A method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images

Abstract

The present invention discloses a method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images, which includes collecting spinal cord resting-state functional images and structural images of the C2-C7 segments; synchronously recording respiratory and heartbeat physiological signals; registering the structural image with a standard template to obtain field transformation map I; preprocessing the functional image, registering the structural image and the functional image with each other to obtain field transformation map II; extracting the standard template of the gray matter horns of C2-C7, and combining field transformation maps I and II to complete the registration of the gray matter horn ROIs from the standard space to the individual functional images; extracting the time series of the gray matter horns corresponding to each ROI, and calculating the motor functional connections and sensory functional connections of each spinal cord segment by using the time series correlation method; performing standardization processing on the low-frequency amplitude with the spinal cord as a mask to obtain the standardized low-frequency amplitude of the spinal cord. The present invention can obtain high-quality and stable images, accurately remove physiological noise, and provide a standardized preprocessing and analysis method, which is of great significance for promoting the application of spinal cord resting-state fMRI in basic scientific research and clinical practice.

Description

Technical Field

[0001] The present invention belongs to the technical field of central nervous system research, and particularly relates to a method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images. Background Art

[0002] The central nervous system is the main component of the nervous system and is composed of the brain and the spinal cord. The spinal cord is a bridge for information conduction between the brain and the peripheral nervous system. It can transmit deep and superficial sensory impulses from the body and most visceral sensory impulses to the brain, thereby generating various sensations. At the same time, the spinal cord can also transmit motor signals sent by the brain to skeletal muscles and internal organs, thereby controlling and coordinating various movements and visceral activities. In addition, the spinal cord is also an independent center for many basic reflexes. The spinal cord plays an important role in human survival and development. Once the spinal cord is damaged, it will lead to functional disorders such as abnormal sensory and motor functions and urinary and fecal incontinence. Therefore, it is of great significance to fully understand the structure and function of the spinal cord nervous system.

[0003] With the development of spinal cord functional magnetic resonance imaging (fMRI) technology, especially the recently emerging spinal cord resting-state fMRI technology, it enables us to non-invasively explore the function of the human spinal cord nervous system. Different from task-based fMRI, when acquiring spinal cord resting-state fMRI images, the subject does not need to perform any sensory or motor tasks, which can reflect the inherent function of the central system and has the characteristics of high efficiency and convenience. This has great application prospects for clinical disease research. However, due to the tissue characteristics of the spinal cord, the complexity of image acquisition sequences and scanning settings, the influence of physiological noise, etc., it is more difficult to control the image quality of spinal cord resting-state fMRI. In addition, there is currently no unified data analysis process, especially the methods for removing physiological noise used in different studies are not the same. Therefore, different from the widely used brain resting-state fMRI, spinal cord resting-state fMRI has many difficulties in current basic scientific research, clinical promotion and use. Summary of the Invention

[0004] In order to achieve the quality control of spinal cord resting-state fMRI images, eliminate the influence of physiological noise and other noises during image acquisition, and obtain stable neural activity signals from the images, the present invention provides a method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images.

[0005] The specific solution is as follows:

[0006] A method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images, setting voxel and Field of view (FOV) parameters, collecting cervical spinal cord images in the sagittal plane, and collecting resting-state functional images and structural images of the spinal cord segments C2-C7; synchronously recording the physiological signals of the subject's respiration and heartbeat; registering the structural image with a standard template to obtain the field transformation map I between the structural image and the standard template; performing denoising preprocessing on the resting-state functional image, registering the structural image and the resting-state functional image with each other to obtain the field transformation map II between the resting-state functional image and the structural image; extracting each gray matter horn of the standard spinal cord template C2-C7, and using the obtained field transformation map I and field transformation map II to complete the registration of each gray matter horn ROI from the standard space to the resting-state functional image; according to the registered gray matter horn ROI, extracting the time series of the corresponding gray matter horn by segment, calculating the motor functional connectivity and sensory functional connectivity (Functional Connectivity, FC) of each spinal cord segment by using the time series correlation method, and performing statistical analysis. Taking the sensory connection as an example, after selecting a certain segment, the time series in each spinal cord horn are extracted respectively, and the specific algorithm is as follows:

