Image processing method and system based on brain white matter magnetic resonance signal intensity

By processing dMRI and T2 Flair images using FSL software, and combining ROI segmentation and generalized linear regression, the problem of quantitative assessment of white matter lesions in existing technologies has been solved, enabling quantitative assessment of white matter lesions and extraction of pathological indicators.

CN118351102BActive Publication Date: 2024-11-19RUIJIN HOSPITAL AFFILIATED TO SHANGHAI JIAO TONG UNIV SCHOOL OF MEDICINE
View PDF 16 Cites 0 Cited by

Patent Information

Application Number
CN202410546518.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-05-06
Publication Date
2024-11-19
Estimated Expiration
2044-05-06

AI Technical Summary

Technical Problem

Existing technologies are insufficient to effectively reflect the signal intensity and pathological degree of white matter lesions through MRI images, and lack quantitative calculation of high signal in white matter and extraction and application of indicators of pathological changes.

Method used

Using an image processing method based on FSL software, combined with dMRI and high-resolution T2 Flair images, the signal intensity ratio of high signal regions in white matter was calculated through edge recognition and ROI segmentation. A mapping model was then established using generalized linear regression to quantify the severity of white matter lesions.

Benefits of technology

It enables quantitative assessment of white matter lesions, reflects the severity of white matter lesions, provides pathological indicators of myelin sheath, and supports quantitative analysis of brain white matter lesions.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118351102B_ABST
    Figure CN118351102B_ABST
Patent Text Reader

Abstract

The invention discloses an image processing method and system based on brain white matter magnetic resonance signal intensity, comprising the following steps: step S1, obtaining a dMRI image and a high-resolution T2 Flair image; step S2, applying FSL software to preprocess the dMRI image; step S3, segmenting the T2 Flair image to obtain a region of white matter high signal WMH; making a WMH seed mask in the whole brain image; step S4, extracting multiple white matter high signal WMH lesion regions using edge recognition technology and identifying their respective boundaries; step S5, obtaining an outerROI of a normal brain region outside the lesion and an innerROI of a region inside the lesion; step S6, obtaining a WMH intensity ratio; step S7, obtaining FWF values ​​of the outerROI and innerROI regions, the ratio of the two being rFWF; step S8, determining a WMH intensity ratio related to rFWF by determining a fitting coefficient; step S9, obtaining a parameter and a mapping model for transforming the WMH intensity ratio to rFWF, and using the model to obtain a quantitative numerical value of the severity of white matter lesions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of image recognition, and in particular to an image processing method and system based on brain white matter magnetic resonance signal intensity. Background Art

[0002] White matter lesions are a common type of brain injury, which is related to physiological factors such as age and gender, and can also be caused by a variety of pathological reasons such as cerebrovascular abnormalities, cerebral ischemia, and neurodegeneration. The main components of white matter are nerve fibers emitted by neurons. When lesions occur, they can cause nerve signal conduction disorders or interruptions, resulting in various clinical symptoms, including limb dysfunction, hemiplegia, cognitive dysfunction, gait abnormalities, emotional disorders, etc. White matter lesions can be imaged by MRI (nuclear magnetic resonance imaging), usually manifested as high signals (WMH) in T2 Flair (fluid-attenuated inversion recovery sequence) images.

[0003] The distribution of white matter hyperintensities in the brain, the size of lesions, and the morphology of lesions can reflect the scope of white matter lesions, and the signal intensity of white matter hyperintensities is related to the pathological degree of white matter lesions. In addition, when white matter fibers are damaged, the diffusion freedom of hydrogen protons in the water in the fibers changes, from the original axial free motion to all-directional free motion, that is, the content and proportion of free water increase. dMRI (multiple b-value diffusion-weighted imaging) can detect the diffusion direction of hydrogen protons and calculate the proportion of free water, thereby judging the degree of damage to nerve fibers. In the past, quantitative calculations of white matter hyperintensities focused on how to segment and rate lesion areas, and could only extract the volume and number of lesions of white matter hyperintensities, but could not reflect the signal intensity of white matter hyperintensities, and lacked the extraction and application of indicators related to pathological changes. Summary of the invention

[0004] The object of the present invention is to provide an image processing method and system based on brain white matter magnetic resonance signal intensity to solve the problems raised in the above background technology.

