Functional magnetic resonance imaging noise reduction method and device

Abstract

The invention discloses a functional magnetic resonance imaging noise reduction method and device, and relates to the technical field of magnetic resonance imaging.The method comprises the steps that a T2 weighted functional magnetic resonance image and a T2 * weighted functional magnetic resonance image of a brain are acquired through a plane echo sequence, the plane echo sequence comprises a first echo chain and a second echo chain, t2 weighted signals collected by the first echo chain correspond to T2 weighted functional magnetic resonance images, T2 * weighted signals collected by the second echo chain correspond to T2 * weighted functional magnetic resonance images, and the T2 weighted signals appear before the T2 * weighted signals; splitting the signal change of each pixel in the T2 * weighted functional magnetic resonance image into short echo time signal change and blood oxygen level dependent signal change; and removing the short echo time signal change as noise to obtain the blood oxygen level dependent signal change. According to the invention, accurate short echo time signals can be acquired, and efficient noise reduction can be realized.

Description

Technical Field

[0001] This invention relates to the field of magnetic resonance imaging technology, and specifically to a method and apparatus for noise reduction in functional magnetic resonance imaging. Background Technology

[0002] This section is intended to provide background or context for the embodiments of the invention set forth in the claims. The description herein is not an admission that it is prior art simply because it is included in this section.

[0003] Functional magnetic resonance imaging (fMRI) has revolutionized brain research and is one of the most important methods for mapping the human brain's functional structure. Echo planar imaging (EPI) sequences are the most commonly used fMRI techniques. For example, gradient echo (GE) planar echo sequences are primarily used in fMRI to detect changes in blood oxygen level dependent (BOLD) signals. BOLD signals are based on the difference in magnetic properties between oxyhemoglobin and deoxyhemoglobin in the blood. Changes in T2* values ​​cause alterations in the magnetic resonance signal, reflecting changes in blood flow in active brain regions and indirectly reflecting brain neural activity. Therefore, the quality of fMRI is crucial for ensuring the effective detection of BOLD signal changes, and fMRI noise reduction is key to achieving high fMRI quality. However, currently, an effective fMRI noise reduction solution is lacking. Summary of the Invention

[0004] In a first aspect, embodiments of the present invention provide a functional magnetic resonance imaging (fMRI) noise reduction method, capable of acquiring accurate short echo time signals and precisely eliminating variations in short echo time signals to achieve efficient noise reduction. The method includes:

[0005] T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were acquired using a planar echo sequence. The planar echo sequence includes a first echo chain and a second echo chain. The T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted fMRI image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted fMRI image. The T2-weighted signal appears before the T2*-weighted signal.

[0006] The signal of each pixel in the T2-weighted functional magnetic resonance image is used as the short echo time signal of the signal of each pixel in the T2*-weighted functional magnetic resonance image.

[0007] The signal changes of each pixel in the T2* weighted functional magnetic resonance imaging are decomposed into short echo time signal changes and blood oxygen level dependent signal changes;

[0008] Short echo time signal changes are removed as noise from the signal changes of pixels in T2*-weighted functional magnetic resonance images to obtain blood oxygen level-dependent signal changes.

[0009] Secondly, embodiments of the present invention provide a functional magnetic resonance imaging (fMRI) noise reduction device capable of acquiring accurate short echo time signals and precisely eliminating variations in short echo time signals to achieve efficient noise reduction. The device includes:

[0010] The image acquisition module is used to acquire T2-weighted functional magnetic resonance images and T2*-weighted functional magnetic resonance images of the brain using a planar echo sequence. The EPI sequence includes a first echo chain and a second echo chain, wherein the T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted functional magnetic resonance image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted functional magnetic resonance image. The T2-weighted signal appears before the T2*-weighted signal.

[0011] The signal decomposition module is used to treat the signal of each pixel in the T2-weighted functional magnetic resonance image as the short echo time signal of the signal of each pixel in the T2*-weighted functional magnetic resonance image; and to decompose the signal change of each pixel in the T2*-weighted functional magnetic resonance image into short echo time signal change and blood oxygen level dependent signal change.

[0012] The denoising module is used to remove short echo time signal changes as noise from the signal changes of pixels in T2* weighted functional magnetic resonance images to obtain blood oxygen level-dependent signal changes.

[0013] Thirdly, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above-described magnetic resonance imaging noise reduction method.

[0014] Fourthly, embodiments of the present invention also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described magnetic resonance imaging noise reduction method.

[0015] Fifthly, embodiments of the present invention provide a computer program product, the computer program product including a computer program, which, when executed by a processor, implements a functional magnetic resonance imaging noise reduction method.

[0016] In this embodiment of the invention, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain are acquired using a planar echo sequence. The planar echo sequence includes a first echo chain and a second echo chain. The T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted fMRI image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted fMRI image. The T2-weighted signal appears before the T2*-weighted signal. The signal of each pixel in the T2-weighted fMRI image is used as the short echo time signal of the signal of each pixel in the T2*-weighted fMRI image. The signal change of each pixel in the T2*-weighted fMRI image is decomposed into short echo time signal change and blood oxygen level-dependent signal change. The short echo time signal change is removed as noise from the signal change of the pixels in the T2*-weighted fMRI image to obtain the blood oxygen level-dependent signal change. In the above process, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were obtained by using a planar echo sequence containing two echo chains. The signal of each pixel in the T2-weighted fMRI image is used as the short echo time signal of the signal of each pixel in the T2*-weighted fMRI image. This is equivalent to acquiring accurate short echo time signals. Since the signal changes of the short echo time signals are caused by noise, these noise changes are extracted from the T2*-weighted fMRI image, thereby achieving a highly efficient noise reduction effect. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0018] Figure 1 This is a flowchart of the functional magnetic resonance imaging noise reduction method in an embodiment of the present invention;

[0019] Figure 2 This is a schematic diagram of the acquisition of two-dimensional excitation imaging of a planar echo sequence, which includes a spin signal echo chain and a gradient signal echo chain, in an embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram of the acquisition of a three-dimensional excitation imaging of a planar echo sequence including a spin signal echo chain and a gradient signal echo chain in an embodiment of the present invention.

