Method and apparatus for predicting degree of disability of multiple sclerosis patient
The method uses a white matter pathway density index (Le-TDI) to predict disability in multiple sclerosis patients, addressing the lack of correlation in traditional brain imaging indicators, providing a more reliable prediction of disability.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- SAMSUNG MEDICAL CENT
- Filing Date
- 2024-12-31
- Publication Date
- 2026-07-02
Smart Images
Figure KR2024021584_02072026_PF_FP_ABST
Abstract
Description
Method and device for predicting the degree of disability in patients with multiple sclerosis
[0001] The present invention relates to a method and apparatus for predicting the degree of disability in patients with multiple sclerosis.
[0002] Multiple sclerosis (MS) is a demyelinating disease of the central nervous system that occurs frequently in young adults, primarily in their 20s to 40s. As a chronic inflammatory disease, multiple sclerosis can lead to long-term neurological impairment.
[0003] The importance of imaging (e.g., MRI) in the diagnosis and monitoring of multiple sclerosis is increasing. It is generally observed that patients with lesions have a worse clinical condition compared to those without. However, in the case of multiple sclerosis, traditional brain imaging findings tend not to correlate well with clinical symptoms and the course of the disease. This tendency is referred to as the "clinico-radiological paradox" in multiple sclerosis.
[0004] As a solution to the clinical-radiological paradox, a Disconnectome approach has been proposed that utilizes indicators (hereinafter referred to as brain connectivity pathology indicators) that quantify the effects or conditions resulting from the disruption of structural connections in the brain. The human brain can be modeled as a network of connected localized regions, and disruptions in brain networks can lead to dysfunction. In other words, the Disconnectome approach is a method that combines brain lesions with the brain's structural connectomics to investigate the effects and conditions caused by brain lesions and the disruption of brain networks. Brain connectivity pathology indicators based on the Disconnectome approach have been shown to be capable of distinguishing between multiple sclerosis patients and a healthy population.
[0005] However, brain connectivity pathology markers derived from the disconnectome approach do not predict the degree of clinical disability in multiple sclerosis. For example, a study using an atlas-based brain lesion disconnectome approach reported that in the case of multiple sclerosis, there was no significant correlation between lesion-derived network metrics and the Expanded Disability Status Scale (EDSS) (Ravano et al., 2021. Validating atlas-based lesion disconnectomics in multiple sclerosis: a retrospective multi-centric study. NeuroImage Clin. 32, 102817). In another example, it has been reported that in the case of multiple sclerosis, no significant association was found between the lesion-derived global network metric and EDSS and PASAT (Paced auditory serial addition test) (Pagani et al., 2020. Structural connectivity in multiple sclerosis and modeling of disconnection. Mult. Scler. J. 26, 220-232).
[0006] Meanwhile, US 11,995,832 discloses a method and system for characterizing the effect of brain lesions on brain connectivity based on a tractography atlas.
[0007] One objective of the present invention is to provide a method and apparatus for predicting the degree of disability of a patient with multiple sclerosis.
[0008] A method for predicting the degree of disability of a multiple sclerosis patient according to exemplary embodiments of the present invention may include: a step of obtaining information about a brain lesion region of a multiple sclerosis patient, which is executed by at least one processor; a step of mapping the brain lesion region to a white matter streamline map that includes a plurality of voxels and information about the number of white matter streamlines passing through each voxel, and identifying the voxels within the brain lesion region in the white matter streamline map; a step of obtaining the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region; a step of determining a white matter path index of the brain lesion region based on the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region; and a step of predicting the degree of disability of the multiple sclerosis patient based on the white matter path index of the brain lesion region using a rule-based model or a trained artificial intelligence model.
[0009] In one embodiment, the step of obtaining information about the brain lesion region includes obtaining a brain lesion region mask representing the brain lesion region of the multiple sclerosis patient, and mapping the brain lesion region to the white matter streamline map may include overlapping the brain lesion region mask with the white matter streamline map.
[0010] In one embodiment, the white matter streamline map may be obtained by averaging the white matter streamline map of a normal group in voxel units.
[0011] In one embodiment, the white matter pathway index of the brain lesion region may be determined based on a simple sum or a weighted sum of the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region.
[0012] In one embodiment, the white matter pathway index of the brain lesion region includes a white matter pathway density index of the brain lesion region, and the white matter pathway density index of the brain lesion region may be defined as a value obtained by dividing the simple sum or weighted sum of the number of white matter streamlines passing through each voxel within the brain lesion region by the number of voxels within the brain lesion region.
[0013] In one embodiment, the degree of impairment may include an indicator that quantifies or categorizes the degree of neurological impairment.
[0014] In one embodiment, the above-described method further comprises the step of obtaining at least one of a brain connectivity pathology indicator and a brain lesion characteristic indicator of the multiple sclerosis patient; and the step of predicting the degree of disability of the multiple sclerosis patient may include predicting the degree of disability of the multiple sclerosis patient based on a white matter pathway indicator of the brain lesion region and at least one of the brain connectivity pathology indicator and the brain lesion characteristic indicator.
