The model will establish a linear equation set about the water signal intensity and the fat signal intensity for each voxel in the data block according to the signal intensity and the phase information at different echo time points. By solving the linear equation set corresponding to each voxel through matrix operation, the water signal and the fat signal of the voxel can be extracted from the mixed signal respectively, the water-fat signal separation is realized, and the interference of the two signals on subsequent imaging caused by mutual superposition is avoided.

[0079] Finally, the separated water signals are inverse Fourier transformed by step S1023 to reconstruct a two-dimensional initial water map, and the separated fat signals are inverse Fourier transformed to reconstruct a two-dimensional non-water phase map. For example, all the water signals of the voxels are extracted from the separation result of step S1022, and these water signals are arranged in the order of spatial positions of the voxels to form a complete water signal data set. The inverse Fourier transform technique is applied to the data set to convert the water signals originally in the frequency domain into image signals in the spatial domain. After post-processing such as image noise reduction and contrast adjustment, a two-dimensional initial water map is generated. In the image, the bright area corresponds to the dense water molecule area, and the water distribution in the target region can be clearly presented. At the same time, the fat signals of all the voxels are extracted and arranged in the same order of spatial positions to form a fat signal data set. The inverse Fourier transform is also applied to convert the fat signals into image signals in the spatial domain. After post-processing, a two-dimensional non-water phase map is generated. In the image, the bright area corresponds to the dense fat tissue area, and the fat distribution in the target region can be directly displayed, laying a foundation for subsequent analysis of the relevant signals of the target tissue.

[0080] In actual application, first, the medical staff determines to take the liver as the scanning target region according to the imaging preliminary examination result of the target region E, selects a B brand magnetic resonance device and loads a multi-echo gradient echo sequence, sets the echo time points to 10, from 0.8 ms to 8.8 ms, with an interval of 0.8 ms, and simultaneously starts the 16-channel radio frequency coil of the device to improve the signal acquisition efficiency. After the scanning starts, the device continuously and dynamically scans the liver region of the target region E. At each echo time point, the 16-channel coil synchronously collects the frequency domain original data. After the scanning is completed, the device automatically integrates the original data of the 16 channels at the 10 echo time points into a multi-channel multi-echo data block and stores it to the image processing terminal of the device. Then, the terminal automatically constructs a phase difference evolution model based on the preset water and fat chemical shift difference value. After the multi-channel multi-echo data block is input into the model, the model separates the water signals and the fat signals one by one through the built-in linear equation system solving algorithm. The terminal displays the separation progress in real time during the separation process to ensure the stability of the data processing. Finally, the terminal performs inverse Fourier transform on the separated water signals and fat signals respectively, and then performs noise reduction, edge enhancement and other processing through the built-in image reconstruction algorithm to generate clear initial water maps and non-water phase maps. The medical staff can clearly distinguish the water molecule distribution and fat distribution of the liver region by viewing the two images through the display interface of the device. During the whole process, the operation software of the device automatically records all the parameter settings and processing steps, which is convenient for subsequent data tracing and review.

[0081] In the overall scheme of step S102, the dynamic scanning is performed by using a multi-echo gradient echo sequence, and the multi-channel multi-echo data blocks are combined, so that abundant echo signal data can be obtained at one time, sufficient original information is provided for water-fat signal separation, and separation errors caused by insufficient data are avoided; by constructing a phase difference evolution model based on the water-fat chemical shift difference value and solving a linear equation set, the signal separation can be accurately realized by using the essential difference between water and fat, the interference of the superposition of water signal and fat signal is effectively eliminated, and the purity of the separated water signal and fat signal is high; by inverse Fourier transform reconstruction of the initial water map and the non-water phase map, the abstract frequency domain signal can be converted into an intuitive visual image, and the distribution of water molecules and fat tissue in the body is clearly presented. The step successfully solves the problem of water-fat signal mixing in traditional MRI imaging, lays a high-quality image foundation for subsequent specific signal analysis of target tissue and accurate identification of target tissue and normal tissue, and the dynamic scanning mode can also meet the needs of the whole real-time dynamic MRI tumor imaging method, ensuring the continuity of data acquisition.

[0082] S103, based on the dynamic signal data, the initial water map and the non-water phase map, performing time-resolved analysis on the multi-echo data to generate a signal intensity curve of the detection device changing with time, and separating specific first type enhancement signal and second type enhancement signal according to the enhancement characteristics of the first type weighted image signal and the second type weighted image signal in the signal intensity curve at different time points.

[0083] Optionally, step S103 can specifically include the following steps:

[0084] S1031, taking the initial water map as a spatial mask, removing the signal values of the background region in the dynamic signal data, marking the fat region in the non-water phase map, and filtering out the signal values of the fat region in the dynamic signal data;

[0085] S1032, based on the signal values obtained after removing the background region and filtering out the fat region, constructing a two-dimensional matrix about time node and spatial position signal intensity, wherein the rows of the two-dimensional matrix correspond to spatial positions, and the columns correspond to time nodes;

[0086] S1033, curve fitting is performed on the data of the signal intensity changing with time in each row of the two-dimensional matrix, and a first type weighted image signal intensity curve and a second type weighted image signal intensity curve of each spatial position are generated respectively;

