A Digital Twin Brain Projection Method with Individual Intrinsic Structural Characteristics

Abstract

The present invention discloses a digital twin brain projection method with individual intrinsic structural characteristics, which is applied to the field of biomedical engineering. Aiming at the problem that the existing digital brain cannot reflect the individual intrinsic characteristics of the user to be analyzed, the present invention uses a fresh corpse with a death time less than 2 hours and no lesions or injuries in the head tissues, which is closest to a normal human body; constructs a head brain voxel multimodal image dataset and a human head tissue projection dataset; then collects the head brain voxel multimodal image data of a normal human body, and based on the established projection data index dataset, automatically establishes the brain tissue mapping function of this normal human body to achieve individual intrinsic brain voxel segmentation and definition; then uses the cloud point reconstruction technology to realize the reconstruction of the individual intrinsic brain, so as to systematically complete the digital twin brain projection with individual intrinsic characteristics.

Description

Technical Field

[0001] The present invention belongs to the field of biomedical engineering, and particularly relates to a digital twin brain projection technology. Background Art

[0002] With the continuous progress of science and technology, people's understanding of the brain has become increasingly urgent. A more comprehensive understanding of the brain can not only effectively analyze the pathogenesis of brain diseases, but also more effectively protect the brain, provide the best treatment plan for the brain when it is traumatized, and avoid secondary injuries.

[0003] The use of medical images for tissue segmentation and three-dimensional modeling technology highly restores the internal organizational structure characteristics of the body. Combining with the parameters of the research object, it can effectively simulate the impact of different scenarios on the body, analyze the conduction mechanism of external stimuli inside the body, and reduce the harm and resource waste caused by in-vivo experiments. Currently, this method has been proficiently applied in various fields such as medicine, physics, and chemistry (see patents: ZL 201710710028.8, ZL202111566360.4, papers: Digital three-dimensional sectional anatomy and visualization research of the head and neck). However, the structural modeling and projection with individual eigen-structure characteristics are the key to accurate analysis. In order to obtain a more accurate analysis of the mechanism of action of external stimuli on the body of an individual in a research scenario, it is necessary to ensure that the individual eigen-structure can be accurately projected into the computer. Therefore, establishing a projection data index data set with approximate body characteristics has become the key to this technology. In order to ensure that the data set is as consistent as possible with the images of real living people, a constant temperature environment is used to collect multi-modal brain voxel images of fresh cadavers. During the collection, on the one hand, the collection parameters of the collection equipment should be recorded, and on the other hand, the cadaver should be kept warm. After completing the brain image collection, the cadaver is frozen, sliced, shaped, restored to temperature, and photographed to obtain a brain cross-section image data set. The two data sets are precisely segmented using manual annotation + deep learning algorithms to establish the mapping relationship between tissues and multi-modal brain voxel images. Establish a mapping relationship for the rapid projection of the individual eigen-brain. Realize the digital twin brain computer projection with individual eigen-structure characteristics.

[0004] The research on existing patent technologies that are relatively close to the present invention is as follows:

[0005] ZL 201710710028.8 "Digital Brain Visualization Method, Device, Computing Device and Storage Medium" provides a digital brain visualization method, device, computing device and storage medium. The method includes: when receiving a digital brain visualization request, obtaining the brain image to be visualized in the digital brain visualization request, where the brain image to be visualized includes a magnetic resonance angiography brain image in TOF modality and a magnetic resonance angiography brain image in T1 modality; using a preset cerebrovascular segmentation algorithm to extract the cerebrovascular structure from the magnetic resonance angiography brain image in TOF modality; using a preset extraction algorithm to extract the brain tissue structure from the magnetic resonance angiography brain image in T1 modality; performing three-dimensional registration on the extracted cerebrovascular structure and brain tissue structure to obtain the registered brain image and draw and output it, thereby improving the accuracy of the registration and fusion of brain tissue and cerebrovascular, improving the accuracy of digital brain visualization, and further improving the digital brain visualization effect.