[0015] In one embodiment, the brain connectivity pathology indicator is an indicator that quantifies the effect of a brain lesion on a brain network or the state of a brain network due to a brain lesion, and may include at least one of the density of the brain network, clustering coefficient, global efficiency, micro-worldliness, metastaticity, and characteristic path length. The brain lesion characteristic indicator may include the volume of the brain lesion.
[0016] An apparatus for predicting the degree of disability of a patient with multiple sclerosis according to exemplary embodiments of the present invention may include at least one memory; and at least one processor that executes instructions stored in the at least one memory.
[0017] The above-mentioned at least one processor can obtain information about a brain lesion region of a multiple sclerosis patient, map the brain lesion region to a white matter streamline map that includes a plurality of voxels and information about the number of white matter streamlines passing through each voxel, identify the voxels within the brain lesion region in the white matter streamline map, obtain the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region, determine a white matter path index of the brain lesion region based on the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region, and control the prediction of the degree of disability of the multiple sclerosis patient based on the white matter path index of the brain lesion region using a rule-based model or a trained artificial intelligence model.
[0018] According to exemplary embodiments of the present invention, an application program stored in a recording medium to execute the above-described method may be provided when operated by at least one processor.
[0019] According to exemplary embodiments of the present invention, a method and apparatus for predicting the degree of disability of a patient with multiple sclerosis may be provided.
[0020] According to exemplary embodiments of the present invention, a white matter pathway index (e.g., a white matter pathway density index) of a brain lesion region in a patient with multiple sclerosis can be calculated. Based on the white matter pathway index of the brain lesion region, the degree of disability (e.g., the degree of neurological disability) of a patient with multiple sclerosis can be reliably predicted.
[0021] Figure 1 is a figure showing the results of analyzing the correlation between traditional brain imaging pathology indicators obtained from multiple sclerosis patients, traditional neurological disability assessment scales, and white matter pathway density indicators of brain lesion regions according to one embodiment of the present invention through graphic LASSO-based network analysis.
[0022] FIG. 2 shows a block diagram of a system for providing a service for predicting the degree of disability of a multiple sclerosis patient according to an embodiment of the present invention.
[0023] FIG. 3 shows a block diagram of a device according to one embodiment of the present invention.
[0024] FIGS. 4 to 6 are flowcharts illustrating a method for predicting the degree of disability of a patient with multiple sclerosis according to one embodiment of the present invention.
[0025] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but can be implemented in various different forms.
[0026] To clearly explain the embodiments of the present invention, parts unrelated to the description may be omitted. Additionally, in describing the embodiments of the present invention, if it is determined that a detailed description of related known components or functions could obscure the essence or description of the present invention, such detailed description may be omitted.
[0027] In this specification, the terms “…part,” “…unit,” and “…module” may refer to a unit that processes at least one function or operation. The “…part,” “…unit,” and “…module” may be implemented in hardware, software, or a combination of hardware and software.
[0028] The classification of components in this specification is merely based on the primary function each component is responsible for. That is, two or more components may be combined into a single component, or a single component may be divided into two or more components based on more subdivided functions. Additionally, each component may perform some or all of the functions of other components in addition to its primary function, and some of the primary functions of each component may be exclusively performed by other components.
[0029] In describing the components in this specification, terms such as "first," "second," etc., may be used. These terms are intended for convenience of explanation to distinguish one component from another, and unless otherwise specifically stated, the nature, order, etc., of the components are not limited by these terms.
[0030] In each step mentioned in this specification, the steps may proceed differently from the specified order unless the context clearly indicates a specific order. That is, the steps may be performed in the same order as specified, substantially simultaneously, or in the reverse order.
[0031] In this specification, "and / or" may mean each of the listed components and a combination of two or more of the components. For example, "A, B and / or C" may be used with the same meaning as "at least one of A, B and C."
[0032] In this specification, "predetermined" may be a concept encompassing "pre-configured." For example, "pre-configured" may be performed by a provider, manager, operator, etc. of a device, system, or service.
[0033] In this specification, regarding "weighted summation" and "weighted average," the weights may be pre-set. For example, the weights may be pre-set by a provider, manager, operator, etc., of a device, system, or service.
[0034] In this specification, "multiple sclerosis (MS)" may be abbreviated as "MS", and "brain lesion" may be abbreviated as "lesion".
[0035] Figure 1 is a figure showing the results of analyzing the correlation between traditional brain imaging pathology indices (i.e., LeVol, Density, ClCo, GlEf, SmWo, Tr, and ChPaLe) obtained from MS patients, traditional neurological impairment assessment scales (i.e., EDSS), and white matter pathway density indices of brain lesion regions (i.e., Le-TDI) according to one embodiment of the present invention through graphic LASSO-based network analysis.