[0087] The step S1033 can specifically include the following processes: extracting the time sequence signal data corresponding to each spatial position in the two-dimensional matrix, establishing the corresponding relationship between the time node and the signal intensity value, and forming a discrete data point set; according to the physical characteristics of the detection device, the signal data is divided into a first signal channel and a second signal channel, wherein the first signal channel corresponds to a fast response feature, and the second signal channel corresponds to a delayed response feature; for the first signal channel, a polynomial function approximation method is used to determine the fitting parameters by the least square method to obtain a continuous function expression of the first signal channel; for the second signal channel, an exponential decay function model is used for fitting, and the decay constant and the initial amplitude parameter are solved by a nonlinear regression method to obtain a continuous function expression of the second signal channel; according to the reliability of the signal intensity, the two continuous function expressions are assigned different weights, and the continuous function expressions processed by the weight factor are smoothed to eliminate high-frequency noise, thereby generating a first type of weighted image signal intensity curve and a second type of weighted image signal intensity curve.

[0088] S1034, according to the rising slope and peak position of the first type of weighted image signal intensity curve, marking the region meeting the preset threshold condition as a specific first type of enhanced signal;

[0089] S1035, according to the falling delay characteristics of the second type of weighted image signal intensity curve, marking the region meeting the preset time delay condition as a specific second type of enhanced signal.

[0090] In the above steps, the time resolution analysis is a processing manner of analyzing signal data combined with time dimension information to capture the change rule of signal over time; the specific first / second type of enhanced signal is a first / second type of weighted image signal only related to the target tissue and having specific time enhancement characteristics; the spatial mask is a tool for screening the target region and excluding irrelevant regions by using the water molecule distribution information in the initial water map; the background region is a region with very low water content in the initial water map and has no clinical significance; the two-dimensional matrix is structured data formed by arranging the processed signal data according to the “spatial position-time node” dimension, with rows corresponding to different spatial positions in the body and columns corresponding to different signal acquisition time nodes; the first signal channel is a signal channel corresponding to the fast response characteristics of the detection device, and the second signal channel is a signal channel corresponding to the delayed response characteristics of the detection device; the weight factor is an importance coefficient assigned to different fitting functions according to the signal intensity reliability, which is used to optimize the curve accuracy; the rising slope is the inclination of the rising stage of the first type of weighted image signal intensity curve, and the peak position is the time point at which the curve reaches the highest signal intensity; the preset threshold condition is a standard for judging whether the first type of signal is a specific enhanced signal, such as the minimum rising slope and the peak position range; the falling delay characteristic is a feature that the second type of weighted image signal intensity curve lags behind the ordinary tissue signal in the falling stage, and the preset time delay condition is a delay time standard for judging whether the second type of signal is a specific enhanced signal.

[0091] In the embodiments of the present application, first, the initial water map is used as a spatial mask to remove the signal values of the background regions in the dynamic signal data by step S1031, and the fat regions are marked in the non-water phase map, and the signal values of the fat regions in the dynamic signal data are filtered out. For example, when processing the abdominal target region of the target region D, the regions with brightness lower than the set value in the initial water map are determined as background regions, such as the air regions at the edges of the abdomen and the non-target organ regions. The signal values corresponding to these background regions in the dynamic signal data are deleted by using the initial water map as a mask. At the same time, in the non-water phase map, the regions with brightness higher than the set value are marked as fat regions, such as the subcutaneous fat layer of the abdomen and the intra-abdominal adipose tissue. The signal values corresponding to these fat regions in the dynamic signal data are also deleted, and only the signal data of the non-background and non-fat regions in the target region is retained, thereby reducing the interference of irrelevant signals on subsequent analysis.

[0092] Secondly, a two-dimensional matrix about time node and spatial position signal intensity is constructed based on the signal values obtained after the background region is eliminated and the fat region is filtered out, wherein the rows of the two-dimensional matrix correspond to spatial positions, and the columns correspond to time nodes. For example, for the target region D abdominal target region, the processed signal data is arranged according to the “spatial position-time node”, that is, each row of the matrix represents a spatial position, such as a voxel in the target region, corresponding to a micro region in the body, and each column represents a signal acquisition time node, such as the previously set 0.5 hours, 1 hour, 1.5 hours and 2 hours. Each element in the matrix is the signal intensity value corresponding to “a spatial position-a time node”. The dispersed signal data is structured by the matrix, so as to facilitate subsequent analysis of the signal change over time according to the spatial position.

[0093] Then, the data of the signal intensity change over time represented by each row of the two-dimensional matrix is curve-fitted by step S1033 to generate the first type of weighted image signal intensity curve and the second type of weighted image signal intensity curve of each spatial position. The specific process is as follows: first, the time sequence signal data corresponding to each spatial position in the two-dimensional matrix is extracted, that is, the signal intensity values of the position arranged in the order of the time nodes are extracted to establish the corresponding relationship between the time nodes and the signal intensity values, and form a discrete data point set; second, according to the physical characteristics of the detection device, the signal data is divided into a first signal channel and a second signal channel, wherein the first signal channel corresponds to the fast response characteristics of the detection device in the target region, such as the rapid rise of the signal in a short time, and the second signal channel corresponds to the delayed response characteristics of the detection device, such as the slow decline of the signal after reaching the peak value; for the first signal channel, a polynomial function approximation method is adopted, and the coefficients of the polynomial are calculated by the least square method to minimize the error sum of squares of the fitting curve and the discrete data points, so as to obtain the continuous function expression of the first signal channel; for the second signal channel, an exponential decay function model is adopted for fitting, and the decay constant and the initial amplitude parameter in the model are solved by a nonlinear regression method, wherein the decay constant reflects the signal decline speed, and the initial amplitude parameter reflects the initial signal intensity, so as to obtain the continuous function expression of the second signal channel; finally, according to the signal intensity reliability of the two signal channels, such as the more stable signal of the first signal channel at the early time nodes and the corresponding weight allocated to the second signal channel at the later time nodes, different weight factors are allocated to the two continuous function expressions, and then the function expressions after weighting are smoothed to eliminate the curve fluctuations caused by high-frequency noise, so as to finally generate the first type of weighted image signal intensity curve and the second type of weighted image signal intensity curve of each spatial position. For example, for a spatial position in the target region D target region, the first type of curve shows a rapid rising trend from 0.5 hours to 1 hour after the detection starts, and the second type of curve shows a slow declining trend after 1 hour.