[0006] This patent relates to the digital brain visualization problem in the field of medical image processing, but the acquisition of data sources is not elaborated in detail, and at the same time, there is no description and protection of the related technology of how the individual intrinsic brain projects onto the computer system.

[0007] ZL202111566360.4 "A Virtual-Reality Fusion Method and System for Digital Humans and Entities" provides a virtual-reality fusion method and system for digital humans and entities, belonging to the field of augmented reality technology. This solution realizes the virtual-reality fusion of digital humans and the mechanical structure of entities. Although this patent mentions the combination of digital humans and entities, it only takes the skeleton as the basis and does not describe other tissues. In particular, the method for obtaining the basic data of digital humans and the production method are not mentioned for protection. At the same time, there is no description and protection of the related technology of how the individual intrinsic brain projects onto the computer system.

[0008] The current research on existing papers with solutions relatively close to the present invention is as follows:

[0009] The first part of the doctoral thesis "Research on Digital Three-Dimensional Sectional Anatomy and Visualization of the Head and Neck" by Dr. Liu Guangjiu of the Third Military Medical University mentions using frozen cadavers to produce digital human image datasets, which is relatively close to the method of this patent in terms of multi-modal brain image acquisition and brain sectional image production. However, this technology uses frozen cadavers that are quite different from living human bodies to complete, and the sliced images are also completed in a frozen state. It does not involve the acquisition of multi-modal brain voxel images of fresh cadaver brains that are closest to real human bodies and the acquisition of brain sectional images after temperature recovery. At the same time, there is no relevant description and protection of how to use this technology to project the individual intrinsic brain with individual differences onto the computer. Summary of the Invention

[0010] To solve the above technical problems, the present invention proposes a digital twin brain projection method with individual eigenstructural characteristics, specifically: using individual eigenbrain voxel multimodal brain images to prepare a projection system for a digital twin brain with individual eigencharacteristics.

[0011] The technical solution adopted by the present invention is: a digital twin brain projection method with individual eigenstructural characteristics, including:

[0012] S1. Collect a multimodal medical image dataset of the brain voxels of the head of a cadaver with a death time less than 2 hours;

[0013] S2. Drain and freeze the cadaver, and after the cadaver is completely hardened, use a CNC milling machine to cut the cadaver into slices with a uniform thickness;

[0014] S3. Fix each slice with a glass slide, then restore the temperature of the slice to 36.5°C, and take a photo with a high-definition digital camera to obtain a cross-sectional photo of each slice, thereby establishing a high-definition image dataset of the cross-section of the cadaver's head;

[0015] S4. Combine anatomical knowledge to identify and label the tissues in each slice of the labeled dataset;

[0016] S5. Project the recognition and labeling results of step S4 into the multimodal medical image data of the brain voxels of the cadaver's head to obtain the indexes of different regions and different tissues in the multimodal images; thereby establishing a multimodal image dataset of brain voxels of the head and a projection dataset of human head tissues;

[0017] S6. Under the same image acquisition parameters, collect the multimodal image data of the brain voxels of the current user to be analyzed;

[0018] S7. According to the multimodal image dataset of brain voxels of the head and the projection dataset of human head tissues established in step S5, realize the segmentation and definition of the eigenbrain voxels of the current user to be analyzed;

[0019] S8. Use the cloud point reconstruction technology to realize the reconstruction of the eigenbrain of the current user to be analyzed, and obtain the digital twin brain projection corresponding to the current user to be analyzed.

[0020] Advantages of the present invention: The present invention uses a multi-modal medical image dataset of the head of a fresh corpse (death time less than 2 hours) (CT image, MRI image dataset, MEG image dataset), combined with the cross-sectional image dataset of a frozen corpse after tomography, shaping, and rewarming. According to the imaging situation of tissues in different physical fields, the head tissues of the corpse are labeled in different physical fields, and a multi-modal imaging tissue index database of a digital twin brain is constructed using a multi-modal fusion method, providing an index reference for the extraction of individual brain image tissues. At the same time, an individual acquires an eigen multi-modal brain image according to the multi-modal image acquisition parameters of the index database, and the obtained image dataset performs tissue extraction according to the index database, projecting the individual eigen brain tissue structure information onto a computer system, thereby obtaining a digital twin brain with individual eigen characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is the implementation process of projecting a digital twin brain with individual eigen characteristics of the present invention;