[0007] FC = corr(TS LD , TS RD ) (3)

[0008] where FC refers to functional connectivity, TS LD is the time series of the left dorsal horn of this segment, TS RD is the time series of the right dorsal horn, and corr is the correlation coefficient;

[0009] Furthermore, the method also includes the extraction of the Amplitude of Low-Frequency Fluctuations (ALFF) index of the spinal cord. The specific method is as follows:

[0010] Calculating the power spectrum (Power Spectrum, PS) of the preprocessed resting-state functional image, and obtaining the ALFF of the spinal cord according to formula (1):

[0011]

[0012] where ALFF is the low-frequency amplitude, PS is the power spectrum, N1 is the number of frames of the power spectrum image corresponding to the lowest frequency in the ALFF frequency band, and N2 is the number of frames of the power spectrum image corresponding to the highest frequency;

[0013] Using the entire spinal cord from C2 to C7 as a mask, standardizing the obtained ALFF according to formula (2):

[0014] zALFF = (ALFF - Mean) / Std (2)

[0015] Wherein, zALFF is the standardized low-frequency amplitude, and Mean and Std are the mean and standard deviation of ALFF of the entire spinal cord;

[0016] Extract the zALFF values of the corresponding segments according to the gray matter templates of different segments after registration, and perform statistical analysis.

[0017] Preferably, the specific method for preprocessing the scanned spinal cord resting-state functional images is as follows:

[0018] Step 1, perform layer-by-layer translational motion correction on the resting-state functional images in the XY plane to remove the head and neck motion interference;

[0019] Step 2, perform time slice correction by using the method of subtracting the first derivative of the time series from the time series;

[0020] Step 3, perform spatial smoothing processing in the XY plane by using an anisotropic kernel;

[0021] Step 4, perform physiological noise modeling (Physiological Noise Model, PNM) on the recorded respiration and heartbeat signals by using Fourier expansion to remove physiological noise;

[0022] Step 5, use high-pass filtering to remove low-frequency signal drift and retain high-frequency signals. Preferably, use a 0.01 Hz high-pass filter.

[0023] Further preferably, when collecting structural images with GRE-ME or MEDIC sequences, directly perform one-step registration from the structural images to the resting-state functional images to complete the extraction of the ROIs of the resting-state functional images.

[0024] When collecting the resting-state functional images, use the ZOOMit-EPI sequence including Siemens small field of view imaging technology, combined with the GRAPPA (generelized autocalibrating spatially parallel acquisitions) parallel acquisition technology and the local shimming box (Shimming Box) to collect the resting-state functional images of the C2-C7 segments of the spinal cord.

[0025] Or preferably, use the EPI sequence of standard field of view imaging combined with the method of outer volume suppression (OVS) to add saturation bands parallel to the shimming box on both sides of the spinal cord in the sagittal direction, and add a saturation band at the lower edge of the scanning box in the coronal direction to suppress the irrelevant signals of the tissues around the spinal cord.

[0026] The collected spinal cord structural images and resting-state functional images are subjected to coordinate normalization, and then the spinal cord is automatically segmented. The positions of the intervertebral discs from C1 / C2 to C7 / T1 are manually marked to complete the mutual registration between the structural images and the standard template PAM50, and the field transformation map I is obtained.

[0027] The technical solution of the present invention has the following advantages:

[0028] A. Considering the important role of spinal cord resting-state fMRI in non-invasively and task-free exploring the functions of the spinal cord central nervous system and related disease mechanisms, the present invention has developed an image acquisition method with high and stable imaging quality, which can accurately remove physiological noise, and is of great significance for promoting the application of spinal cord resting-state fMRI in basic scientific research and clinical practice.