[0005] To achieve the above-mentioned object of the invention, one aspect of the present invention provides an image processing method based on brain white matter magnetic resonance signal intensity, comprising the following steps:

[0006] Step S1, performing an MRI scan on the subject to obtain a dMRI image and a high-resolution T2 Flair image;

[0007] Step S2, in the Linux operating system, using FSL software to perform dMRI image preprocessing;

[0008] Step S3, in the Linux operating system, the FSL-BIANCA function module is used to segment the T2 Flair image to obtain the area of ​​white matter high signal WMH; in the whole brain image, the signal value of the white matter high signal area is set to 1, and the signal value of the other normal areas is set to 0 to make a WMH seed mask;

[0009] Step S4, using edge recognition technology to extract multiple white matter high signal WMH lesion areas and identify their respective boundaries, and any connected points, surfaces, and lines of each voxel cube are considered to be the same white matter high signal area;

[0010] Step S5, using 3D expansion and contraction technology, at the edge of a single region, obtain the outerROI of the normal brain region outside the lesion and the innerROI of the region inside the lesion;

[0011] Step S6, collecting the average T2Flair weighted values ​​of the outerROI of the normal brain region outside the lesion and the innerROI of the region inside the lesion, and defining the ratio of the two as the WMH intensity ratio;

[0012] Step S7, use the image registration tool of FSL to register the high-resolution T2 Flair image to the dMRI, and simultaneously complete the corresponding registration of the outerROI and innerROI, and obtain the FWF (free water fraction) values ​​of the outerROI and innerROI regions, the ratio of which is rFWF;

[0013] Step S8, fitting the WMH intensity ratios of the eight white matter high signal areas to the rFWF of the area by generalized linear regression, and determining the WMH intensity ratios related to the rFWF by determining the fitting coefficient;

[0014] Step S9, obtaining the parameters and mapping model for transforming the WMH intensity ratio to rFWF by fitting the function, and using the model to obtain the quantitative value of the severity of white matter lesions.

[0015] Furthermore, in step S1, the dMRI image is based on gradient echo planar imaging, where TR / TE=

[0016] 3500 / 77.3ms; voxel size = 1.5×1.5×1.5mm 3 ; Number of slices: 100; Diffusion direction = 32 / 64; Two b values: 1500 and 3000; Phase encoding direction = PA; 4 pairs of b0 images were acquired in opposite phase encoding directions to correct for susceptibility-induced distortion.

[0017] Furthermore, in step S1, a high-resolution T2 Flair image, wherein TR = 6000ms, TE = 408ms, and TI = 1841ms, and voxel size = 0.75×0.75×0.75mm 3 .

[0018] Furthermore, the dMRI image preprocessing process is to correct the original diffusion MRI image for susceptibility-induced distortion using the Top-Up method of FSL software, and then use the Eddy tool in FSL to correct the head motion and eddy current.

[0019] Furthermore, in step S4, all WMH lesion areas are subjected to image edge recognition and clustering. Specifically, voxels marked as 1 are explored one by one, and it is explored in 3D space whether there are voxels connected to the voxel that are also marked as 1. If so, they are classified into the same cluster until the edge is explored.

[0020] Furthermore, in step S5, based on the WMH lesion edge identified in step S4, a distance of 2 voxels is extended away from the center of the Cluster, and the area of ​​this extended segment is defined as the outerROI, which is the normal brain area outside the lesion; conversely, 2 voxels are extended toward the center of the Cluster, and the area within the lesion edge is defined as the innerROI.

[0021] Furthermore, in step S8, the eight white matter high signal areas are the periventricular white matter high signals and deep white matter high signals in the frontal, temporal, parietal and occipital lobes.

[0022] Furthermore, in step S8, the region that cannot be fitted indicates that the WMH intensity ratio cannot reflect the pathological changes of rFWF, and this region is not included in the fitting calculation model.

[0023] Another aspect of the present invention provides an image processing system based on brain white matter magnetic resonance signal intensity, the system comprising an image acquisition module, a brain white matter high signal area and contrast area segmentation module, and a WMH pathology degree analysis module, wherein:

[0024] An image acquisition module, used for performing MRI scan on the subject to obtain high-resolution T2 Flair images and dMRI images;

[0025] The brain white matter pathology fitting module is used to apply the FSL tool to preprocess T2 Flair images and dMRI images, perform WMH segmentation, and align innerROI and outerROI, and calculate the myelin case index rFWF of the WMH area based on dMRI, and establish a mapping relationship between WMH intensity value and rFWF;

[0026] The white matter pathology analysis module is used to obtain the white matter high signal area and intensity ratio, as well as the myelin pathology index of each WMH area.