[0021] Figure 4 This is a schematic diagram of the acquisition of a two-dimensional excitation imaging of a planar echo sequence, which includes a spin signal echo chain and multiple gradient signal echo chains, in an embodiment of the present invention.

[0022] Figure 5 Utilizing in the embodiments of the present invention Figure 4 One T2-weighted functional magnetic resonance imaging (fMRI) image from SE EPI and three T2*-weighted fMRI images from GE EPI acquired in the middle sequence;

[0023] Figure 6 This is a schematic diagram of the functional magnetic resonance imaging noise reduction device in an embodiment of the present invention;

[0024] Figure 7 This is a schematic diagram of a computer device in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the embodiments of the present invention will be further described in detail below with reference to the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention.

[0026] The acquisition, storage, use, and processing of data in this application all comply with the relevant provisions of national laws and regulations.

[0027] EPI (Epidermal Imaging) sequence is a magnetic resonance imaging (MRI) method. The inventors discovered that, taking the GE (Genomic Electron Imaging) EPI sequence as an example, the signal generates echoes through changes in the gradient field, belonging to the T2* weighted signal category. This method is highly sensitive to magnetic field inhomogeneities, including local magnetic field changes caused by oxyhemoglobin and deoxyhemoglobin. Therefore, the GE EPI sequence can sensitively detect changes in the BOLD (Brain Magnetic Resonance) signal caused by brain activity. A conventional GE EPI sequence uses a single echo train for signal acquisition. The functional magnetic resonance signal expression is as follows:

[0028] S(TE) = S0 × exp(-TE × R2) * )

[0029] The echo time TE refers to the time difference between the excitation pulse and the center of the K space corresponding to the collected signal; S(TE) is the magnitude of the functional magnetic resonance signal acquired at the echo time TE; S0 is the functional magnetic resonance signal at the moment when the echo time TE is equal to 0 or infinitely close to 0, which can be called the short echo time signal; R2* is the reciprocal of the time constant T2* of the functional magnetic resonance signal decay.

[0030] The relative change of the functional magnetic resonance signal S(TE) can be approximated as:

[0031] ΔS / S=ΔS0 / S0-TE×ΔR2 *

[0032] Where ΔR2 *This includes changes caused by brain tissue activation in functional magnetic resonance imaging (fMRI). In experiments on 3T MRI systems, a longer TE (transmission time) of 30ms is typically used to acquire relatively large signal changes ΔS / S for easier detection. S0, theoretically unaffected by brain tissue activation, is considered to be caused by noise. This noise includes instability in the MRI hardware and software, as well as signal changes caused by subject movement and physiological noise. Current techniques can fit S0, but the obtained S0 is inaccurate. Furthermore, current techniques cannot accurately acquire the S0 signal. If the S0 signal could be accurately acquired and its relative change ΔS0 / S0 calculated, this portion could be separated from the fMRI signal, thus achieving precise noise reduction. S0 can be considered as a short echo time functional magnetic resonance signal, thus forming a short echo time image. Therefore, the principle of functional magnetic resonance imaging noise reduction in this embodiment of the invention is based on the short echo time image. While acquiring the conventional long echo time EPI signal, a short echo time EPI functional magnetic resonance signal is also acquired to obtain its signal change. This signal change is considered to be caused by noise, and it is then separated from the functional magnetic resonance EPI signal, thereby achieving the noise reduction effect.

[0033] Figure 1 The flowchart of the functional magnetic resonance imaging noise reduction method in this embodiment of the invention includes:

[0034] Step 101: T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain are acquired using a planar echo sequence. The planar echo sequence includes a first echo chain and a second echo chain. The T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted fMRI image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted fMRI image. The T2-weighted signal appears before the T2*-weighted signal.

[0035] Step 102: The signal of each pixel in the T2-weighted functional magnetic resonance image is used as the short echo time signal of the signal of each pixel in the T2*-weighted functional magnetic resonance image;

[0036] Step 103: Decompose the signal change of each pixel in the T2* weighted functional magnetic resonance image into short echo time signal change and blood oxygen level dependent signal change;

[0037] Step 104: Remove the short echo time signal change as noise from the signal changes of pixels in the T2* weighted functional magnetic resonance image to obtain the blood oxygen level dependent signal change.

[0038] In this embodiment of the invention, a planar echo sequence containing two echo chains is used to obtain T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain. The signal of each pixel in the T2-weighted fMRI image is used as the short echo time signal of the signal of each pixel in the T2*-weighted fMRI image. This is equivalent to acquiring an accurate short echo time signal. The signal change of the short echo time signal is caused by noise, and this noise is removed from the T2*-weighted fMRI image, thereby achieving a highly efficient noise reduction effect.

[0039] In one embodiment, the first echo chain is a spin signal echo chain, and the second echo chain is a gradient signal echo chain.

[0040] The gradient signal echo train is applied after the spin signal echo train.

[0041] It should be noted that the planar echo sequence, i.e., the EPI sequence, used for functional magnetic resonance imaging noise reduction in this embodiment of the invention includes at least two echo chains, that is, it can contain more than two echo chains. The spin signal echo chain and the gradient signal echo chain appear sequentially in time. The one that appears first in time must be the spin signal echo chain.

[0042] The spin signal echo chain refers to the echo combination following an excitation pulse and a refocusing pulse, which is responsible for acquiring the refocused spin echo signal, i.e., the T2 weighted signal of SE EPI.