[0029] B. The parallel acquisition and shimming techniques adopted by the present invention take into account the acquisition speed while obtaining high-quality images, and record the physiological noise for subsequent noise removal. It is applicable to different magnetic resonance platforms and can be widely used in basic scientific research and clinical research related to the intrinsic functions of the spinal cord.

[0030] C. Since the comparison of resting-state results between different studies requires a unified data processing process, the present invention has developed a standardized preprocessing method for spinal cord resting-state fMRI data, providing a basis for the analysis and comparison of different studies. More importantly, spinal cord resting-state fMRI is vulnerable to physiological noise interference. The present invention provides an integrated solution for recording and standardizing the removal of spinal cord physiological noise, which can relatively accurately remove physiological noise and increase the reliability of subsequent statistical analysis.

[0031] D. The spinal cord resting-state functional connectivity and standardized low-frequency amplitude extraction and analysis solutions provided by the present invention can effectively distinguish the differences in the intrinsic functional states of patients and healthy controls, and the two resting-state functional indicators can specifically predict the pain characteristics of patients, further proving the reliability of the data acquisition and analysis solutions of the present invention, and may be used as an effective solution for evaluating the intrinsic functions of the spinal cord in clinical spinal cord lesion patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the specific embodiments of the present invention, the drawings required for the specific embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0033] Figure 1-1 It is a diagram of the T2 structural image scanning settings provided by the present invention;

[0034] Figure 1-2It is a scanning setup diagram of the EPI sequence with ZOOMit function provided by the present invention;

[0035] Figure 1-3 It is a scanning setup diagram of the standard EPI sequence provided by the present invention;

[0036] The markings in Figure 1 are indicated as follows:

[0037] 1 - saturation band; 2 - physiological signal recording; 3 - scanning frame; 4 - shimming frame.

[0038] Figure 2 It is the preprocessing process of spinal cord resting-state fMRI data provided by the present invention;

[0039] Figure 3-1 It is a diagram showing the extraction of sensory and motor resting-state functional connectivity indexes in the spinal cord gray matter region provided by the present invention;

[0040] Figure 3-2 It is a diagram showing the extraction of ALFF values in the spinal cord gray matter region provided by the present invention;

[0041] Figure 4-1 It is the original T2 structural image provided by the present invention;

[0042] Figure 4-2 It is the resting-state fMRI image collected by the standard EPI sequence provided by the present invention and the layered acquisition effect;

[0043] Figure 4-3 It is the resting-state fMRI image collected by the ZOOMit-EPI sequence provided by the present invention and the layered acquisition effect;

[0044] Figure 5-1 It is a diagram showing the sensory functional connectivity analysis of patients and healthy people provided by the present invention;

[0045] Figure 5-2 It is a diagram showing the correlation between sensory functional connectivity and pain sensitivity of patients and healthy people provided by the present invention;

[0046] Figure 6-1 It is a diagram showing the ALFF analysis of patients and healthy people provided by the present invention;

[0047] Figure 6-2 It is a diagram showing a significant negative correlation between the ALFF of the C3 segment of patients and the pain duration provided by the present invention. Detailed implementation manners

[0048] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of them. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0049] The present invention provides a method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images. Voxel and field of view parameters are set, and sagittal images of the cervical spinal cord are collected to acquire the resting-state functional images and structural images of the spinal cord segments C2-C7; the physiological signals of the subject's respiration and heartbeat are synchronously recorded; the structural image is registered with a standard template to obtain the field transformation map I between the structural image and the standard template; the resting-state functional image is preprocessed, and the structural image and the resting-state functional image are mutually registered to obtain the field transformation map II between the resting-state functional image and the structural image; each gray matter horn of the standard spinal cord template C2-C7 is extracted, and using the obtained field transformation map I and field transformation map II, the registration of each gray matter horn ROI from the standard space to the resting-state functional image is completed; according to the registered gray matter horn ROI, the time series of the corresponding gray matter horn are extracted segment by segment, and the motion functional connection and sensory functional connection of each spinal cord segment are calculated using the time series correlation method, and statistical analysis is performed.