[0036] In Figure 1, variables are represented as nodes, and the relationships between variables are represented as edges. LeVol is a lesion characteristic indicator representing the volume of the lesion. Density, ClCo, GlEf, SmWo, Tr, and ChPaLe are brain connectivity pathology indicators that quantify the impact and state of the disruption of brain networks caused by the lesion, based on a lesion region mask and a normative connectome atlas. Density represents the density of the brain network, ClCo represents the clustering coefficient, GlEf represents global efficiency, SmWo represents small-worldness, Tr represents transitiviry, and ChPaLe represents characteristic path length. The Expanded Disability Status Scale (EDSS) is a traditional neurological disability assessment scale. Since each indicator and scale is a known concept, a detailed description is omitted in this specification.
[0037] According to the results of prior research, in the case of MS, there is no significant correlation between traditional brain imaging pathology indicators (e.g., brain structural pathology indicators such as lesion characteristic indicators, brain connectivity pathology indicators, etc.) and clinical symptoms (e.g., degree of disability, etc.). The inventors of the present invention have also confirmed through research that there is no direct significant correlation between traditional brain imaging pathology indicators and the degree of disability (e.g., EDSS) in MS.
[0038] However, according to the research results of the inventors of the present invention, regarding MS, the lesion-derived white matter tract density index (Le-TDI) of a brain lesion region according to one embodiment of the present invention was found to be capable of performing a bridging function that correlates the degree of disability with traditional brain imaging pathology indicators (see Fig. 1). That is, while it is difficult to reliably predict the degree of disability of MS patients using only traditional brain imaging pathology indicators, it was confirmed that the degree of disability of MS patients can be predicted more reliably by utilizing the Le-TDI according to one embodiment of the present invention.
[0039] Accordingly, according to exemplary embodiments of the present invention, a method and apparatus for predicting the degree of disability of an MS patient based on a white matter pathway indicator of a brain lesion region (e.g., a white matter pathway density indicator of a brain lesion region, i.e., Le-TDI, etc.) may be provided.
[0040] FIG. 2 illustrates a block diagram of a system (10) for predicting the degree of disability of an MS patient according to an embodiment of the present invention. For convenience of explanation, the system for predicting the degree of disability of an MS patient may be abbreviated as 'MS disability prediction system' and the service for predicting the degree of disability of an MS patient may be abbreviated as 'MS disability prediction service'.
[0041] Referring to FIG. 2, the MS failure prediction system (10) may include a user device (100) and a server device (200).
[0042] The user device (100) may be a device of a user who wishes to receive an MS failure prediction service. In one example, the user device (100) may be a device of a medical professional, medical institution, etc. who wishes to receive an MS failure prediction service to check the degree of disability of a subject (i.e., an MS patient, etc.). In another example, the user device (100) may be a device of a subject who wishes to receive an MS failure prediction service to check their own degree of disability directly. The user device (100) may interact with the server device (200) via a network, such as transmitting or receiving certain data. For example, the user device (100) may provide information related to the subject (e.g., information regarding the subject's brain images or brain lesion areas, etc.) to the server device (200) and receive a prediction result regarding the subject's degree of disability from the server device (200).
[0043] The user device (100) may be a computing device that can be implemented in various forms, such as a smartphone, laptop, desktop computer, or tablet PC. Although FIG. 1 illustrates only one user device (100), multiple user devices may connect to the server device (200) to use the MS failure prediction service.
[0044] The server device (200) may be a device of a service provider that provides MS failure prediction services. The server device (200) may interact with the user device (100), a third device (not shown), etc., by transmitting or receiving certain data through a network. The server device (200) may receive information related to the subject (e.g., information on the subject's brain images or brain lesion areas, etc.) through a network from the user device (100), the third device (not shown), etc., and may provide the prediction result regarding the subject's degree of disability to the user device (100). The server device (200) may store a rule-based model, a trained artificial intelligence model, etc., for predicting the subject's degree of disability. The server device (200) may provide the prediction result regarding the subject's degree of disability to the user device (100) in the form of a platform upon the request of the user device (100).
[0045] The server device (200) may be a computing device that can be implemented in various forms, such as a standard server, a server group, or a rack server system.
[0046] The relationship between the server device (200) and the user device (100) as illustrated in FIG. 2 can be established when the MS failure prediction service is provided in the form of a platform. According to another embodiment of the present invention, the MS failure prediction service may be provided in the form of a non-network-based program or application rather than a platform. In this case, unlike as illustrated in FIG. 2, the MS failure prediction service may be provided by a program or application installed on the user device (100) without a server device (200). In this case, the server device (200) may be the entity providing the program or application.
[0047] FIG. 3 illustrates the structure of a device according to an embodiment of the present invention. FIG. 3 illustrates an example of the structure of a user device (100) or a server device (200) of FIG. 2. In the description of FIG. 3, the user device (100) and the server device (200) may be collectively referred to as 'devices'.
[0048] Referring to FIG. 3, the device may include a control unit (301), a storage unit (302), and a communication unit (303).