[0043] The gradient signal echo chain refers to the echo combination applied after the spin signal echo chain, which is responsible for acquiring the gradient echo signal, i.e., the T2* weighted signal of GE EPI.

[0044] The minimum number of gradient signal echo chains is one, and the number is not limited.

[0045] The T2* weighted signal is derived from the preceding T2 weighted signal. It can be considered as a T2* weighted signal with TE=0, meaning the T2 weighted signal of SE EPI is considered as S0 of the subsequent T2* weighted attenuated signal of GE EPI. Since the T2 weighted signal of SE EPI is insensitive to changes in the BOLD signal and uses the shortest possible echo time, S0 is theoretically unaffected or minimally affected by brain tissue activation. Its relative change ΔS0 / S0 is considered to be noise-induced.

[0046] Therefore, the echo time of the gradient signal echo chain of GE EPI can be redefined as the time difference between their echo time and that of the spin signal echo chain of SE EPI.

[0047] In one embodiment, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain are acquired using a planar echo sequence, including:

[0048] When the brain is in a functional task state or a resting state, a pre-set repetition time TR (repetition time) is continuously repeated using a planar echo sequence, so that the first echo chain acquires the T2-weighted signal and the second echo chain acquires the T2*-weighted signal.

[0049] Fourier transform image reconstruction was performed on the k-space data composed of T2-weighted signals to obtain T2-weighted functional magnetic resonance images;

[0050] Fourier transform image reconstruction was performed on the k-space data composed of T2* weighted signals to obtain T2* weighted functional magnetic resonance images.

[0051] Assuming there are M TRs, and the sequence contains one spin signal echo chain and N gradient signal echo chains, then the spatial location of each layer will generate M T2-weighted functional magnetic resonance images and M×N T2*-weighted functional magnetic resonance images. No specific limitations are placed on the duration of the task state or the number of TRs. The T2-weighted magnetic resonance images correspond to the T2-weighted signals and can both be represented by S0.

[0052] The M T2-weighted functional magnetic resonance images and the M×N T2*-weighted functional magnetic resonance images recorded the dynamic information of neural activity at each time point.

[0053] In one embodiment, the method further includes:

[0054] If the planar echo sequence uses two-dimensional excitation imaging, T2-weighted signals and T2*-weighted signals are acquired layer by layer, and Fourier transform image reconstruction is performed on the k-space data composed of T2-weighted signals and T2*-weighted signals layer by layer.

[0055] If a planar echo sequence is used for three-dimensional excitation imaging, by continuously modifying the phase encoding gradient in the layer direction, T2-weighted signals and T2*-weighted signals are acquired. After phase encoding is completed, three-dimensional Fourier transform image reconstruction is performed on the entire three-dimensional k-space data composed of T2-weighted signals and the entire three-dimensional k-space data composed of T2*-weighted signals.

[0056] In one embodiment, the spin signal echo train and the gradient signal echo train are acquired using either a bipolar gradient signal echo train or a unipolar gradient signal echo train.

[0057] When the spin signal echo chain and gradient signal echo chain adopt the bipolar gradient signal echo chain method, the acquired T2-weighted signal and T2*-weighted signal are phase-corrected.

[0058] The unipolar gradient signal echo train method refers to acquiring magnetic resonance signals using only positive or negative gradients, while the bipolar gradient signal echo train method uses both positive and negative gradients to acquire magnetic resonance signals, which can increase the acquisition efficiency of magnetic resonance signals. However, phase correction is required for the T2-weighted signals and T2*-weighted signals acquired by the positive and negative gradients for subsequent steps.

[0059] In one embodiment, the length of the echo train and the number of echoes it contains are not specifically limited and are generally determined by the image spatial resolution. When full data acquisition is achieved, the size of the image matrix in the phase encoding direction of the T2-weighted functional magnetic resonance image and the T2*-weighted functional magnetic resonance image is consistent with the length of the echo train. However, partial Fourier transform techniques, multiple excitation techniques, small field-of-view imaging techniques, and parallel imaging techniques can shorten the echo train length as needed, thereby shortening the echo time interval. Partial Fourier transform techniques utilize the conjugate symmetry of the k-space data in magnetic resonance imaging to acquire only a portion of the k-space data; multiple excitation techniques acquire a portion of the k-space data each time using a short echo train, and the data acquired multiple times are merged into the entire k-space; small field-of-view imaging techniques and parallel imaging techniques both use special excitation pulses to excite only a small image region, requiring only a smaller amount of k-space data to be acquired at the same spatial resolution, thus also shortening the echo train length.

[0060] In one embodiment, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain are acquired using a planar echo sequence, including:

[0061] Based on multi-slice scanning technology, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were acquired using planar echo sequence acquisition.

[0062] Simultaneous multislice imaging techniques (SMS) can reduce the number of triggers while keeping the number of slices constant, which can significantly reduce the time required to acquire whole-brain data and solve the problem of extended time in multi-echo chain techniques.

[0063] In one embodiment, the signal variation of each pixel in a T2*-weighted functional magnetic resonance image is decomposed into short-echo time signal variation and blood oxygen level-dependent signal variation, including:

[0064] Principal component analysis and independent component analysis were performed on the signal of each pixel in the T2* weighted functional magnetic resonance image to obtain multiple components;

[0065] Each component was fitted and compared according to the short echo time signal change formula and the blood oxygen level dependent signal change formula;

[0066] Based on the fitting comparison results, determine whether the signal change component of each pixel in the T2* weighted functional magnetic resonance image is a short echo time signal change or a blood oxygen level dependent signal change.