[0050] The spinal cord T2-weighted structural image is realized by adjusting the existing high spatial resolution T2 SPACE sequence. Specifically, image acquisition is performed in the sagittal plane, and the voxel resolution is set to 0.8*0.8*0.8mm 3 , the FOV is set to 46.8*280mm 2 (TR = 1500ms, TE = 125ms, FA = 140°), so as to realize the acquisition of the structural image of the cervical spinal cord (C2-C7), as Figure 1-1 shown.

[0051] Among them, the spinal cord resting-state functional images include two imaging methods: ZOOMit-EPI and standard EPI. Method 1 uses Siemens small field of view imaging technology ZOOMit, combined with GRAPPA parallel acquisition technology (taken from the Siemens work-in-progress WIP scientific imaging sequence package), and a local shim box to achieve fast and high-quality acquisition of the resting-state functional images of the cervical spinal cord (C2-C7), as Figure 1-2 . Method 2 is for magnetic resonance instruments without the ZOOMit function. Using the standard field of view imaging technology, in order to reduce the interference of irrelevant signals, this method recommends combining the OVS method, adding saturation bands on both sides of the spinal cord to suppress the signals of the surrounding tissues of the spinal cord, as Figure 1-3As shown below. In addition, in order to obtain fine axial imaging while reducing partial volume effects, both methods use a combination of high planar resolution and relatively large slice thickness to adapt to the special structure of the spinal cord, thereby further improving image quality. The specific parameters are as follows:

[0052] For magnetic resonance instruments with ZOOMit function, the ZOOMit-EPI sequence scheme is adopted: the phase encoding direction is P→A, TR = 2000 ms, TE = 25 ms, Echo spacing = 0.64 ms, FOV = 114 mm, FOV phase = 36.8%, 20 slices, slice thickness 5 mm, in-slice resolution = 1.0 * 1.0 mm 2 , FA = 80°, GRPPA = 2.

[0053] For magnetic resonance instruments without ZOOMit function, the standard EPI sequence combined with the OVS scheme is adopted: the phase encoding direction is P→A, TR = 2140 ms, TE = 28 ms, Echo spacing = 1.05 ms, FOV = 128 mm, FOV phase = 100%, 20 slices, slice thickness 5 mm, in-slice resolution = 1.1 * 1.1 mm 2 , FA = 80°, GRPPA = 2.

[0054] Before the spinal cord resting-state functional image scan, the subject needs to wear a cervical collar to keep the neck stable. At the same time, a respiratory recording strap and a pulse oximeter are required to record respiratory and heartbeat data for subsequent physiological noise removal.

[0055] As Figure 2 shown, the present invention discloses a general spinal cord resting-state fMRI preprocessing process:

[0056] Step 1, motion correction uses layer-by-layer translation correction in the XY direction to reduce the interference of head and neck movement on spinal cord images.

[0057] Step 2, temporal layer correction subtracts the first-order derivative of the time series from the time series to reduce the influence of acquisition time differences.

[0058] Step 3, anisotropic spatial smoothing is adopted, that is, smoothing is only performed in the XY plane and not in the Z direction to reduce the interference of cerebrospinal fluid and white matter signals on gray matter signals.

[0059] Step 4, physiological noise modeling is performed using the recorded respiratory and heartbeat signals to achieve the purpose of noise removal.

[0060] Step 5, a high-pass filter of 0.01 Hz is used to remove low-frequency signal drift, retain high-frequency signals, and enhance the detection of resting-state functional connectivity.