[0049] The control unit (301) can control the overall functions and operations of the device. The control unit (301) can control the components of the device, for example, the storage unit (302), the communication unit (303), etc. That is, the control unit (301) can provide information or data necessary for the operation of the components of the device and can perform operations based on the information or data generated or managed by the components. The control unit (301) may include at least one processor, at least one circuit, etc. For example, the at least one processor may include at least one of a central processing unit (CPU), a graphics processing unit (GPU), and a neural network processing unit (NPU). The control unit (301) can perform the necessary control to operate the device according to various embodiments described in this specification. For example, the control unit (301) can control the operation of the device by executing software, programs, or instructions stored in the storage unit (302).
[0050] The storage unit (302) can store data used in the device and software, programs, commands, etc. for the operation of the device. The storage unit (302) can provide stored data under the control of the control unit (301). The storage unit (302) can store applications, drivers, etc. to be driven by the control unit (301). For example, the storage unit (302) may include RAM such as DRAM, SRAM, etc.; ROM; EEPROM; HDD; SSD; flash storage means, etc.
[0051] The communication unit (303) can perform the function of transmitting and receiving signals with other devices. The communication unit (303) performs wired communication or wireless communication and can process signals according to the control of the control unit (301). For example, the communication unit (303) may include an RF circuit, an antenna, etc. for wireless communication, or a connection terminal, a modem, a driver module, etc. for wired communication. For example, the communication unit (303) may support at least one of various communication protocols, such as cellular communication like LTE and 5G, short-range wireless communication like WiFi and Bluetooth, and short-range wired communication like Ethernet.
[0052] Although not illustrated in FIG. 3, the device may further include at least one of a power supply (e.g., battery, power supply, etc.), a display device (e.g., display, etc.), an input device (e.g., touch panel, button, etc.), and an output device, depending on the type of device.
[0053] FIGS. 4 to 6 are flowcharts illustrating a method for predicting the degree of disability of an MS patient according to an embodiment of the present invention. FIGS. 4 to 6 illustrate an example where an MS disability prediction service is provided in the form of a platform. That is, FIGS. 4 to 6 may represent operations performed by a server device (200). However, as previously described, the MS disability prediction service may be provided by a program or application installed on a user device (100). In this case, FIGS. 4 to 6 may represent the operation of the user device (100), and a person skilled in the art will clearly understand that some of the descriptions below may be implemented in a modified form without departing from the essential characteristics of the present invention.
[0054] Referring to FIG. 4, the server device (200) can obtain information about the brain lesion area of a subject (i.e., an MS patient, etc.) (e.g., S401).
[0055] The server device (200) can send a request for information provision to a device that provides information related to the subject (e.g., information about the subject's brain images or brain lesion areas, etc.) and receive information related to the subject (however, the sending of the request for information provision may be omitted). The server device (200) can receive information related to the subject from a user device (100), a third device (not shown; an MRI scanner, a computing device associated therewith, etc.). In one example, a dedicated platform or website designed for the exchange of information may be used. The server device (200) can provide an interface for uploading information related to the subject online and receive information related to the subject through said interface.
[0056] Information regarding the brain lesion region may include shape information, size information, location information, etc. of the brain lesion region. In one example, information regarding the brain lesion region may be obtained through an image in which the brain lesion region is marked, a brain lesion region mask, etc. That is, obtaining information regarding the brain lesion region may include obtaining an image in which the brain lesion region is marked, a brain lesion region mask, etc. For example, the brain lesion region may refer to an MS lesion region.
[0057] In one example, the server device (200) can acquire a subject's brain image (e.g., MRI, etc.) from a user device (100), a third device, etc. The server device (200) can generate a brain lesion region mask representing a brain lesion region based on the brain image using a trained artificial intelligence model (hereinafter referred to as a lesion region extraction model) (i.e., automated brain lesion segmentation). Since various lesion region extraction models are already known, a detailed description is omitted in this specification. In another example, the server device (200) can receive and acquire a brain lesion region mask generated by the user device (100), a third device, etc. In yet another example, the server device (200) can receive and acquire a brain lesion region mask generated by experts directly analyzing the subject's brain image (e.g., MRI, etc.) (i.e., manual brain lesion segmentation).
[0058] The server device (200) can map a brain lesion region to a white matter streamline map and identify the region corresponding to the brain lesion region in the white matter streamline map (e.g., S402).
[0059] A white matter streamline map may include a plurality of voxels. For example, a white matter streamline map may consist of a total of N voxels, from the first voxel to the Nth voxel (where N is a natural number). A voxel may have a predetermined volume. For example, a voxel may have a volume of l mm × w mm × h mm. l, w, and h may each be integers from 0.5 to 10. l, w, and h may be the same or different from each other.
[0060] A white matter streamline map may include information regarding the number of white matter streamlines passing through each voxel. The white matter streamline map may include information indicating the number of white matter streamlines passing through each voxel. For example, the white matter streamline map may display the number of streamlines passing through each voxel on a voxel-by-voxel basis. For example, the white matter streamline map may include, from the number of white matter streamlines passing through the first voxel C1 to the number of white matter streamlines passing through the Nth voxel C N Each voxel may contain information indicating the number of white matter streamlines passing through that voxel.