[0067] Denoising T2*-weighted functional magnetic resonance imaging (fMRI) images involves treating the relative change ΔS0 / S0 of the T2-weighted signal in the SE EPI as noise. Then, mathematical algorithms are used to separate ΔS0 / S0 from the relative change in the subsequent GE EPI T2*-weighted signal, thus achieving noise reduction in the fMRI signal. Therefore, the primary step in noise reduction is to decompose the signal change of each pixel in the T2*-weighted fMRI image into short echo time (TE) signal changes and blood oxygen level-dependent (BOLD) signal changes. This includes performing principal component analysis (PCA) and independent component analysis (ICA) on the signal of each pixel in the T2*-weighted fMRI image. After obtaining multiple components, they are fitted and compared according to the formulas for TE and BOLD signal changes. The signal fluctuation of the noise component does not change with increasing TE, while the change in BOLD signal changes linearly with TE. This allows for the distinction between noise and BOLD signals. ΔS0 / S0 equal to 0 indicates no noise, while ΔS0 / S0 not equal to 0 indicates noise, requiring noise reduction.

[0068] The signal variation of each pixel in a T2*-weighted functional magnetic resonance imaging (fMRI) image can be decomposed into short-echo time signal variation and blood oxygen level-dependent signal variation. In one embodiment, the formula for short-echo time signal variation is:

[0069] ΔS / S=ΔS0 / S0

[0070] Where ΔS / S represents the signal variation of each pixel in the T2* weighted functional magnetic resonance image, ΔS0 / S0 represents the short echo time signal variation, and S0 represents the short echo time signal.

[0071] The formula for blood oxygen level dependent on signal changes is:

[0072] ΔS / S=-TE×ΔR2 *

[0073] Where TE is the echo time, and R2 is the echo duration. * ΔR2 is the reciprocal of the time constant T2* of the magnetic resonance signal decay. * R2 * The changes.

[0074] In one embodiment, removing short echo time signal variations as noise from the signal variations of pixels in a T2*-weighted functional magnetic resonance image includes:

[0075] The fluctuations in the amplitude of pixels corresponding to the short echo time signal changes on the time axis are used to perform signal regression to remove the short echo time signal changes.

[0076] Signal regression can remove instability of magnetic resonance imaging hardware and software, as well as signal changes caused by subject movement and physiological noise, from T2*-weighted images, finally obtaining the BOLD signal.

[0077] In one embodiment, the method further includes:

[0078] When there are multiple T2* weighted functional magnetic resonance images, assign a weighting factor to each obtained oxygenation level-dependent signal.

[0079] An optimized blood oxygen level dependent signal is obtained by mixing each blood oxygen level dependent signal and its corresponding weighting factor.

[0080] If multiple T2*-weighted images from GE EPI are available, they can be blended to optimize fMRI image quality and BOLD contrast. Assigning specific weighting factors to images acquired at different echo times by GE EPI depends on a variety of factors, including the parameters of the GEEPI sequence and the brain regions where neural activity is being studied.

[0081] In one embodiment, the weighting factor for the blood oxygen level-dependent signal is obtained using the following formula or determined based on the region of interest in the brain:

[0082] Q = TE × exp(-TE × R2) * )

[0083] Where Q is the weighting factor, TE is the echo time, and R2 is the weighting factor. * It is the reciprocal of the time constant T2* for the decay of the magnetic resonance signal;

[0084] R2 * It is obtained by fitting the following formula using the least squares method:

[0085] S(TE) = S0 × exp(-TE × R2) * )

[0086] S(TE) is the magnitude of the T2* weighted signal, and S0 is the short echo time signal.

[0087] The above presents two schemes for obtaining the weighting factor of the BOLD signal.

[0088] The first method, calculated according to the formula, helps improve R² because this method can directly acquire S0 compared to traditional multi-echo GE EPI (Multi-echo GEEPI) sequences. * The accuracy of the fitting results is high, therefore the obtained Q is very accurate.

[0089] The second approach involves identifying regions of interest in the brain. For example, regions prone to artifacts (e.g., the prefrontal and temporal lobes) may benefit from higher weighting in images acquired with shorter echo times. The process of blending images can be performed using a variety of image fusion techniques, ranging from simple linear combinations to more complex image processing algorithms, designed to extract and combine the most relevant features from each image type to improve the signal-to-noise ratio.

[0090] In addition to the methods mentioned above for obtaining short echo time signals, two more schemes are given below.

[0091] In one embodiment, the method further includes:

[0092] When the brain is in a functional task state or resting state, a planar echo sequence is applied, and multiple k spaces are continuously filled in a spiral manner after an excitation pulse, wherein the distance from the excitation pulse to the k space corresponding to the first T2* weighted functional magnetic resonance image is infinitely close to zero or equal to zero.

[0093] Fourier transform image reconstruction is performed on the continuously filled k-space to obtain multiple T2* weighted functional magnetic resonance images, including long echo time T2* weighted functional magnetic resonance images and short echo time T2* weighted functional magnetic resonance images.

[0094] The short echo time signal was obtained from the first short echo time T2* weighted functional magnetic resonance image.

[0095] Among them, the short echo time T2* weighted functional magnetic resonance imaging is used to detect changes in the BOLD signal. The K space is filled by spiral out, and the excitation pulse design is optimized so that the distance from the excitation pulse to the K space corresponding to the first T2* weighted functional magnetic resonance imaging is infinitely close to zero or equal to zero, usually less than 3ms, without restriction, so as to achieve the effect that the echo time of the first T2* weighted functional magnetic resonance imaging is UTE or ZTE (Ultrashort echo time or Zero echo time).