[0061] Step 6: Register the standard spatial gray matter angle template to the individual spatial structural image, and then register the individual spatial structural image to the individual spatial functional image to complete the registration of the ROI through two-step registration. When the structural image is acquired with the GRE-ME or MEDIC sequence, clear gray matter can be directly segmented, and the standard template is not required. The extraction of the ROI of the resting-state functional image can be completed through one-step registration from the structure to the resting-state functional image.

[0062] Certainly, the present invention also includes the extraction of the spinal cord ALFF index, and the specific method is as follows:

[0063] Calculate the PS of the preprocessed resting-state functional image, and obtain the ALFF of the spinal cord according to formula (1):

[0064]

[0065] Where ALFF is the low-frequency amplitude, PS is the power spectrum, N1 is the number of frames corresponding to the power spectrum image with the lowest frequency in the ALFF frequency band, and N2 is the number of frames corresponding to the power spectrum image with the highest frequency;

[0066] Use the entire spinal cord from C2 to C7 as a mask, and standardize the obtained ALFF according to formula (2):

[0067] zALFF = (ALFF - Mean) / Std (2)

[0068] Where: zALFF is the standardized low-frequency amplitude, Mean and Std are the mean and standard deviation of the ALFF of the entire spinal cord;

[0069] Extract the zALFF values of the corresponding segments according to the gray matter templates of different segments after registration, and perform statistical analysis.

[0070] Example

[0071] Recruit 37 patients with lumbar disc herniation (14 females, average age 45.8 years) to participate in the magnetic resonance experimental study. All subjects are right-handed, are required not to have taken analgesic drugs within one week, and have a pain duration of more than 3 months. Recruit 32 healthy subjects (6 females, average age 43.7 years) who are matched in gender and age and have no pain symptoms. During the experiment, the subjects underwent spinal cord MRI scans, including T2 structural images and resting-state fMRI. In addition, collect the pain sensitivity questionnaire and McGill pain questionnaire of the subjects, and record the pain course of the patients to verify the effectiveness of the spinal cord resting-state extraction index.

[0072] The acquisition steps of the spinal cord resting-state functional image are as follows:

[0073]

A01

[0074]

A02

[0075]

A03

[0076]

A04

[0077]

A05

[0078] Perform the following preprocessing process on the collected spinal cord magnetic resonance images:

[0079]

B01

[0080]

B02

[0081]

B03

[0082]

B04

[0083]

B05

[0084]

B06

[0085]

B07

[0086]

B08

[0087]

B09

[0088] Spinal cord resting-state fMRI index extraction and analysis process:

[0089]

C01

B01

B08

[0090]

C02

[0091] Using the above spinal cord resting-state magnetic resonance acquisition and processing procedures, significant differences in spinal cord sensory functional connectivity and ALFF between patients with lumbar disc herniation and healthy people were found, and the resting-state functional indicators of patients could significantly predict pain indicators. Specifically, among the spinal cord resting-state functional connectivity indicators, the sensory functional connectivity of patients with lumbar disc herniation was significantly higher than that of healthy people (t = 2.39, p = 0.02, Figure 5-1 ). The sensory connection of patients was significantly negatively correlated with pain sensitivity (r = -0.44, p = 0.02, Figure 5-2 right), while that of healthy people was significantly positively correlated (r = 0.40, p = 0.03, Figure 5-2 left). However, the sensory connection was not correlated with pain intensity, and the motor connection was not correlated with pain sensitivity, indicating that the association between sensory functional connectivity and pain sensitivity is specific. In the comparison of ALFF, the ALFF value of the C3 segment of patients was significantly lower than that of healthy people (t = 2.70, p = 0.01, Figure 6-1 ), and was significantly negatively correlated with the pain time of patients (r = -0.45, p = 0.03, Figure 6-2 ). However, the ALFF index was not correlated with pain intensity and pain sensitivity, indicating that the association between ALFF and pain time is also specific.