[0094] Then, the region satisfying the preset threshold condition is marked as specific first type enhanced signal according to the rising slope and peak position of the first type weighted image signal intensity curve in step S1034. For example, the rising slope threshold of the first type curve is preset, such as the minimum rising slope, and the peak position threshold is preset, such as the peak position range. The first type curve of all spatial positions in the target region Dtarget region is analyzed: if the curve rising slope of a spatial position reaches or exceeds the preset threshold, and the peak position is within the preset time range, it is indicated that the signal of the position meets the fast response feature of the detection device of the target tissue region, and the first type signal of the position is marked as specific first type enhanced signal; otherwise, if the rising slope is insufficient or the peak position deviates, it is determined as the related signal of the non-target tissue.

[0095] Finally, the region satisfying the preset time delay condition is marked as specific second type enhanced signal according to the descending delay characteristic of the second type weighted image signal intensity curve in step S1035. For example, the descending delay threshold of the second type curve is preset, such as the time from the signal reaching the peak to the half intensity of the peak is not less than a certain value. The second type curve of all spatial positions in the target region Dtarget region is analyzed: if the curve of a spatial position descends to the half intensity of the peak for a time reaching or exceeding the preset threshold, it is indicated that the signal of the position meets the delay attenuation feature of the detection device of the target tissue region, and the normal tissue signal descends faster, and the second type signal of the position is marked as specific second type enhanced signal; if the descending speed is too fast and does not reach the delay threshold, it is determined as the related signal of the non-target tissue.

[0096] In practical application, first, the staff imports the previously obtained target region E liver region dynamic signal data, initial water map and non-water phase map into A brand medical image processing software, the software automatically takes the initial water map as a spatial mask, removes the intestinal gas background region signal around the liver, and at the same time marks the fat envelope region of the liver surface in the non-water phase map, filters out the signal of the region, and only retains the signal data of the liver parenchyma region. Then, the software constructs a two-dimensional matrix based on the processed signal data: the matrix row corresponds to 1000 spatial positions in the liver parenchyma, which is divided by voxel, and the column corresponds to four time nodes of 0.4 hours, 0.8 hours, 1.2 hours and 1.6 hours after starting detection, and each matrix element fills the signal intensity value of the corresponding position and time. Subsequently, the software performs curve fitting on each row of matrix data: extract the time series data of each spatial position, divide the signal into a first signal channel and a second signal channel, the first signal channel corresponds to a rapid response of 0.4-0.8 hours, and the second signal channel corresponds to a delayed response of 0.8-1.6 hours, a cubic polynomial function is fitted to the first channel by least squares method, and an exponential decay function is fitted to the second channel by nonlinear regression, then according to the channel reliability, the first channel weight is 0.6, the second channel weight is 0.4, and the curve is smoothed to generate two types of signal intensity curves. Then, the software analyzes according to the preset threshold: for the first type of curve, if the rising slope is ≥0.5 signal units / hour and the peak value is within 0.7-0.9 hours, it is marked as specific first type of enhanced signal; for the second type of curve, if the time of falling to half the peak value is ≥0.5 hours, it is marked as specific second type of enhanced signal. During the whole process, the software displays the marking results in real time, and the staff can check the signal distribution of the suspected target tissue region through the interface.

[0097] In the overall scheme of the above step S103, by using the initial water map and the non-water phase map to remove the background and fat signal, the interference of irrelevant tissue signal is effectively excluded, and the subsequent analysis focuses on the effective signal of the target region; by constructing a two-dimensional matrix, the dispersed signal data is structured according to the "space-time" dimension, which facilitates systematic analysis of the signal change rule of each position; by curve fitting, the discrete signal data is converted into continuous signal intensity curve, which clearly presents the time response characteristics of the detection device at different positions, and through channel fitting and weight optimization, the accuracy of the curve is improved; by marking the specific signal through the specific enhancement characteristics of the curve, only the signal related to the target tissue is accurately selected, and the situation that the normal tissue signal is misjudged as the target tissue signal is avoided. Overall, through layer-by-layer processing and analysis, this step extracts the specific signal of the target tissue from complex signal data, provides a key basis for generating the image of the target tissue in the subsequent, and greatly improves the accuracy of target tissue signal recognition.

[0098] S104, according to the specific first type of enhanced signal and the second type of enhanced signal, and in combination with the time dimension information in the dynamic signal data, an MRI image is reconstructed by using a compressed sensing algorithm to obtain a dynamic enhanced image of the target tissue, and the specific evolution mode of the detection device in the time dimension is used to identify and suppress the artifact interference in the dynamic enhanced image.