[0096] In one embodiment, the method further includes:

[0097] When the brain is in a functional task state or a resting state, a planar echo sequence is applied. After an excitation pulse, multiple K-space lines are continuously filled using a radial planar echo pattern. Multiple T2*-weighted functional magnetic resonance images are obtained using a K-space weighted image contrast (KWIC) reconstruction method. These multiple T2*-weighted functional magnetic resonance images include T2*-weighted functional magnetic resonance images with long echo times and T2*-weighted functional magnetic resonance images with short echo times. KWIC selectively filters K-space lines with different echo times to manipulate the echo time and contrast of the T2*-weighted functional magnetic resonance images, so that the distance from the excitation pulse to the K-space corresponding to the first T2*-weighted functional magnetic resonance image is infinitely close to zero or equal to zero.

[0098] The short echo time signal was obtained from the first short echo time T2* weighted functional magnetic resonance image.

[0099] Similarly, the distance from the excitation pulse to the K-space corresponding to the first T2* weighted functional magnetic resonance image is infinitely close to zero or equal to zero, usually less than 3ms, and is not limited, so as to achieve the effect that the echo time of the first T2* weighted functional magnetic resonance image is UTE or ZTE (Ultrashort echo time or Zero echo time).

[0100] Three specific implementation examples are given below.

[0101] Example 1:

[0102] Figure 2 This is a schematic diagram of the acquisition of a two-dimensional excitation imaging of a planar echo sequence, which includes a spin signal echo chain and a gradient signal echo chain, in an embodiment of the present invention.

[0103] In a TR (Transmission Response) system, a spin signal echo train and a gradient signal echo train appear sequentially. This embodiment does not limit the number of each type of echo train.

[0104] Both the spin signal echo chain and the gradient signal echo chain use a bipolar gradient signal echo chain to acquire the T2-weighted signal and the T2*-weighted signal at each echo moment.

[0105] The first 90-degree RF pulse flips the magnetization vector to the transverse plane, while the second 180-degree RF pulse refocuses the spins that have lost phase due to magnetic field inhomogeneity. After a period of time (TE1, which is twice the time difference between the 90-degree and 180-degree RF pulses) of signal attenuation (T2), a spin echo is generated at the spin signal echo chain.

[0106] Then, using the form of gradient echo, the first spin echo signal undergoes a T2* signal attenuation of duration (TE2-TE1), generating a gradient echo at the gradient signal echo chain.

[0107] In this embodiment, TE2 is greater than TE1, and TE1 needs to be as short as possible. There is no restriction on the size of TE2, which can be modified according to experimental requirements.

[0108] The k-space data composed of T2-weighted signals and T2*-weighted signals in the spin signal echo train and gradient signal echo train are obtained, and Fourier transform image reconstruction is performed on them respectively to generate T2-weighted functional magnetic resonance images and T2*-weighted functional magnetic resonance images.

[0109] The amplitude of the signal of each pixel in the T2-weighted functional magnetic resonance imaging (fMRI) image can be used as S0 to perform noise reduction processing on subsequent T2*-weighted fMRI images. The noise reduction method can be adjusted according to experimental requirements; this example does not impose specific limitations.

[0110] Example 2:

[0111] Figure 3 This is a schematic diagram of the acquisition of a three-dimensional excitation imaging of a planar echo sequence comprising a spin signal echo train and a gradient signal echo train, as described in an embodiment of the present invention. Compared to... Figure 2 Instead of using two-dimensional excitation imaging, Example 2 employs three-dimensional excitation imaging, simultaneously exciting the entire brain and adding a phase-encoding gradient in the Gz direction for encoding the layer direction. The k-space data, composed of T2-weighted and T2*-weighted signals acquired from the spin signal echo train and gradient signal echo train, are subjected to three-dimensional Fourier transform image reconstruction, thereby generating three-dimensional T2-weighted functional magnetic resonance imaging (fMRI) images and three-dimensional T2*-weighted fMRI images. The T2-weighted signal of SE EPI can be considered as S0 of the subsequent T2*-weighted signal of GE EPI, and noise reduction processing is performed on the T2*-weighted fMRI images generated by GE EPI.

[0112] Example 3:

[0113] Figure 4 This is a schematic diagram illustrating the acquisition of a two-dimensional excitation imaging system for a planar echo sequence comprising a spin signal echo train and multiple gradient signal echo trains, as described in an embodiment of the present invention. Compared to... Figure 2In Example 3, multiple gradient signal echo chains are applied after the spin signal echo chain in the EPI sequence. T2* weighted signals are acquired at each of these applied gradient signal echo chains using gradient echo methods. Two-dimensional Fourier transform image reconstruction is performed on the k-space data composed of the T2-weighted and T2*-weighted signals acquired from one spin signal echo chain and multiple gradient signal echo chains, respectively, to generate one T2-weighted magnetic resonance image and multiple T2*-weighted magnetic resonance images. The amplitude of the signal of each pixel in the T2-weighted functional magnetic resonance image can be used as S0 to denoise the multiple T2*-weighted functional magnetic resonance images. After denoising, the multiple T2*-weighted functional magnetic resonance images can be used individually or in combination in a specific manner.

[0114] Figure 5 Utilizing in the embodiments of the present invention Figure 4 The dataset consists of one SE EPI T2-weighted functional magnetic resonance imaging (fMRI) image and three GE EPI T2*-weighted fMRI images acquired in the mid-sequence. In this embodiment, the mixing method involves averaging the SE EPI T2-weighted fMRI image and the GE EPI T2*-weighted fMRI image. This embodiment does not limit the combination method. The imaging parameters are: TR / TE1 / TE2 / TE3 / TE4 = 3s / 20 / 35 / 50 / 65ms. Using the SE EPI T2-weighted fMRI image as the S0 of the three GE EPI T2*-weighted fMRI images, their TE values ​​are 0 / 15 / 30 / 45ms respectively, and the imaging field of view (FOV) is 224*224mm. 2 Spatial resolution = 3.5 × 3.5 × 3.5 mm 3 The number of slices is 37, the parallel imaging acceleration factor is 2, the spin signal echo train length is 32, and the gradient signal echo train length is 32. This embodiment does not limit the imaging parameters.