[0022] Figure 2 It is the establishment process of the brain voxel multi-modal image dataset of the present invention;

[0023] Figure 3 It is the establishment process of the brain end-face image dataset;

[0024] Figure 4 It is the establishment process of the projection index dataset;

[0025] Figure 5 It is the computer projection process of the individual eigen digital twin brain. SPECIFIC EMBODIMENTS

[0026] To facilitate those skilled in the art to understand the technical content of the present invention, the content of the present invention will be further explained below with reference to the accompanying drawings.

[0027] To ensure a high degree of consistency between the digital twin brain and the physical brain, the characteristics of the digital twin brain (such as organizational structure characteristics, tissue parameter characteristics, and tissue distribution characteristics, etc.) need to be consistent with the characteristics of the individual's intrinsic brain. To obtain a multi-physical-field brain image dataset that approximates the actual human body, the brain voxel multi-modal imaging is collected in a constant temperature + heat preservation manner, and the device acquisition parameters are recorded during the acquisition, providing a reference basis for the later acquisition of individual intrinsic brain voxel images. After the brain voxel multi-modal imaging of the entity is completed, the brain cross-section images are obtained by using the methods of freezing, slicing, restoring temperature, and taking pictures. The algorithm of personal annotation + deep learning is implemented for the two types of datasets to realize the indexing of the head brain voxel multi-modal image dataset and the human head tissue projection data. Finally, after obtaining the individual intrinsic brain voxel multi-modal image dataset, using the mapping relationship between the obtained cadaver head brain voxel multi-modal image dataset and the human head tissue projection data index, in terms of structure, the projection of the individual intrinsic brain digital twin brain is realized. The detailed process is as Figure 1 shown.

[0028] Generally, the multi-modal imaging tissue index database is the core part of this technology. Therefore, it is necessary to make the index of the index database highly consistent with the multi-modal imaging development value and be able to fully analyze human tissues. The present invention provides a method for establishing a high-precision digital twin brain tissue index. As Figure 2 , first, a fresh cadaver that has died normally within 2 hours and whose head tissues have not undergone lesions or injuries is obtained. After obtaining the cadaver, it is subjected to heat preservation treatment at 35-37 degrees Celsius to ensure that the cadaver remains within the normal human body temperature range (36-37.5°C) during multi-modal imaging acquisition. During the acquisition, the imaging acquisition environment is controlled at the normal experimental temperature, and the normal experimental temperature is 17-25°C, and the cadaver is subjected to temperature maintenance treatment at 35-37 degrees Celsius. A multi-modal image dataset of the fresh cadaver head brain voxel is obtained. The device control parameters of different modalities of the device need to be recorded during the acquisition, providing a reference for the later multi-modal imaging acquisition of individuals. Here, the multi-modal specifically includes MRI images, CT images, and MEG image datasets.

[0029] As Figure 3 , after the multi-modal image data acquisition is completed, the cadaver needs to be drained to prevent organ rupture caused by the phase change of body fluids during freezing. After draining about 1 / 15 of the body weight of the liquid, the cadaver is frozen for about 1-2 weeks. After the cadaver is completely hardened, it is cut into 1-mm-thick slices using a CNC milling machine. After the cut slices are shaped using glass slides with a thickness less than 1 mm, their temperature is restored. The temperature restoration is completed in a constant temperature restoration chamber at 36.5°C. When the overall temperature of the slices reaches 36.5°C, high-definition digital cameras are used to take pictures to obtain the cross-section photos of each slice, thereby constructing a high-definition image dataset of the fresh cadaver head cross-section.