[0027] Compared with the prior art, the present system and method have the following advantages:

[0028] The present invention obtains high-resolution T2 Flair images and dMRI images by performing MR scanning on subjects. The multimodal images are analyzed and processed using FSL according to the above steps, and the relative values ​​of the T2Flair intensity of the WMH area and the WMH lesions can be located. The rFWF obtained by dMRI is used as the standard for myelin pathological changes, the model parameters for mapping the WMH intensity ratio to the rFWF are calculated, and it is determined whether WMH will cause changes in the myelin components of the white matter in the area; at the same time, the T2 Flair images of other populations in this white matter area can be input into the mapping model generated by the present invention to obtain the myelin pathology index of the severity of white matter lesions in the T2 Flair images of MR of the general population. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 The figure is a flow chart of the image processing method based on the intensity of brain white matter magnetic resonance signals.

[0030] Figure 2 Schematic diagram of the image processing system based on the intensity of brain white matter magnetic resonance signals. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] like Figure 1 The flowchart of the method of the present invention is shown, and the multimodal image processing method comprises the following steps:

[0033] Step S1, performing an MRI scan on a subject to obtain a dMRI image and a high-resolution T2 Flair image.

[0034] Step S2: In the Linux operating system, FSL software is used to perform dMRI image preprocessing.

[0035] Step S3, in the Linux operating system, the FSL-BIANCA function module is used to segment the T2 Flair image to obtain the white matter high signal WMH area; in the whole brain image, the signal value of the white matter high signal area is set to 1, and the other normal areas are set to 0 to make a WMH seed mask.

[0036] Step S4, using edge recognition technology, can extract multiple white matter high signal WMH lesion areas and identify their respective boundaries. Any connected points, surfaces, and lines of each voxel cube are considered to be the same white matter high signal area, and if they are not connected, they are judged to be different white matter high signal areas.

[0037] Step S5, using 3D expansion and contraction technology, at the edge of a single region, expand 2 voxels away from the center to obtain the outerROI of the normal brain region outside the lesion; at the edge of a single region, expand 2 voxels toward the center to obtain the innerROI of the region inside the lesion.

[0038] Step S6, collecting the average T2Flair weighted values ​​of the outerROI of the normal brain region outside the lesion and the innerROI of the region inside the lesion, and defining the ratio of the two as the WMH intensity ratio.

[0039] Step S7, use FSL's image registration tool to register the high-resolution T2 Flair image to the dMRI, and simultaneously complete the corresponding registration of the outerROI and innerROI. In the dMRI space, the dMRI image is processed by the AMICO tool and the FWF values ​​of the outerROI and innerROI regions are calculated, and the ratio of the two is defined as rFWF.

[0040] Step S8, for the eight different white matter areas of paraventricular white matter hyperintensities and deep white matter hyperintensities in the frontal, temporal, parietal, and occipital lobes, the WMH intensity ratio of each white matter hyperintensity area is fitted to the rFWF of the area by generalized linear regression, and factors such as age and gender are corrected. By determining the fitting coefficient, the WMH intensity ratio associated with rFWF is determined. The area that cannot be fitted indicates that the WMH intensity ratio cannot reflect the pathological changes of rFWF, and the area is not included in the fitting calculation model.

[0041] Step S9, obtaining the parameters and mapping model for transforming the WMH intensity ratio to rFWF by fitting the function, thereby applying the model to other T2 Flair images of the white matter area to obtain a quantitative value of the severity of white matter lesions.

[0042] In step 1, the dMRI image is based on gradient echo planar imaging, and the specific parameters are: TR / TE = 3500 / 77.3ms; voxel size = 1.5×1.5×1.5mm 3 ; Number of slices: 100; Diffusion direction = 32 / 64; Two b values ​​= 1500, 3000; Phase encoding direction = PA (posterior-anterior). 4 pairs of b0 images were acquired with opposite phase encoding directions (anterior-posterior, posterior-anterior) to correct for susceptibility-induced distortion.

[0043] In addition, the high-resolution T2 Flair image, the specific parameters are: TR = 6000ms, TE = 408ms, and TI = 1841ms, voxel size = 0.75×0.75×0.75mm 3 .

[0044] In step 2, the dMRI image preprocessing process is as follows: the original diffusion MRI image is corrected for susceptibility-induced distortion using the Top-Up method of FSL software, and then the head motion and eddy current are corrected using the Eddy tool in FSL.