[0061] A white matter streamline may refer to a virtual line that visually represents the white matter tracts of the brain through diffusion MRI, tractography techniques, etc. Since the concepts of white matter streamlines and white matter tracts are already known, a detailed description is omitted in this specification.
[0062] In some examples, the white matter streamline map may be generated by averaging (e.g., arithmetic mean) multiple white matter streamline maps obtained from a normal group (i.e., MS healthy control group) on a voxel basis. That is, the number of white matter streamlines C passing through the M-th voxel of the white matter streamline map M≠
[0063] In some examples, mapping brain lesion regions to a white matter streamline map may include spatially aligning brain lesion regions to the white matter streamline map. In one example, the mapping may include marking brain lesion regions at locations on the white matter streamline map corresponding to location information for brain lesion regions. In another example, the mapping may include overlaying a brain lesion region mask on the same coordinate system as the white matter streamline map. In this case, preprocessing (e.g., resampling, spatial alignment, etc.) may be performed on the brain lesion region mask so that the brain lesion region mask and the white matter streamline map share the same coordinate system.
[0064] The server device (200) can identify voxels within the brain lesion region by mapping the brain lesion region to the white matter streamline map. In one example, the server device (200) can identify voxels within the region where the brain lesion region mask and the white matter streamline map overlap.
[0065] In a specific example, multiple registration can be performed between FLAIR MRI and T1-weighted MRI (T1w MRI), and between T1-weighted MRI and the Montreal Neuroscience Institute (MNI) standard template. Multiple registration can be performed using mutual information metrics based on ANTs software. After registration is completed, a lesion region can be extracted from FLAIR MRI, and the extracted lesion region can be mapped to the MNI template to generate a lesion region mask. By superimposing the lesion region mask onto the white matter streamline map, regions corresponding to the lesion region in the white matter streamline map can be identified. That is, voxels located within the lesion region can be identified.
[0066] The server device (200) can obtain the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region (e.g., S403).
[0067] Multiple voxels may be located within the brain lesion region. For example, within the brain lesion region, the first lesion voxel (referred to as a lesion voxel to distinguish it from the first voxel mentioned above) to the nth lesion voxel may be located (where n is a natural number). The server device (200) can obtain the total number of voxels (i.e., n) within the brain lesion region.
[0068] The white matter streamline map is from the number of white matter streamlines passing through the first lesion voxel L1 to the number of white matter streamlines passing through the nth lesion voxel L n It may include information indicating. The server device (200) indicates the number of white matter streamlines passing through each voxel within the brain lesion region (i.e., from L1 to L n You can obtain ).
[0069] The server device (200) can determine the white matter path indicator of the brain lesion region based on the number of voxels in the brain lesion region and the number of white matter streamlines passing through each voxel in the brain lesion region (e.g., S404).
[0070] For example, the server device (200) can calculate the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region using a predetermined mathematical formula (e.g., simple sum, weighted sum, simple average, weighted average, other functions, etc.) to calculate the white matter path index of the brain lesion region.
[0071] In some examples, the white matter pathway index of a brain lesion region may be determined based on the number of voxels within the brain lesion region; and the simple sum or weighted sum of the number of white matter streamlines passing through each voxel within the brain lesion region. For example, the white matter pathway index of a brain lesion region may be calculated by performing an operation on the number of voxels within the brain lesion region; and the simple sum or weighted sum of the number of white matter streamlines passing through each voxel within the brain lesion region; using a predetermined mathematical formula (e.g., multiplication, division, or other functions).
[0072] In some examples, the white matter tract index of the brain lesion region may include the lesion-derived white matter tract density index (Le-TDI). Le-TDI may be defined as the value obtained by dividing the simple sum or weighted sum of the number of white matter streamlines passing through each voxel within the brain lesion region by the number of voxels within the brain lesion region. That is, Le-TDI may be the arithmetic mean or weighted mean of the number of white matter streamlines passing through each voxel.
[0073] In one example, when a total of n voxels located within a brain lesion region are referred to as the first lesion voxel through the nth lesion voxel, Le-TDI can be calculated according to Equation 1.
[0074] [Equation 1]
[0075]
[0076] In Equation 1, n is the number of voxels within the brain lesion region, and L k is the number of white matter streamlines passing through the k-th lesion voxel. where k is a natural number and 1 ≤ k ≤ n.
[0077] In another example, when a total of n voxels located within a brain lesion region are defined as the first lesion voxel through the nth lesion voxel, Le-TDI can be calculated according to Equation 2.
[0078] [Equation 2]
[0079]
[0080] In Equation 2, n is the number of voxels within the brain lesion region, and L k is the number of white matter streamlines passing through the k-th lesion voxel, and w k is L k It is a pre-set value as a weight applied to. However, k is a natural number, and 1 ≤ k ≤ n. For example, the weight can be pre-set per voxel based on the position of the voxel.