[0099] Optionally, as shown in Figure 2 S104 can specifically include the following steps:

[0100] S1041, the specific first type of enhanced signal and the specific second type of enhanced signal are combined into a joint enhanced signal, and the spatial position of the joint enhanced signal is consistent with the dynamic signal data;

[0101] S1042, a time evolution constraint function is constructed based on the time dimension information of the dynamic signal data, and the time evolution constraint function describes the prior mode of the signal intensity of the detection device changing with time;

[0102] S1043, the joint enhanced signal is taken as the initial input, and the time evolution constraint function is taken as the sparsity constraint to construct a target optimization function, and the signal estimation value of each spatial position at the continuous time node is updated by iteratively solving the target optimization function;

[0103] In the step S1043, the joint enhanced signal is assigned to the signal estimation value as an initial value, and the time evolution constraint function is taken as a sparsity evaluation basis to construct the target optimization function. The target optimization function includes a data matching part between the signal estimation value and the dynamic signal data and a mode compliance part of the signal estimation value under the time evolution constraint function. The signal estimation value is updated through a loop calculation process. In each loop, the target optimization function value corresponding to the current signal estimation value is calculated, and the signal estimation value is adjusted based on the target optimization function value. The loop calculation process is repeated until the change of the target optimization function value is lower than a preset stop condition, and the updated signal estimation value is output.

[0104] S1044, the updated signal estimation value is compared with the dynamic signal data for residual error, and if the residual error does not converge, the iteration is repeated, and when the residual error converges, the signal estimation value is Fourier transformed to reconstruct a dynamic enhanced image of the target tissue;

[0105] S1045, in the reconstruction process, abnormal fluctuation points in the signal intensity curve that do not comply with the time evolution constraint function are detected, the spatial position corresponding to the abnormal fluctuation points is marked as an artifact interference region, and the signal intensity of the artifact interference region in the dynamic enhanced image is set to zero.

[0106] In the above steps, the joint enhanced signal is a comprehensive signal formed by combining the specific first-type enhanced signal and the specific second-type enhanced signal according to the spatial positions, which is completely matched with the dynamic signal data in spatial position and can reflect the signal characteristics of the target tissue; the time evolution constraint function is a prior mode describing the change of the signal strength of the detection device with time, which is constructed in advance based on the physical characteristics of the detection device and the response law of the target tissue microenvironment, and is used to limit the reasonable change range of the signal; the target optimization function is a mathematical function used to iteratively optimize the signal estimation value, which includes a data matching part and a mode matching part, the data matching part measures the difference between the signal estimation value and the dynamic signal data, and the mode matching part measures the degree of fit between the signal estimation value and the time evolution constraint function; the signal estimation value is a predicted value used to approximate the real target tissue signal, which is updated in the iteration process; the preset stop condition is a standard for judging whether the iteration is terminated, which is usually that the change of the target optimization function value is less than a certain minimum value; the residual comparison is a process of calculating the difference between the updated signal estimation value and the dynamic signal data, and the residual convergence means that the difference is less than a preset threshold, indicating that the signal estimation value is close enough to the real value; the abnormal fluctuation point is a signal point in the signal strength curve that deviates from the time evolution constraint function and has no reasonable physiological meaning; the artifact interference region is the spatial position corresponding to the abnormal fluctuation point, and the signal in these regions does not come from the target tissue, but is caused by motion, device noise and other factors.

[0107] In the embodiments of the present application, first, the specific first-type enhanced signal and the specific second-type enhanced signal are combined into a joint enhanced signal through step S1041, and the spatial position of the joint enhanced signal is consistent with the dynamic signal data. For example, when processing the abdominal target region of the target region D, the specific first-type enhanced signal and the specific second-type enhanced signal marked with all spatial positions are extracted first, and are combined according to the “spatial position-time node” dimension: the two types of signals at the same spatial position and the same time node are superimposed according to a preset proportion, such as 0.5 for the first-type signal and 0.5 for the second-type signal, to form the joint enhanced signal value at the position and the time point; after the joint enhanced signal values of all positions and time nodes are integrated, it is ensured that the spatial position distribution is completely consistent with the dynamic signal data, so as to avoid errors in subsequent analysis caused by position misalignment.

[0108] Secondly, a time evolution constraint function is constructed based on the time dimension information of the dynamic signal data in step S1042. The time evolution constraint function describes the prior mode of the signal intensity of the detection device changing with time, for example, the historical response law of the detection device in the target tissue region in the dynamic signal data of the target region D, such as the rapid rise of the signal in 0.5-1 hour and the slow decline in 1-2 hours. The time evolution constraint function is constructed: the function takes time as the independent variable and signal intensity as the dependent variable, limits the rising rate of the signal in 0.5-1 hour to be not lower than a certain value and the declining rate in 1-2 hours to be not higher than a certain value, and excludes sudden changes and sudden drops that do not conform to the characteristics of the detection device. The function can be realized in the form of a polynomial fitting or a piecewise function, providing a reasonable mode reference for subsequent signal optimization.