[0030] To obtain more accurate voxel information of brain tissue structures, precise image segmentation becomes very important. For example Figure 4 , to obtain a more accurate tissue index of multimodal medical image data, using the sectional image dataset as the annotation dataset and the multimodal medical image dataset as the training set, combining anatomical knowledge, and using the method of visual interpretation + deep learning to identify and label the sectional tissues. The sectional image dataset is labeled for tissue types and tissue boundaries through visual interpretation, and the labeled results are projected into the multimodal images through deep learning, so as to obtain the indexes of different regions and different tissues in the multimodal images. Thus, a multimodal image dataset of head brain voxels and a human head tissue projection dataset are established.

[0031] After the projection index dataset is established, under the same image acquisition parameters, a mapping relationship will be automatically established between the brain voxel multimodal image and the projection index of any individual. For example Figure 5 as shown, after any individual completes image acquisition according to the acquisition parameters of the fresh cadaver brain voxel multimodal image, an individual eigenbrain image dataset will be established. This dataset automatically establishes a tissue mapping function using the projection data index dataset established from the fresh cadaver to realize the segmentation and definition of the individual eigenbrain voxels. After the segmentation and definition are completed, the projection image of the individual eigenbrain structure is established, and the reconstruction of the individual eigenbrain is realized using the cloud point reconstruction technology, so as to complete the digital twin brain projection with individual eigen characteristics by the system. At this time, the individual eigenbrain is digitally projected into the computer to realize the digital twin in structure.

[0032] Those of ordinary skill in the art will realize that the embodiments described herein are to assist the reader in understanding the principles of the present invention and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. For those skilled in the art, various changes and modifications can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the scope of the claims of the present invention.

Claims

1. A digital twin brain projection method with individual intrinsic structural characteristics, characterized in that, Including: S1. Collect a multi-modal medical image dataset of the brain voxels of the head of a corpse with a death time less than 2 hours; S2. Drain and freeze the corpse. After the corpse is completely hardened, use a CNC milling machine to cut the head into slices with a consistent thickness; S3. Fix each slice with a glass slide, then restore the slice temperature to 36.5 °C, and take pictures with a high-definition digital camera to obtain the cross-sectional photos of each slice, thereby establishing a high-definition image dataset of the corpse cross-section; S4. Combine anatomical knowledge to identify and label the tissues in each slice of the high-definition image dataset of the corpse cross-section; S5. Project the recognition and annotation results of step S4 into the multi-modal medical image data of the brain voxels of the corpse head to obtain the indices of different regions and different tissues in the multi-modal image; Step S5 uses the method of deep learning. Take the high-definition image dataset of the corpse cross-section after being recognized and labeled in step S4 as the labeled dataset, and take the multi-modal medical image dataset in step S1 as the training set. Through deep learning, project the labeled results into the multi-modal image to obtain the indices of different regions and different tissues in the multi-modal image; S6. Collect the multi-modal image data of the brain voxels of any individual; S7. According to the index established in step S5, realize the segmentation and definition of the eigenbrain voxels of any individual; S8. Use the cloud point reconstruction technology to realize the reconstruction of the eigenbrain of any individual, and obtain the digital twin brain projection corresponding to any individual.

2. A digital twin brain projection method with individual intrinsic structural characteristics according to claim 1, characterized in that When collecting in step S1, the multi-modal medical image acquisition environment is controlled at the normal experimental temperature, and the corpse is maintained at a temperature of 35-37 degrees Celsius.

3. A digital twin brain projection method with individual intrinsic structural characteristics according to claim 2, characterized in that, Step S1 specifically obtains the MRI image, CT image and MEG image corresponding to the brain voxels of the head by using an MRI device, a CT device and a MEG device, thereby obtaining a multi-modal medical image dataset.

4. A digital twin brain projection method with individual intrinsic structural characteristics according to claim 3, characterized in that, Step S1 also includes recording the device control parameters of the MRI device, CT device and MEG device.

5. A digital twin brain projection method with individual intrinsic structural characteristics according to claim 4, characterized in that, Step S4 specifically identifies and labels the tissue type and tissue boundary.

6. A digital twin brain projection method with individual intrinsic structural characteristics according to claim 5, characterized in that, In step S6, under the same device control parameters of the MRI device, CT device and MEG device as in step S1, collect the multi-modal image data of the brain voxels of any individual.

Images