[0081] The server device (200) can predict the degree of disability of a subject (i.e., an MS patient, etc.) based on white matter pathway indicators of the brain lesion region of the subject using a preset rule-based model or a trained artificial intelligence model (e.g., S405).
[0082] The degree of disability may refer to an indicator that quantifies the degree of disability (e.g., a score, etc.) or a categorized indicator (e.g., a grade, etc.). In one example, the degree of disability may be an indicator established based on a traditional disability assessment scale (e.g., EDSS as a traditional neurological disability assessment scale). In another example, the degree of disability may be an indicator established according to criteria set by experts based on specific data (e.g., test results, responses to a survey, etc.).
[0083] The pre-configured rule-based model may include a regression model or a classification model depending on the type of disability degree. For example, the pre-configured rule-based model can take white matter pathway indicators of a brain lesion region as input and output the disability degree.
[0084] The trained AI model may include a regression model or a classification model depending on the type of disability degree. The trained AI model may be a model trained to predict the disability degree based on white matter pathway indicators in brain lesion regions. For example, the trained AI model may be trained using white matter pathway indicators in brain lesion regions with labeled disability degrees as a training dataset. For example, the trained AI model may be constructed by fine-tuning a known AI model as a backbone.
[0085] As illustrated in FIG. 5, according to another embodiment of the present invention, the degree of disability of a subject (i.e., an MS patient, etc.) can be predicted.
[0086] Referring to FIG. 5, the server device (200) can obtain a white matter pathway indicator of the subject's brain lesion region (e.g., S501).
[0087] White matter pathway indicators of the brain lesion region can be obtained according to the aforementioned S401, S402, S403, and S404.
[0088] The server device (200) can acquire at least one of the subject's brain connectivity pathology indicator and brain lesion characteristic indicator (e.g., S502).
[0089] Brain connectivity pathology markers may refer to indicators that quantify the impact of brain lesions on brain networks and the state of brain networks resulting from brain lesions.
[0090] In some examples, brain connectivity pathology indicators may include traditional brain connectivity pathology indicators. For example, brain connectivity pathology indicators may include lesion-derived network metrics. For example, brain connectivity pathology indicators may include brain network density, clustering coefficient, global efficiency, small-worldness, transitiviry, characteristic path length, etc. Since each indicator is a known concept, a specific description is omitted in this specification.
[0091] In one example, the server device (200) can receive and obtain the subject's brain connectivity pathology indicators from the user device (100), a third device, etc.
[0092] In another example, the server device (200) can obtain brain connectivity pathology indicators using a standard network atlas based on the subject's brain image (e.g., MRI, etc.), information about brain lesion regions, etc.
[0093] In a specific example, the server device (200) maps the lesion to a standard network atlas using a lesion region mask, etc., to estimate the disconnection of the brain network caused by the lesion, and can analyze the impact of the disconnection caused by the lesion on the brain network using a lesion quantification toolkit. The lesion quantification toolkit can extract a matrix of disconnections caused by the lesion and topological metrics. The topological metrics can be used as pathological indicators of brain connectivity. The topological metrics may include the aforementioned brain network density, clustering coefficient, global efficiency, micro-worldness, transferability, characteristic path length, etc.
[0094] Brain lesion characteristic indicators may refer to indicators representing the structural characteristics, spatial characteristics, etc., of the brain lesion itself.
[0095] In some examples, brain lesion characteristic indicators may include the volume, size, etc. of the brain lesion. For example, a server device (200) may obtain brain lesion characteristic indicators based on a subject's brain image (e.g., MRI, etc.).
[0096] In one example, the server device (200) can obtain the volume of the brain lesion by multiplying the number of voxels within the brain lesion region obtained in the aforementioned S403 by the volume of the voxels.
[0097] The server device (200) can predict the degree of impairment based on at least one of a brain connectivity pathology indicator and a brain lesion characteristic indicator, and a white matter pathway indicator of the brain lesion region using a preset rule-based model or a trained artificial intelligence model (e.g., S503).
[0098] The pre-configured rule-based model may include a regression model or a classification model depending on the type of disability degree. For example, the rule-based model may take at least one of a brain connectivity pathology indicator and a brain lesion characteristic indicator, along with a white matter pathway indicator of the brain lesion region, as input and output a disability degree.
[0099] The trained artificial intelligence model may include a regression model or a classification model depending on the type of disability degree. The trained artificial intelligence model may be a model trained to predict the disability degree based on at least one of a brain connectivity pathology indicator and a brain lesion characteristic indicator, and a white matter pathway indicator of the brain lesion region. For example, the trained artificial intelligence model may be trained using at least one of a brain connectivity pathology indicator and a brain lesion characteristic indicator, a white matter pathway indicator of the brain lesion region, and the disability degree as a label as a training dataset.
[0100] As illustrated in FIG. 6, the degree of disability of a subject (i.e., an MS patient, etc.) can be predicted according to another embodiment of the present invention. In the following description of FIG. 6, content that overlaps with the previously mentioned content may be omitted.
[0101] Referring to FIG. 6, the server device (200) can obtain the subject's brain connectivity pathology indicator (e.g., S601).