[0109] Next, the joint enhanced signal is taken as the initial input and the time evolution constraint function is taken as the sparsity constraint to construct a target optimization function in step S1043. The signal estimation value of each spatial position at a continuous time node is updated by iteratively solving the target optimization function. The specific process is as follows: first, all values of the joint enhanced signal are assigned to the signal estimation value as the initial value; then, the time evolution constraint function is taken as the basis for sparsity evaluation to construct the target optimization function. The data matching part of the function is the mean square error of the signal estimation value and the dynamic signal data, and the mode compliance part is the square sum of the deviation of the signal estimation value from the time evolution constraint function. The two parts are added together according to the preset weight, such as the data matching part weight 0.6 and the mode compliance part weight 0.4, to obtain the total value of the target optimization function. Subsequently, the loop calculation process is entered: in each loop, the target optimization function value corresponding to the current signal estimation value is calculated first, and then the signal estimation value is adjusted by the gradient descent algorithm, such as fine-tuning the signal estimation value to reduce the function value for the position with a larger function value. The loop calculation process is repeated until the change amount of the target optimization function value in two consecutive loops is less than the preset stop condition, such as the change amount being less than 0.001, and the updated signal estimation value at this time is output, for example, the signal intensity of the target region D at 1.2 hours is 100, which is adjusted to 92 after 3 iterations, and the target optimization function value is reduced from 5.8 to 1.2, satisfying the stop condition, and the signal estimation value of the position at the time point is determined to be 92.

[0110] Then, the updated signal estimation value is compared with the dynamic signal data by step S1044, and if the residual does not converge, iteration is repeated, and when the residual converges, the signal estimation value is Fourier transformed to reconstruct a dynamic enhancement image of the target tissue. For example, the residual of the signal estimation value of all spatial positions in the target region D and the dynamic signal data is calculated, the residual is the absolute value of the difference between the two, if the average residual of a certain position is 0.8, which is higher than the preset convergence threshold 0.5, then return to step S1043 to re-iterate; if the average residual of all positions is reduced to 0.3, which meets the convergence condition, then the signal estimation value of all spatial positions is sorted by time node, and the Fourier transform is performed on the signal estimation value of each time node to convert the frequency domain signal to spatial domain image signal, and then the dynamic enhancement image is generated after contrast adjustment, edge enhancement and other processing. In the image, only the region with signal meeting the characteristics of the target tissue is displayed, and the normal tissue region appears low brightness due to no specific signal, and the target tissue is clearly distinguished.

[0111] Finally, in the reconstruction process by step S1044, abnormal fluctuation points in the signal intensity curve that do not meet the time evolution constraint function are detected, the spatial position corresponding to the abnormal fluctuation point is marked as a artifact interference region, and the signal intensity of the artifact interference region in the dynamic enhancement image is set to zero. For example, in the dynamic enhancement image reconstruction process of the target region D, the signal intensity curve of each spatial position is monitored in real time: if the signal intensity of a certain spatial position suddenly rises from 80 to 150 in 1 hour, which deviates from the "peak value is reached after about 1 hour and then slowly decreases" mode defined by the time evolution constraint function, it is determined that the point is an abnormal fluctuation point; the spatial position corresponding to the point is marked as an artifact interference region; in the dynamic enhancement image, the signal intensity of the region is forced to zero to eliminate the interference of artifacts on the image, and ensure that only the signal region of the real target tissue is retained in the image.

[0112] In practical applications, first, the staff merges the specific first and second enhanced signals of the target region E liver region into a joint enhanced signal according to the spatial position by using the A brand medical image processing software, and ensures that the two types of signals of each voxel are correspondingly superimposed; then, the software starts to detect 0.4 hours, 0.8 hours, 1.2 hours and 1.6 hours based on the dynamic signal data of the target region E, constructs a time evolution constraint function to limit the signal to rise from 0.4 hours to 0.8 hours and slowly decline from 0.8 hours to 1.6 hours; then, the software constructs a target optimization function with the joint enhanced signal as the initial value, updates the signal estimate value by iterative solution, calculates the function value after each iteration until the function value changes less than 0.002 to stop, and iterates 5 times in total; after that, the software calculates the residual error between the signal estimate value and the dynamic signal data, and the average residual error is 0.28, which meets the convergence threshold 0.3, and performs Fourier transform on the signal estimate value to reconstruct the dynamic enhanced image; during the reconstruction process, the software detects that the signal of a certain region of the liver edge drops sharply at 1.0 hour, which does not meet the constraint function, marks it as an artifact region and sets the signal to zero; finally, the generated dynamic enhanced image clearly displays two target tissue regions in the liver without obvious artifacts.

[0113] In the overall scheme of the above step S104, by merging the two types of specific enhanced signals, the signal characteristics of the target tissue are concentrated, avoiding the limitations of a single signal type, providing a comprehensive signal basis for accurate imaging; by constructing the time evolution constraint function, the signal change range is limited according to the real response law of the detection device to exclude unreasonable signal interference and ensure the physiological reasonableness of the signal; by iterative optimization, the signal estimate value is constantly adjusted to not only fit the original dynamic signal data but also meet the evolution mode of the detection device, improving the signal accuracy; by residual error comparison and Fourier transform, the reliability of the reconstructed image is ensured, and only the target tissue is presented; by artifact detection and elimination, the image quality is further purified, and misdiagnosis caused by artifacts is avoided. Overall, through the complete process of signal merging, constraint, optimization, reconstruction and artifact removal, the dynamic enhanced image of the target tissue with high precision and high definition is finally generated.