[0115] This invention also proposes a functional magnetic resonance imaging (fMRI) noise reduction device, the principle of which is similar to the fMRI noise reduction method, and will not be described in detail here.

[0116] Figure 6 This is a schematic diagram of the functional magnetic resonance imaging noise reduction device in an embodiment of the present invention, comprising:

[0117] Image acquisition module 601 is used to acquire T2-weighted functional magnetic resonance images and T2*-weighted functional magnetic resonance images of the brain using planar echo sequences. The EPI sequence includes a first echo chain and a second echo chain, wherein the T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted functional magnetic resonance image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted functional magnetic resonance image. The T2-weighted signal appears before the T2*-weighted signal.

[0118] The signal splitting module 602 is used to treat the signal of each pixel in the T2-weighted functional magnetic resonance image as the short echo time signal of the signal of each pixel in the T2*-weighted functional magnetic resonance image; and to split the signal change of each pixel in the T2*-weighted functional magnetic resonance image into short echo time signal change and blood oxygen level dependent signal change.

[0119] The denoising module 603 is used to remove short echo time signal changes as noise from the signal changes of pixels in T2* weighted functional magnetic resonance images to obtain blood oxygen level dependent signal changes.

[0120] In one embodiment, the image acquisition module is specifically used for:

[0121] When the brain is in a functional task state or a resting state, the planar echo sequence is repeatedly repeated for a preset duration, so that the first echo chain collects T2-weighted signals and the second echo chain collects T2*-weighted signals.

[0122] Fourier transform image reconstruction was performed on the k-space data composed of T2-weighted signals to obtain T2-weighted functional magnetic resonance images;

[0123] Fourier transform image reconstruction was performed on the k-space data composed of T2* weighted signals to obtain T2* weighted functional magnetic resonance images.

[0124] In one embodiment, the image acquisition module is specifically used for:

[0125] If the planar echo sequence uses two-dimensional excitation imaging, T2-weighted signals and T2*-weighted signals are acquired layer by layer, and Fourier transform image reconstruction is performed on the k-space data composed of T2-weighted signals and T2*-weighted signals layer by layer.

[0126] If a planar echo sequence is used for three-dimensional excitation imaging, by continuously modifying the phase encoding gradient in the layer direction, T2-weighted signals and T2*-weighted signals are acquired. After phase encoding is completed, three-dimensional Fourier transform image reconstruction is performed on the entire three-dimensional k-space data composed of T2-weighted signals and the entire three-dimensional k-space data composed of T2*-weighted signals.

[0127] In one embodiment, the first echo chain is a spin signal echo chain, and the second echo chain is a gradient signal echo chain.

[0128] The gradient signal echo train is applied after the spin signal echo train.

[0129] In one embodiment, the spin signal echo train and the gradient signal echo train are acquired using either a bipolar gradient signal echo train or a unipolar gradient signal echo train.

[0130] When the spin signal echo chain and gradient signal echo chain adopt the bipolar gradient signal echo chain method, the acquired T2-weighted signal and T2*-weighted signal are phase-corrected.

[0131] In one embodiment, the image acquisition module is specifically used for:

[0132] Based on multi-slice scanning technology, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were acquired using planar echo sequence acquisition.

[0133] In one embodiment, the signal splitting module is specifically used for:

[0134] Principal component analysis and independent component analysis were performed on the signal of each pixel in the T2* weighted functional magnetic resonance image to obtain multiple components;

[0135] Each component was fitted and compared according to the short echo time signal change formula and the blood oxygen level dependent signal change formula;

[0136] Based on the fitting comparison results, determine whether the signal component of each pixel in the T2* weighted functional magnetic resonance image is a short echo time signal change or a blood oxygen level dependent signal change.

[0137] The signal variation of each pixel in a T2*-weighted functional magnetic resonance imaging (fMRI) image can be decomposed into short-echo time signal variation and blood oxygen level-dependent signal variation. In one embodiment, the formula for short-echo time signal variation is:

[0138] ΔS / S=ΔS0 / S0

[0139] Where ΔS / S represents the signal variation of each pixel in the T2* weighted functional magnetic resonance image, ΔS0 / S0 represents the short echo time signal variation, and S0 represents the short echo time signal.

[0140] The formula for blood oxygen level dependent on signal changes is:

[0141] ΔS / S=-TE×ΔR2 *

[0142] Where TE is the echo time, and R2 is the echo duration. *ΔR2 is the reciprocal of the time constant T2* of the magnetic resonance signal decay. * R2 * The changes.

[0143] In one embodiment, the noise reduction module is specifically used for:

[0144] The fluctuations in the amplitude of pixels corresponding to the short echo time signal changes on the time axis are used to perform signal regression to remove the short echo time signal changes.

[0145] In one embodiment, the noise reduction module is specifically used for:

[0146] When there are multiple T2* weighted functional magnetic resonance images, assign a weighting factor to each obtained oxygenation level-dependent signal.

[0147] An optimized blood oxygen level dependent signal is obtained by mixing each blood oxygen level dependent signal and its corresponding weighting factor.

[0148] In one embodiment, the weighting factor for the blood oxygen level-dependent signal is obtained using the following formula or determined based on the region of interest in the brain:

[0149] Q = TE × exp(-TE × R2) * )

[0150] Where Q is the weighting factor, TE is the echo time, and R2 is the weighting factor. * It is the reciprocal of the time constant T2* for the decay of the magnetic resonance signal;

[0151] R2 * It is obtained by fitting the following formula using the least squares method:

[0152] S(TE) = S0 × exp(-TE × R2) * )

[0153] S(TE) is the magnitude of the T2* weighted signal, and S0 is the short echo time signal.