[0092] The above results indicate that the spinal cord resting state image acquisition and data analysis method described in the present invention can not only acquire high-quality spinal cord resting state data, but also find out the differences in the intrinsic functions of the spinal cord between healthy individuals and pain patients through reasonable preprocessing and extraction of resting state indicators. In addition, the extracted functional indicators can specifically reflect pain indicators in different dimensions. Therefore, the present invention can be used to promote the research on the spinal cord nerve function status of healthy individuals and patients with spinal cord lesions.

[0093] Obviously, the above embodiments are merely examples for clear illustration and not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to list all the implementation manners here. And the obvious changes or variations derived therefrom are still within the protection scope of the present invention.

Claims

1. A method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images, characterized in that Set voxel and field of view parameters, and acquire cervical spinal cord images in the sagittal plane, acquiring the resting-state functional images and structural images of the spinal cord segments C2-C7; synchronously record the physiological signals of the subject's respiration and heartbeat; register the structural image with the standard template to obtain the field transformation map I between the structural image and the standard template; perform denoising preprocessing on the resting-state functional images, and register the structural image and the resting-state functional image with each other to obtain the field transformation map II between the resting-state functional image and the structural image; extract each gray matter horn of the standard spinal cord template C2-C7, and use the obtained field transformation map I and field transformation map II to complete the registration of each gray matter horn ROI from the standard space to the individual space resting-state functional image; according to the registered gray matter horn ROI, extract the time series of the corresponding gray matter horn by segment, calculate the motor functional connection and sensory functional connection of each spinal cord segment using the time series correlation method, and perform statistical analysis; The method further includes the extraction of the spinal cord low-frequency amplitude index, and the specific method is as follows: Calculate the power spectrum of the preprocessed resting-state functional image, and obtain the low-frequency amplitude of the spinal cord according to formula (1): ； where ALFF is the low-frequency amplitude, PS is the power spectrum, N1 is the frame number corresponding to the power spectrum image of the lowest frequency in the ALFF frequency band, and N2 is the frame number corresponding to the power spectrum image of the highest frequency; Use the entire spinal cord from C2 to C7 as a mask, and perform normalization processing on the obtained low-frequency amplitude according to formula (2): zALFF=(ALFF-Mean) / Std (2) where: zALFF is the normalized low-frequency amplitude, Mean and Std are the mean and standard deviation of ALFF of the entire spinal cord; Extract the low-frequency amplitude values of the corresponding segments according to the gray matter templates of different segments after registration, and perform statistical analysis; The specific method for preprocessing the scanned spinal cord resting-state functional image is as follows: Step 1, perform layer-by-layer translational motion correction on the resting-state functional image in the XY plane to remove the head and neck motion interference; Step 2, perform time slice correction using the method of subtracting the first derivative of the time series from the time series; Step 3, perform spatial smoothing processing in the XY plane using an anisotropic kernel; Step 4, use Fourier expansion to model the physiological noise of the recorded respiration and heartbeat signals to remove the physiological noise; Step 5, use high-pass filtering to remove the low-frequency signal drift and retain the high-frequency signals.

2. The method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images according to claim 1, wherein When acquiring the structural image with the GRE-ME or MEDIC sequence, directly perform one-step registration from the structural image to the resting-state functional image to complete the extraction of the resting-state functional image ROI.

3. The method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images according to claim 1, wherein When acquiring the resting-state functional image, use the Siemens small field of view imaging technology ZOOMit, combined with the GRAPPA parallel acquisition technology and the local shim frame, to acquire the resting-state functional image of the spinal cord segments C2-C7.

4. The method for acquiring and analyzing spinal cord resting-state functional magnetic resonance images according to claim 1, wherein When acquiring the resting-state functional image, use the EPI sequence of standard field of view imaging combined with the external saturation band suppression method, add saturation bands parallel to the shim frame on both sides of the spinal cord in the sagittal direction, and add a saturation band at the lower edge of the scan frame in the coronal direction to suppress the irrelevant signals of the tissues around the spinal cord.

Images