[0114] The following is a complete embodiment for steps S101 to S104:

[0115] As Figure 3As shown, first, the detection device adopts the compressed sensing technology to accurately collect the dynamic signal data of the detection device on the first type of weighted image (T1 weighted image) and the second type of weighted image (T2 weighted image) at four time points of 0.2 hours, 0.4 hours, 0.6 hours and 0.8 hours when the detection device starts detection, and focuses on capturing the signal changes in the micro lesion area. Then, the multi-echo water-fat separation sequence of the magnetic resonance device is started to dynamically scan the pancreas and the surrounding area of the target region C, and the multi-echo data is synchronously acquired. According to the chemical shift difference between water and fat, the water signal of the pancreas and the fat signal of the surrounding fat tissue are accurately separated, and the clear initial water map and non-water phase map are generated by MRI image reconstruction. Then, time resolution analysis is carried out combined with the dynamic signal data, the initial water map and the non-water phase map: the initial water map is used to remove the background signals outside the pancreas such as the gastrointestinal gas area, the non-water phase map is used to filter out the surrounding fat signals, and the signal intensity curve of each spatial position is generated based on the processed data; the T1 weighted image signal of the micro lesion area in the pancreas is found to rise rapidly at 0.2-0.4 hours, and the T2 weighted image signal shows a delayed downward characteristic at 0.4-0.8 hours, while the normal pancreas tissue signal has no such regularity, thereby separating the specific T1 enhancement signal and the specific T2 enhancement signal. Finally, the two types of specific enhancement signals are combined with the time dimension information of the dynamic signal data, and the compressed sensing algorithm is used to reconstruct the MRI image, which clearly presents the active pancreatic cancer lesion with a diameter of about 3mm. At the same time, the time evolution mode of the detection device is used to identify and suppress the artifacts in the image caused by abdominal peristalsis, and to accurately distinguish the lesion from the surrounding normal pancreas tissue.

[0116] The real-time dynamic MRI tumor imaging method based on compressed sensing provided by the present application solves the problem of insufficient signal specificity of traditional MRI by using the detection device to target the microenvironment of the target tissue and combining compressed sensing technology to efficiently collect specific dynamic signals. The water-fat signal interference is eliminated by using the multi-echo water-fat separation sequence to generate high-quality basic images, which lays a foundation for subsequent signal analysis. The specific enhancement signal of the target tissue is accurately separated by time resolution analysis, which effectively distinguishes the target tissue from the normal tissue. Finally, the dynamic enhancement image representing the target tissue is reconstructed by using the compressed sensing algorithm, and the artifacts are suppressed, which greatly improves the accuracy and clarity of the target tissue imaging.

[0117] Figure 4 A specific implementation structure diagram of a real-time dynamic MRI tumor imaging system based on compressed sensing provided by an embodiment of the present application is shown in Figure 4 The system can include:

[0118] The acquisition module 41 adopts the compressed sensing technology to collect the dynamic signal data of the detection device on the first type of weighted image and the second type of weighted image caused by structural transformation at different time points;

[0119] a separation module 42 configured to perform a dynamic scan on a target region using a multi-echo water-fat separation sequence, to synchronously acquire multi-echo data during the dynamic scan, to separate the multi-echo data based on a chemical shift difference between water and fat to obtain water signals and fat signals, and to generate an initial water map based on the separated water signals and a non-water phase map based on the separated fat signals;

[0120] a generation module 43 configured to perform time-resolved analysis on the multi-echo data based on the dynamic signal data, the initial water map and the non-water phase map, to generate a signal intensity curve of the detection device over time, and to separate specific first and second type of enhanced signals according to enhancement characteristics of first and second type of weighted image signals at different time points in the signal intensity curve;

[0121] a reconstruction module 44 configured to perform MRI image reconstruction using a compressed sensing algorithm based on the specific first and second type of enhanced signals and in combination with time dimension information in the dynamic signal data to obtain a dynamic enhancement image of the target tissue, and to identify and suppress artifacts in the dynamic enhancement image using a specific evolution pattern of the detection device in the time dimension.

[0122] The real-time dynamic MRI tumor imaging system based on compressed sensing according to the embodiments of the present application is used to implement the real-time dynamic MRI tumor imaging method based on compressed sensing as described above, and thus the specific embodiments of the real-time dynamic MRI tumor imaging system based on compressed sensing can be seen from the foregoing embodiments of the real-time dynamic MRI tumor imaging method based on compressed sensing, and the specific embodiments can be described with reference to the descriptions of the respective embodiments, which will not be repeated here.

[0123] The present application also provides an electronic device, which comprises a memory configured to store a computer program and a processor configured to implement the steps of the real-time dynamic MRI tumor imaging method based on compressed sensing as described above when executing the computer program.

[0124] The present application also provides a computer readable storage medium having a computer program stored thereon, and the computer program is configured to implement the steps of the real-time dynamic MRI tumor imaging method based on compressed sensing as described above when executed by a processor.

[0125] In an exemplary embodiment, the computer readable storage medium as described above can include, but is not limited to, a U disk, a read-only memory, a random access memory, a mobile hard disk, a magnetic disk or an optical disk, and various media that can store computer programs.

[0126] The embodiment of the present application further provides a computer program product, the computer program product comprising a computer program, the computer program being executed by a processor to implement the steps in any of the above-mentioned embodiments of the real-time dynamic MRI tumor imaging method based on compressed sensing.