[0102] The server device (200) can predict the white matter pathway indicator of the subject's brain lesion region based on the subject's brain connectivity pathology indicator using a preset first rule-based model or a first trained artificial intelligence model (e.g., S602).
[0103] In some examples, brain connectivity pathology indicators may include at least one, two, or three of the aforementioned brain network density, clustering coefficient, global efficiency, micro-worldliness, metastaticity, and characteristic path length.
[0104] The pre-configured first rule-based model may include a regression model. For example, the first rule-based model may take a brain connectivity pathology indicator as input and output a white matter pathway indicator of a brain lesion region.
[0105] The first trained artificial intelligence model may include a regression model. The first trained artificial intelligence model may be a model trained to predict white matter pathway indicators of a brain lesion region based on brain connectivity pathology indicators. For example, the first trained artificial intelligence model may be trained using white matter pathway indicators of a brain lesion region as a training dataset, as brain connectivity pathology indicators and labels.
[0106] The server device (200) can predict the degree of disability of a subject based on white matter path indicators of the subject's brain lesion region using a pre-set second rule-based model or a second trained artificial intelligence model (e.g., S603).
[0107] The pre-configured second rule-based model may include a regression model or a classification model depending on the type of disability degree. For example, the pre-configured second rule-based model may take white matter pathway indicators of a brain lesion region as input and output a disability degree.
[0108] The second trained artificial intelligence model may include a regression model or a classification model depending on the type of disability degree. The second trained artificial intelligence model may be a model trained to predict the disability degree based on white matter pathway indicators of brain lesion regions. For example, the second trained artificial intelligence model may be trained using white matter pathway indicators of brain lesion regions labeled with the disability degree as a training dataset.
[0109] In some examples, in S602, the server device (200) can predict a white matter pathway indicator of a brain lesion region based on a first brain connectivity pathology indicator including at least one of the acquired brain connectivity pathology indicators by using a preset first rule-based model or a first trained artificial intelligence model. In S603, the server device (200) can predict the degree of impairment based on a second brain connectivity pathology indicator including at least one of the acquired brain connectivity pathology indicators and a white matter pathway indicator of a brain lesion region by using a preset second rule-based model or a second trained artificial intelligence model. The first brain connectivity pathology indicator and the second brain connectivity pathology indicator may be the same or different, and if there are multiple indicators, some of them may be the same.
[0110] In this case, the pre-configured first rule-based model and second rule-based model may be redesigned as the target input data differs; since this can be clearly understood by referring to the foregoing, a redundant explanation is omitted. Additionally, each of the first trained artificial intelligence model and the second trained artificial intelligence model may be trained using a training data set consisting of target input data and target output data (i.e., labels). Since this can be clearly understood by referring to the foregoing, a redundant explanation is omitted.
[0111] In some other examples, at S601, the server device (200) may further acquire the subject's brain lesion characteristic indicators. At S602, the server device (200) may predict the white matter pathway indicators of the brain lesion region based on the brain connectivity pathology indicators and the brain lesion characteristic indicators using a pre-configured first rule-based model or a first trained artificial intelligence model. For example, the server device (200) may predict the white matter pathway indicators of the brain lesion region based on a first brain connectivity pathology indicator including at least one of the acquired brain connectivity pathology indicators and a first brain lesion characteristic indicator including at least one of the acquired brain lesion characteristic indicators. At S603, the server device (200) may predict the degree of impairment based on the brain connectivity pathology indicators, the brain lesion characteristic indicators, and the white matter pathway indicators of the brain lesion region using a pre-configured second rule-based model or a second trained artificial intelligence model. For example, the server device (200) can predict the degree of impairment based on a second brain connectivity pathology indicator including at least one of the acquired brain connectivity pathology indicators, a second brain lesion characteristic indicator including at least one of the acquired brain lesion characteristic indicators, and a white matter pathway indicator of the brain lesion region. The first brain connectivity pathology indicator and the second brain connectivity pathology indicator may be the same or different, and if there are multiple indicators, some of them may be the same. The first brain lesion characteristic indicator and the second lesion characteristic indicator may be the same or different, and if there are multiple indicators, some of them may be the same.
[0112] In this case, the pre-configured first rule-based model and second rule-based model may be redesigned as the target input data differs; since this can be clearly understood by referring to the foregoing, a redundant explanation is omitted. Additionally, each of the first trained artificial intelligence model and the second trained artificial intelligence model may be trained using a training data set consisting of target input data and target output data (i.e., labels). Since this can be clearly understood by referring to the foregoing, a redundant explanation is omitted.
[0113] In this specification, the designation of 'one embodiment' of the principles of the present invention and various variations of such expression means that specific features, structures, characteristics, etc., associated with this embodiment are included in at least one embodiment of the principles of the present invention. Accordingly, the expression 'in one embodiment' and any other variations disclosed throughout this specification do not necessarily refer to the same embodiment.