[0154] In one embodiment, the image acquisition module is further configured to:

[0155] When the brain is in a functional task state or resting state, a planar echo sequence is applied, and multiple k spaces are continuously filled in a spiral manner after an excitation pulse, wherein the distance from the excitation pulse to the k space corresponding to the first T2* weighted functional magnetic resonance image is infinitely close to zero or equal to zero.

[0156] Fourier transform image reconstruction is performed on the continuously filled k-space to obtain multiple T2* weighted functional magnetic resonance images, including long echo time T2* weighted functional magnetic resonance images and short echo time T2* weighted functional magnetic resonance images.

[0157] The short echo time signal was obtained from the first short echo time T2* weighted functional magnetic resonance image.

[0158] In one embodiment, the image acquisition module is further configured to:

[0159] When the brain is in a functional task state or a resting state, a planar echo sequence is applied. After an excitation pulse, multiple K-space lines are continuously filled using a radial planar echo pattern. Multiple T2*-weighted functional magnetic resonance images are obtained using a K-space-weighted image contrast reconstruction method. These multiple T2*-weighted functional magnetic resonance images include T2*-weighted functional magnetic resonance images with long echo times and T2*-weighted functional magnetic resonance images with short echo times. KWIC selectively filters K-space lines with different echo times to manipulate the echo time and contrast of the T2*-weighted functional magnetic resonance images, so that the distance from the excitation pulse to the K-space corresponding to the first T2*-weighted functional magnetic resonance image is infinitely close to zero or equal to zero.

[0160] The short echo time signal was obtained from the first short echo time T2* weighted functional magnetic resonance image.

[0161] In summary, the method and apparatus proposed in this invention utilize planar echo sequences to acquire T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain. The planar echo sequences include a first echo chain and a second echo chain. The T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted fMRI image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted fMRI image. The T2-weighted signal appears before the T2*-weighted signal. The signal of each pixel in the T2-weighted fMRI image is used as the short echo time signal of each pixel in the T2*-weighted fMRI image. The signal change of each pixel in the T2*-weighted fMRI image is decomposed into short echo time signal change and blood oxygen level-dependent signal change. The short echo time signal change is removed as noise from the signal change of pixels in the T2*-weighted fMRI image to obtain the blood oxygen level-dependent signal change. In the above process, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were obtained by using a planar echo sequence containing two echo chains. The signal of each pixel in the T2-weighted fMRI image is used as the short echo time signal of the signal of each pixel in the T2*-weighted fMRI image. This is equivalent to acquiring accurate short echo time signals. Since the signal changes of the short echo time signals are caused by noise, these noise changes are extracted from the T2*-weighted fMRI image, thereby achieving a highly efficient noise reduction effect.

[0162] This invention addresses some limitations of current BOLD imaging, such as poor motion sensitivity and spatial accuracy, providing a new imaging approach for the study of brain function and its diseases. Besides functional magnetic resonance imaging, this imaging method can also be applied to other fields such as diffusion imaging, perfusion imaging, and quantitative relaxation.

[0163] This invention also provides a computer device. Figure 7 This is a schematic diagram of a computer device in an embodiment of the present invention. The computer device 700 includes a memory 710, a processor 720, and a computer program 730 stored in the memory 710 and executable on the processor 720. When the processor 720 executes the computer program 730, it implements the above-mentioned functional magnetic resonance imaging noise reduction method.

[0164] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described magnetic resonance imaging noise reduction method.

[0165] This invention also provides a computer program product, which includes a computer program that, when executed by a processor, implements the above-described magnetic resonance imaging noise reduction method.

[0166] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0167] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0168] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0169] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0170] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for noise reduction in functional magnetic resonance imaging, characterized in that, include: T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were acquired using a planar echo sequence. The planar echo sequence includes a first echo chain and a second echo chain. The T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted fMRI image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted fMRI image. The T2-weighted signal appears before the T2*-weighted signal. The signal of each pixel in the T2-weighted functional magnetic resonance image is used as the short echo time signal of the signal of each pixel in the T2*-weighted functional magnetic resonance image. The signal changes of each pixel in the T2* weighted functional magnetic resonance imaging are decomposed into short echo time signal changes and blood oxygen level dependent signal changes; Short echo time signal changes are removed as noise from the signal changes of pixels in T2*-weighted functional magnetic resonance imaging to obtain blood oxygen level-dependent signal changes.

2. The method as described in claim 1, characterized in that, T2-weighted and T2*-weighted functional magnetic resonance imaging (fMRI) images of the brain were acquired using planar echo sequences, including: When the brain is in a functional task state or a resting state, a planar echo sequence is used so that the first echo chain acquires the T2-weighted signal and the second echo chain acquires the T2*-weighted signal. Fourier transform image reconstruction was performed on the k-space data composed of T2-weighted signals to obtain T2-weighted functional magnetic resonance images; Fourier transform image reconstruction was performed on the k-space data composed of T2* weighted signals to obtain T2* weighted functional magnetic resonance images.

3. The method as described in claim 2, characterized in that, Also includes: If the planar echo sequence uses two-dimensional excitation imaging, T2-weighted signals and T2*-weighted signals are acquired layer by layer, and Fourier transform image reconstruction is performed on the k-space data composed of T2-weighted signals and T2*-weighted signals layer by layer. If a planar echo sequence is used for three-dimensional excitation imaging, by continuously modifying the phase encoding gradient in the layer direction, T2-weighted signals and T2*-weighted signals are acquired. After phase encoding is completed, three-dimensional Fourier transform image reconstruction is performed on the entire three-dimensional k-space data composed of T2-weighted signals and the entire three-dimensional k-space data composed of T2*-weighted signals.

4. The method as described in claim 1, characterized in that, The first type of echo train is a spin signal echo train, and the second type of echo train is a gradient signal echo train. The gradient signal echo train is applied after the spin signal echo train.