[0127] Those skilled in the art will further appreciate that the functions of the examples described herein-based units and algorithm steps can be implemented using electronic hardware, computer software, or any combination thereof. When the functions are implemented in software, the functions can be stored on or transmitted over a computer-readable medium, such as an optical, magnetic or semiconductor storage medium. The order of execution or the arrangement of code or steps can be changed, or individual code or steps can be combined or broken apart, or additional code or steps can be added, without departing from the scope of the application. Thus, the functions described herein-based examples can be implemented using any number of hardware and software configurations.

[0128] The above provides a kind of based on compressed sensing real-time dynamic MRI tumor imaging method and system provided in the application.The principle and implementation of the present application are described in this paper by applying specific examples, the above example is only used to help understand the method of the present application and its core idea.It should be pointed out that, for the ordinary skilled in the art, under the premise of not departing from the principle of the present application, the present application can be improved and modified, these improvements and modifications also fall within the scope of the present application.

Claims

1. A method for real-time dynamic MRI tumor imaging based on compressed sensing, characterized in that, The application relates to a method for reconstructing dynamic contrast-enhanced magnetic resonance imaging (MRI) images of a target tissue, comprising: acquiring, by using a compressed sensing technique, dynamic signal data of the first type of weighted images and the second type of weighted images of a detection device caused by structural transformation at different time points; performing dynamic scanning on a target region by using a multi-echo water-fat separation sequence, synchronously acquiring multi-echo data in the dynamic scanning process, separating the multi-echo data according to the chemical shift difference between water and fat to obtain water signals and fat signals, and generating an initial water image by MRI image reconstruction based on the separated water signals and a non-water phase image by MRI image reconstruction based on the separated fat signals; performing time resolution analysis on the multi-echo data based on the dynamic signal data, the initial water image and the non-water phase image, generating a signal intensity curve of the detection device changing with time, and separating specific first type of enhanced signals and second type of enhanced signals according to the enhancement characteristics of the first type of weighted image signals and the second type of weighted image signals at different time points in the signal intensity curve; reconstructing the MRI image by using a compressed sensing algorithm according to the specific first type of enhanced signals and the second type of enhanced signals and combining the time dimension information in the dynamic signal data, obtaining a dynamic contrast-enhanced image of the target tissue, and identifying and suppressing the artifacts in the dynamic contrast-enhanced image by using the specific evolution mode of the detection device in the time dimension. In the method, the time resolution analysis on the multi-echo data based on the dynamic signal data, the initial water image and the non-water phase image, the generation of the signal intensity curve of the detection device changing with time, and the separation of the specific first type of enhanced signals and the second type of enhanced signals according to the enhancement characteristics of the first type of weighted image signals and the second type of weighted image signals at different time points in the signal intensity curve, comprise the following steps: taking the initial water image as a spatial mask to remove the signal values of the background region in the dynamic signal data, marking the fat region in the non-water phase image to filter the signal values of the fat region in the dynamic signal data; constructing a two-dimensional matrix about the signal intensity of the time node and the spatial position based on the signal values obtained after the background region is removed and the fat region is filtered, wherein the rows of the two-dimensional matrix correspond to the spatial positions and the columns correspond to the time nodes; performing curve fitting on the data of the signal intensity changing with time in each row of the two-dimensional matrix to generate the first type of weighted image signal intensity curve and the second type of weighted image signal intensity curve of each spatial position; and marking the region meeting the preset threshold condition as the specific first type of enhanced signals according to the rising slope and the peak position of the first type of weighted image signal intensity curve, and marking the region meeting the preset time delay condition as the specific second type of enhanced signals according to the delay characteristics of the second type of weighted image signal intensity curve.

2. The method of claim 1, wherein, In the method, the reconstruction of the MRI image by using the compressed sensing algorithm according to the specific first type of enhanced signals and the second type of enhanced signals and combining the time dimension information in the dynamic signal data, the obtaining of the dynamic contrast-enhanced image of the target tissue, and the identification and suppression of the artifacts in the dynamic contrast-enhanced image by using the specific evolution mode of the detection device in the time dimension, comprise the following steps: merge the specific first type of enhanced signal and the specific second type of enhanced signal into a joint enhanced signal, a spatial position of the joint enhanced signal being consistent with the dynamic signal data; construct a time evolution constraint function based on time dimension information of the dynamic signal data, the time evolution constraint function describing a prior mode of signal intensity of the detection device changing over time; construct a target optimization function with the joint enhanced signal as an initial input and the time evolution constraint function as a sparsity constraint, update signal estimation values of each spatial position at consecutive time nodes by iteratively solving the target optimization function; compare the updated signal estimation values with the dynamic signal data for residual error, repeat iteration if the residual error does not converge, and when the residual error converges, perform Fourier transform on the signal estimation values to reconstruct a dynamic enhanced image of the target tissue; in the reconstruction process, detect abnormal fluctuation points in the signal intensity curve that do not conform to the time evolution constraint function, mark the spatial positions corresponding to the abnormal fluctuation points as artifact interference regions, and set the signal intensity of the artifact interference regions to zero in the dynamic enhanced image.