[0114] The implementation of the device and system according to the various embodiments of the present invention described above may be realized with digital electronic circuits, integrated circuits, ASICs (application specific integrated circuits), hardware, firmware, software, or a combination thereof.
[0115] The method according to the various embodiments of the present invention described above may be implemented as a computer program or mobile application and stored on a medium so as to be executed in combination with hardware. The steps of the method or algorithm described in relation to the embodiments of the present invention may be implemented directly in hardware, implemented as a software module executed by hardware, or implemented by a combination thereof. The software module may reside in RAM, ROM, EPROM, EEPROM, flash memory, hard disk, removable disk, CD-ROM, or any form of computer-readable recording medium well known in the art to which the present invention belongs. Additionally, the algorithm may be produced in the form of an installation file and provided in the form of an online download, and for this purpose, may be stored on a server accessible through an online software market.
[0116] All embodiments and conditional examples disclosed herein are intended to help those skilled in the art to understand the principles and concepts of the invention. Those skilled in the art will understand that the invention may be implemented in modified forms without departing from the essential nature of the invention. Therefore, the disclosed embodiments should be considered in an illustrative rather than a limiting sense. The scope of the invention is defined by the claims, not by the foregoing description, and all variations within the scope of the claims should be interpreted as being included in the invention.
Claims
1. A method executed by at least one processor, A step of obtaining information about brain lesion regions in a multiple sclerosis patient; A step of identifying voxels within the brain lesion region in the white matter streamline map by mapping the brain lesion region to a white matter streamline map that includes a plurality of voxels and contains information on the number of white matter streamlines passing through each voxel; A step of obtaining the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region; A step of determining a white matter pathway index of a brain lesion region based on the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region; and A step of predicting the degree of disability of a multiple sclerosis patient based on white matter pathway indicators of the brain lesion region using a rule-based model or a trained artificial intelligence model; comprising A method for predicting the degree of disability in patients with multiple sclerosis.
2. In Claim 1, The step of obtaining information about the brain lesion region includes obtaining a brain lesion region mask representing the brain lesion region of the multiple sclerosis patient, and Mapping the brain lesion region to the white matter streamline map comprises overlapping the brain lesion region mask with the white matter streamline map. A method for predicting the degree of disability in patients with multiple sclerosis.
3. In Claim 1, The above white matter streamline map is obtained by averaging the white matter streamline map of the normal group in voxel units, A method for predicting the degree of disability in patients with multiple sclerosis.
4. In Claim 1, The white matter pathway index of the brain lesion region is determined based on the number of voxels within the brain lesion region; and the simple sum or weighted sum of the number of white matter streamlines passing through each voxel within the brain lesion region. A method for predicting the degree of disability in patients with multiple sclerosis.
5. In Claim 1, The white matter pathway indicator of the brain lesion region includes a white matter pathway density indicator of the brain lesion region. The white matter pathway density index of the brain lesion region is defined as the value obtained by dividing the simple sum or weighted sum of the number of white matter streamlines passing through each voxel within the brain lesion region by the number of voxels within the brain lesion region. A method for predicting the degree of disability in patients with multiple sclerosis.
6. In Claim 1, The above degree of disability includes an indicator that quantifies or categorizes the degree of neurological disability. A method for predicting the degree of disability in patients with multiple sclerosis.
7. In Claim 1, The method further comprises the step of obtaining at least one of the brain connectivity pathology indicator and brain lesion characteristic indicator of the multiple sclerosis patient; and The step of predicting the degree of disability of the multiple sclerosis patient comprises predicting the degree of disability of the multiple sclerosis patient based on: a white matter pathway indicator of the brain lesion region; and at least one of the brain connectivity pathology indicator and the brain lesion characteristic indicator. A method for predicting the degree of disability in patients with multiple sclerosis.
8. In Claim 7, The above-mentioned brain connectivity pathology indicator is an indicator that quantifies the effect of a brain lesion on a brain network or the state of a brain network due to a brain lesion, and includes at least one of the brain network density, clustering coefficient, global efficiency, micro-worldliness, metastaticity, and characteristic path length. The above brain lesion characteristic indicator includes the volume of the brain lesion, A method for predicting the degree of disability in patients with multiple sclerosis.
9. At least one memory; and It includes at least one processor that executes instructions stored in at least one memory; and The above-mentioned at least one processor is, Acquire information on brain lesion regions in multiple sclerosis patients, and A white matter streamline map comprising a plurality of voxels and containing information on the number of white matter streamlines passing through each voxel is mapped to the brain lesion region to identify the voxels within the brain lesion region in the white matter streamline map, and Obtain the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region, and Determine the white matter pathway index of the brain lesion region based on the number of voxels within the brain lesion region and the number of white matter streamlines passing through each voxel within the brain lesion region, and Predicting the degree of disability of a multiple sclerosis patient based on white matter pathway indicators of the brain lesion region using a rule-based model or a trained artificial intelligence model, A device for predicting the degree of disability in patients with multiple sclerosis.
10. An application program stored on a recording medium to execute the method according to claim 1 when operated by at least one processor.