5. The method as described in claim 4, characterized in that, Spin signal echo trains and gradient signal echo trains are acquired using either bipolar gradient signal echo trains or unipolar gradient signal echo trains. When the spin signal echo chain and gradient signal echo chain adopt the bipolar gradient signal echo chain method, the acquired T2-weighted signal and T2*-weighted signal are phase-corrected.

6. The method as described in claim 1, characterized in that, T2-weighted and T2*-weighted functional magnetic resonance imaging (fMRI) images of the brain were acquired using planar echo sequences, including: Based on multi-slice scanning technology, T2-weighted functional magnetic resonance imaging (fMRI) images and T2*-weighted fMRI images of the brain were acquired using planar echo sequence acquisition.

7. The method as described in claim 1, characterized in that, The signal variation of each pixel in a T2*-weighted functional magnetic resonance imaging (fMRI) image is decomposed into short-echo time-dependent signal variation and blood oxygenation level-dependent signal variation, including: Principal component analysis and independent component analysis were performed on the signal of each pixel in the T2* weighted functional magnetic resonance image to obtain multiple components; Each component was fitted and compared according to the short echo time signal change formula and the blood oxygen level dependent signal change formula; Based on the fitting comparison results, determine whether the signal component of each pixel in the T2* weighted functional magnetic resonance image is a short echo time signal change or a blood oxygen level dependent signal change.

8. The method as described in claim 7, characterized in that, The formula for the change of short echo time signal is: ΔS / S=ΔS0 / S0 Where ΔS / S represents the signal variation of each pixel in the T2* weighted functional magnetic resonance image, ΔS0 / S0 represents the short echo time signal variation, and S0 represents the short echo time signal. The formula for blood oxygen level dependent on signal changes is: ΔS / S=-TE×ΔR2 * Where TE is the echo time, and R2 is the echo duration. * ΔR2 is the reciprocal of the time constant T2* of the magnetic resonance signal decay. * R2 * The changes.

9. The method as described in claim 1, characterized in that, Short echo time signal variations are removed as noise from the pixel signal variations in T2*-weighted functional magnetic resonance imaging (fMRI) images, including: The fluctuations in the amplitude of pixels corresponding to short echo time signal changes on the time axis are used to perform signal regression to remove the short echo time signal changes.

10. The method as described in claim 1, characterized in that, Also includes: When there are multiple T2* weighted functional magnetic resonance images, assign a weighting factor to each obtained oxygenation level-dependent signal. An optimized blood oxygen level dependent signal is obtained by mixing each blood oxygen level dependent signal and its corresponding weighting factor.

11. The method as described in claim 10, characterized in that, The weighting factor for blood oxygenation level-dependent signals is obtained using the following formula or determined based on the region of interest in the brain: Q=TE×exp(-TE×R2 * ) Where Q is the weighting factor, TE is the echo time, and R2 is the weighting factor. * It is the reciprocal of the time constant T2* for the decay of the magnetic resonance signal; R2 * It is obtained by fitting the following formula using the least squares method: S(TE)=S0×exp(-TE×R2 * ) S(TE) is the magnitude of the T2* weighted signal, and S0 is the short echo time signal.

12. The method as described in claim 1, characterized in that, Also includes: When the brain is in a functional task state or resting state, a planar echo sequence is applied, and multiple k spaces are continuously filled in a spiral manner after an excitation pulse, wherein the distance from the excitation pulse to the k space corresponding to the first T2* weighted functional magnetic resonance image is infinitely close to zero or equal to zero. Fourier transform image reconstruction is performed on the continuously filled k-space to obtain multiple T2* weighted functional magnetic resonance images, including long echo time T2* weighted functional magnetic resonance images and short echo time T2* weighted functional magnetic resonance images. The short echo time signal was obtained from the first short echo time T2* weighted functional magnetic resonance image.

13. The method as described in claim 1, characterized in that, Also includes: When the brain is in a functional task state or a resting state, a planar echo sequence is applied. After an excitation pulse, multiple K-space lines are continuously filled using a radial planar echo pattern. Multiple T2*-weighted functional magnetic resonance images are obtained using a K-space-weighted image contrast reconstruction method. These multiple T2*-weighted functional magnetic resonance images include T2*-weighted functional magnetic resonance images with long echo times and T2*-weighted functional magnetic resonance images with short echo times. KWIC selectively filters K-space lines with different echo times to manipulate the echo time and contrast of the T2*-weighted functional magnetic resonance images, so that the distance from the excitation pulse to the K-space corresponding to the first T2*-weighted functional magnetic resonance image is infinitely close to zero or equal to zero. The short echo time signal was obtained from the first short echo time T2* weighted functional magnetic resonance image.

14. A functional magnetic resonance imaging noise reduction device, characterized in that, include: The image acquisition module is used to acquire T2-weighted functional magnetic resonance images and T2*-weighted functional magnetic resonance images of the brain using a planar echo sequence. The EPI sequence includes a first echo chain and a second echo chain, wherein the T2-weighted signal acquired by the first echo chain corresponds to the T2-weighted functional magnetic resonance image, and the T2*-weighted signal acquired by the second echo chain corresponds to the T2*-weighted functional magnetic resonance image. The T2-weighted signal appears before the T2*-weighted signal. The signal decomposition module is used to treat the signal of each pixel in the T2-weighted functional magnetic resonance image as the short echo time signal of the signal of each pixel in the T2*-weighted functional magnetic resonance image; and to decompose the signal change of each pixel in the T2*-weighted functional magnetic resonance image into short echo time signal change and blood oxygen level dependent signal change. The denoising module is used to remove short echo time signal changes as noise from the signal changes of pixels in T2* weighted functional magnetic resonance images to obtain blood oxygen level-dependent signal changes.

15. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 13.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method according to any one of claims 1 to 13.

17. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 13.

Images