3. The method of claim 2, wherein, construct a target optimization function with the joint enhanced signal as an initial input and the time evolution constraint function as a sparsity constraint, update signal estimation values of each spatial position at consecutive time nodes by iteratively solving the target optimization function, including: assign the joint enhanced signal to the signal estimation values as initial values, and construct the target optimization function based on the time evolution constraint function as a sparsity evaluation basis; the target optimization function includes a data matching part between the signal estimation values and the dynamic signal data and a mode conforming part of the signal estimation values under the time evolution constraint function; update the signal estimation values through a loop calculation process, calculate a target optimization function value corresponding to the current signal estimation values in each loop, and adjust the signal estimation values based on the target optimization function value; repeat the loop calculation process until the change of the target optimization function value is lower than a preset stop condition, and output the updated signal estimation values.

4. The method of claim 1, wherein, perform curve fitting on the data of signal intensity changing over time represented by each row in the two-dimensional matrix to generate a first type of weighted image signal intensity curve and a second type of weighted image signal intensity curve of each spatial position, including: extract time sequence signal data corresponding to each spatial position in the two-dimensional matrix to establish a corresponding relationship between time nodes and signal intensity values, forming a discrete data point set; according to the physical characteristics of the detection device, divide the signal data into a first signal channel and a second signal channel, wherein the first signal channel corresponds to a fast response feature, and the second signal channel corresponds to a delayed response feature; for the first signal channel, a polynomial function approximation method is used to determine the fitting parameters by the least square method to obtain a continuous function expression of the first signal channel; for the second signal channel, an exponential decay function model is used for fitting, and the decay constant and initial amplitude parameters are solved by a nonlinear regression method to obtain a continuous function expression of the second signal channel; According to the reliability of signal intensity, different weights are assigned to the two continuous function expressions, and the continuous function expressions processed by the weight factor are smoothed to eliminate high-frequency noise, generating a first type of weighted image signal intensity curve and a second type of weighted image signal intensity curve.

5. The method of claim 1, wherein, The dynamic signal data of the detection device on the first type of weighted image and the second type of weighted image caused by structural transformation at different time points is collected by using the compressed sensing technology, including: At each time node, a first type of excitation pulse sequence is transmitted using the radio frequency coil of the magnetic resonance device to excite the first type of signal response of the detection device with the frequency parameter corresponding to the first type of weighted image, and the first type of frequency domain data is collected based on the random frequency sampling template; At the same time node, switch to the frequency parameter corresponding to the second type of weighted image, transmit a second type of excitation pulse sequence to excite the second type of signal response of the detection device, and collect the second type of frequency domain data based on the random frequency sampling template; The first type of frequency domain data and the second type of frequency domain data at each time node are integrated in time sequence to form dynamic signal data containing spatial coordinates, time nodes and signal type identification.

6. The method of claim 1, wherein, The target region is dynamically scanned by using a multi-echo water-fat separation sequence, and multi-echo data is synchronously acquired during the dynamic scanning process. The multi-echo data is separated according to the chemical shift difference between water and fat to obtain water signals and fat signals. An initial water map is generated based on the separated water signals through MRI image reconstruction, and a non-water phase map is generated based on the separated fat signals through MRI image reconstruction, including: The target region is continuously scanned by using a multi-echo gradient echo sequence, and frequency domain raw data is synchronously acquired at each echo time point. The frequency domain raw data at each echo time point is combined into a multi-channel multi-echo data block; Based on the preset chemical shift difference between water and fat, a phase difference evolution model containing phase accumulation functions at different echo time points is constructed, the multi-echo data block is input into the phase difference evolution model, and the water signals and fat signals of each voxel are separated by solving linear equations; The separated water signals are inverse Fourier transformed to reconstruct a two-dimensional initial water map, and the separated fat signals are inverse Fourier transformed to reconstruct a two-dimensional non-water phase map.

7. A real-time dynamic MRI tumor imaging system based on compressed sensing for performing the method of any one of claims 1-6. It includes: The acquisition module is used to collect the dynamic signal data of the detection device on the first type of weighted image and the second type of weighted image caused by structural transformation at different time points by using the compressed sensing technology; The separation module is used to dynamically scan the target region by using a multi-echo water-fat separation sequence, synchronously acquire multi-echo data during the dynamic scanning process, separate the multi-echo data according to the chemical shift difference between water and fat to obtain water signals and fat signals, and generate an initial water map based on the separated water signals through MRI image reconstruction, and generate a non-water phase map based on the separated fat signals through MRI image reconstruction; The generating module is configured to perform time resolution analysis on the multi-echo data based on the dynamic signal data, the initial water map and the non-water map, to generate a signal intensity curve of the detection device changing over time, and to separate specific first type enhancement signals and second type enhancement signals according to enhancement characteristics of first type weighted image signals and second type weighted image signals at different time points in the signal intensity curve. The reconstructing module is configured to perform MRI image reconstruction by using a compressed sensing algorithm according to the specific first type enhancement signals and the second type enhancement signals and in combination with time dimension information in the dynamic signal data, to obtain a dynamic enhancement image of the target tissue, and to identify and suppress interference of artifacts in the dynamic enhancement image by using a specific evolution mode of the detection device in the time dimension.

8. An electronic device, comprising: The computer program is stored in the computer readable storage medium and is executed by the processor to implement the steps of the real-time dynamic MRI tumor imaging method based on compressed sensing according to any one of claims 1 to 6. The computer program is stored in the computer readable storage medium and is executed by the processor to implement the steps of the real-time dynamic MRI tumor imaging method based on compressed sensing according to any one of claims 1 to 6. ​ 9. A computer-readable storage medium, characterized in that, ​

Images