[0045] In step 4, all WMH lesion areas are subjected to image edge recognition and clustering. Specifically, the voxels marked as 1 are explored voxel by voxel, and whether there are voxels connected to the voxel that are also marked as 1 in the 3D space. If so, they are classified into the same cluster until the edge is explored.

[0046] In step 5, based on the WMH lesion edge identified in step 4, the distance of 2 voxels is extended away from the center of the Cluster, and the area of ​​this extended segment is defined as the outerROI, which is the normal brain area outside the lesion; conversely, 2 voxels are extended toward the center of the Cluster, and the area within the lesion edge is defined as the innerROI.

[0047] Those skilled in the art will appreciate that all or part of the steps in the above-mentioned embodiment method can be implemented by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes the steps of the above-mentioned embodiment method, and the storage medium can be: ROM / RAM, magnetic disk, optical disk, memory card, etc.

[0048] Therefore, in contrast to the method of the present invention, the present invention also includes an image processing system based on brain white matter magnetic resonance signal intensity, which can implement each step of the image processing method based on brain white matter magnetic resonance signal intensity. Figure 2 As shown, the system includes an image acquisition module, a white matter pathology fitting module, and a white matter pathology analysis module, wherein:

[0049] An image acquisition module, used for performing MRI scan on the subject to obtain high-resolution T2 Flair images and dMRI images;

[0050] The brain white matter pathology fitting module is used to apply the FSL tool to preprocess T2 Flair images and dMRI images, perform WMH segmentation, and align innerROI and outerROI, and calculate the myelin case index rFWF of the WMH area based on dMRI, and establish a mapping relationship between WMH intensity value and rFWF;

[0051] The white matter pathology analysis module is used to obtain the white matter high signal area and intensity ratio, as well as the myelin pathology index of each WMH area.

[0052] The present invention provides an image processing method of multimodal MR (including T2Flair, multi-b value dMRI) capable of reflecting pathological destruction of nerve fiber myelin sheath, which can be based on the white matter high signal area WMH located by T2 Flair, adopt edge recognition technology to scale the lesion voxel by voxel, collect the signal value of normal tissue inside and around the WMH lesion at the same time, and then calculate the FWF of normal tissue inside and around the WMH lesion by dMRI, and finally obtain the contribution rate β of pathological index rFWF to WMH intensity ratio, and calculate the myelin pathological index rFWF by WMH intensity ratio after reverse mapping, so as to achieve the effect of extracting pathological index of white matter high signal. The present invention also provides an analysis module, which can analyze the T2 Flair image obtained when the general individual performs MRI scanning, and can provide the index of pathological change of nerve fiber myelin sheath in the white matter high signal area without obtaining dMRI.

[0053] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An image processing method based on brain white matter magnetic resonance signal intensity, characterized in that: The following steps are involved: Step S1, performing an MRI scan on the subject to obtain a dMRI image and a high-resolution T2 Flair image; Step S2, in the Linux operating system, using FSL software to perform dMRI image preprocessing; Step S3, in the Linux operating system, the FSL-BIANCA function module is used to segment the T2 Flair image to obtain the area of ​​white matter high signal WMH; in the whole brain image, the signal value of the white matter high signal area is set to 1, and the signal value of the other normal areas is set to 0 to make a WMH seed mask; Step S4, using edge recognition technology to extract multiple white matter high signal WMH lesion areas and identify their respective boundaries, and any connected points, surfaces, and lines of each voxel cube are considered to be the same white matter high signal area; Step S5, using 3D expansion and contraction technology, at the edge of a single region, obtain the outerROI of the normal brain region outside the lesion and the innerROI of the region inside the lesion; Step S6, collecting the average T2Flair weighted values ​​of the outerROI of the normal brain region outside the lesion and the innerROI of the region inside the lesion, and defining the ratio of the two as the WMH intensity ratio; Step S7, use the image registration tool of FSL to register the high-resolution T2 Flair image to the dMRI, and simultaneously complete the corresponding registration of the outerROI and innerROI, and obtain the FWF values ​​of the outerROI and innerROI regions, and the ratio of the two is rFWF; Step S8, fitting the WMH intensity ratios of the eight white matter high signal areas to the rFWF of the area by generalized linear regression, and determining the WMH intensity ratios related to the rFWF by determining the fitting coefficient; Step S9, obtaining the parameters and mapping model for transforming the WMH intensity ratio to rFWF by fitting the function, and using the model to obtain the quantitative value of the severity of white matter lesions.

2. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: In step S1, the dMRI image is based on gradient echo planar imaging, where TR / TE=3500 / 77.3 ms; voxel size=1.5×1.5×1.5 mm 3 ; Number of slices: 100; Diffusion direction = 32 / 64; Two b values: 1500 and 3000; Phase encoding direction = PA; 4 pairs of b0 images were acquired in opposite phase encoding directions to correct for susceptibility-induced distortion.

3. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: In step S1, high-resolution T2 Flair images, where TR = 6000ms, TE = 408ms, and TI = 1841ms, voxel size = 0.75×0.75×0.75mm 3 .

4. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: The dMRI image preprocessing process is to correct the magnetic susceptibility induced distortion of the original diffusion MRI image by the Top-Up method of FSL software, and then use the Eddy tool in FSL to correct the head motion and eddy current.

5. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: In step S4, all WMH lesion areas are subjected to image edge recognition and clustering. Specifically, voxels marked as 1 are explored one by one, and whether there are voxels connected to the voxel that are also marked as 1 in the 3D space. If so, they are classified into the same cluster until the edge is explored.

6. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: In step S5, based on the WMH lesion edge identified in step S4, a distance of 2 voxels is extended away from the center of the Cluster, and the area of ​​this extended segment is defined as the outerROI, which is the normal brain area outside the lesion; conversely, 2 voxels are extended toward the center of the Cluster, and the area within the lesion edge is defined as the innerROI.

7. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: In step S8, the eight brain white matter high signal areas are the periventricular white matter high signal areas and deep brain white matter high signal areas of the frontal, temporal, parietal and occipital lobes.

8. The image processing method based on brain white matter magnetic resonance signal intensity according to claim 1, characterized in that: In step S8, the region that cannot be fitted indicates that the WMH intensity ratio cannot reflect the pathological changes of rFWF, and this region is not included in the fitting calculation model.

9. An image processing system based on brain white matter magnetic resonance signal intensity, characterized in that: The system includes an image acquisition module, an image preprocessing module, an image segmentation module, an image recognition module, a ROI module, a T2 weighting module, an image registration module, a white matter pathology fitting module, and a white matter pathology analysis module, wherein: An image acquisition module, used for performing MRI scan on the subject to obtain high-resolution T2 Flair images and dMRI images; An image preprocessing module is used to perform dMRI image preprocessing using FSL software in the Linux operating system; An image segmentation module is used to segment the T2 Flair image using the FSL-BIANCA function module in the Linux operating system to obtain the area of ​​white matter high signal WMH; in the whole brain image, the signal value of the white matter high signal area is set to 1, and the other normal areas are set to 0 to make a WMH seed mask; Image recognition module, used to extract multiple white matter high signal WMH lesion areas using edge recognition technology and identify their respective boundaries. Any connected points, surfaces, and lines of each voxel cube are considered to be the same white matter high signal area; The ROI module is used to obtain the outerROI of the normal brain region outside the lesion and the innerROI of the region inside the lesion at the edge of a single region by using 3D expansion and contraction technology; The T2 weighted module is used to collect the average T2 Flair weighted values ​​of the outerROI of the normal brain area outside the lesion and the innerROI of the area inside the lesion, and the ratio of the two is defined as the WMH intensity ratio; Image registration module, used to register high-resolution T2 Flair images to dMRI using FSL's image registration tool, synchronously complete the corresponding registration of outerROI and innerROI, and obtain the FWF values ​​of the outerROI and innerROI regions. The ratio of the two is Rfwf; The brain white matter pathology fitting module is used to fit the WMH intensity ratios of 8 brain white matter high signal areas to the rFWF of the area through generalized linear regression, and determine the WMH intensity ratios related to rFWF by determining the fitting coefficient; The brain white matter pathology analysis module is used to obtain the parameters and mapping model for transforming the WMH intensity ratio to rFWF through a fitting function, and use the model to obtain the quantitative value of the severity of white matter lesions.

Citation Information

Patent Citations

  • White matter high signal segmentation method, device and equipment and storage medium

    CN114419066A

  • Patient light stimulation parameter selection method and system based on brain function network

    CN114984457A

  • Universal MR nerve function / metabolism imaging navigation method and system

    CN115359004A

  • RET (transfection rearrangement) for assessment of stroke

    CN115698721A

  • Method for automatically segmenting and analyzing white matter high signals based on MR deep learning

    CN116543159A