Networking method for nervous system degenerative diseases

Abstract

A method of determining a patient's response to a neuropharmacological intervention, a method of determining a likelihood of a patient having one or more neurological disorders, and a system for use in the methods are disclosed. The method of determining the likelihood of a patient suffering from one or more neurological disorders comprises the steps of: obtaining data indicative of electrical activity within the brain of the patient; generating a network based at least in part on the obtained data, the network comprising a plurality of nodes and directional connections between the nodes, where the network indicates a flow of electrical activity within the brain of the patient; calculating, for each node, a difference between the number and / or strength of connections entering the node and the number and / or strength of connections exiting the node; and determining a likelihood that the patient suffers from one or more neurological disorders using the calculated differences.

Description

[0001] This application is a divisional application of the patent application with Chinese Application No. 201880099123.1, invention title "Network method for neurodegenerative diseases", and application date of September 5, 2018 (PCT Application No. PCT / EP2018 / 073906). Technical Field

[0002] The present invention relates to the application of network methods in the study of neurocognitive disorders. Background Art

[0003] Brain models as complex networks of interconnected subunits have improved the understanding of normal brain tissue and have enabled the resolution of functional changes in neurological disorders. These subunits form so-called brain modules, i.e., groups of regions with high connection density within them and lower connection density between the groups. It has been proposed that the modular organization of the brain supports the effective integration between spatially segregated neural processes, thus supporting various cognitive and behavioral functions. Changes in the brain network can help identify patients with Alzheimer's disease (AD) and behavioral variant frontotemporal dementia (bvFTD).

[0004] For example, alterations in regional volume have been identified in schizophrenia patients through studies of structural networks in health and disease, where pairwise correlations of cortical regional volume or thickness, such as in vivo measurements derived from T1-weighted magnetic resonance imaging (MRI), have been examined. This method has shown clinical relevance by revealing alterations in regional volume in schizophrenia patients. However, volume metrics representing the product of cortical thickness (CT) and surface area (SA) may confound underlying differences. For example, consideration of changes in cortical thickness can provide insights into how the disease alters the size, density, and arrangement of cells within the cortical layer. On the other hand, changes in surface area can provide information about perturbations in the functional integration between column groups in the diseased brain.

[0005] Previously, whole-brain or lobar volumes have been used to monitor the effects of neuropharmacological interventions. However, this is a relatively crude level of analysis.

[0006] As another example of the use of networks in the understanding of the brain, as discussed in WO 2017 / 118733 (the entire content of which is incorporated herein by reference), EEG data collected from patients can be used to detect the intensity and directionality of electrical currents within the brain.

[0007] A poster entitled "Organisation of cortical thickness networks in Alzheimer's disease and behavioural variant frontotemporal dementia across brain lobes" was presented by Vuksanovic et al. at the 6th Cambridge University Neuroscience Symposium, Neural Networks in Health and Disease, on 7 - 8 September 2017.

[0008] Another poster entitled "Divergent changes in structural correlation networks in Alzheimer's disease and behavioural variant frontotemporal dementia" was presented by Vuksanovic et al. at the ARUK Conference in London, UK, on 20 - 21 March 2018.

[0009] A presentation entitled "Modular organisation of cortical thickness and surface area structural correlation networks in Alzheimer's disease (AD) and behavioural variant frontotemporal dementia (bvFTD)" was given by Vuksanovic, V at the 10th SINAPSE Annual Scientific Meeting in Edinburgh on 25 June 2018. SUMMARY OF THE INVENTION

[0010] In a first aspect, the present invention provides a method for determining a patient's response to a neuropharmacological intervention, the method comprising the steps of:

[0011] Obtaining structural neurology data from a plurality of patients prior to the neuropharmacological intervention, the structural neurology data indicating the physical structure of a plurality of cortical regions;

[0012] Generating a first correlation matrix from the structural neurology data by:

[0013] Assign a plurality of structural nodes corresponding to cerebral cortical regions; and

[0014] Determine pairwise correlations between the structural nodes at least in part based on corresponding data of the structural neurological data;

[0015] Obtain further structural neurological data from the plurality of patients after a neuropharmacological intervention, the further structural neurological data indicating the physical structure of the plurality of cortical regions; and

[0016] Generate a second correlation matrix from the further structural neurological data by:

[0017] Determine pairwise correlations between the structural nodes at least in part based on corresponding data of the further structural neurological data;

[0018] The method includes:

[0019] Compare the first correlation matrix with the second correlation matrix to thereby determine the patient's response to the neuropharmacological intervention.

[0020] Optional features of the invention will now be set forth. These are applicable either individually or in any combination with any aspect of the invention.

[0021] The correlation matrix can mean generating a structural correlation network, which can then be represented by a matrix.

[0022] In one embodiment, the patient response can be in the context of a clinical trial, such as for evaluating the efficacy of a drug in the treatment of a neurocognitive disorder. Thus, the group of patients (the plurality of patients) can be a treatment group that has been diagnosed with the disorder, or can be a control ("normal") group. Finally, the efficacy of the drug can be evaluated in whole or in part based on the patient group response determined according to the invention, optionally compared with a comparison group that has not received the intervention.

[0023] The measured or obtained physical structure can be cortical thickness and / or surface area. The value of cortical thickness and / or surface area can be the average value obtained from the structural neurological data. The structured neurological data can be collected from magnetic resonance imaging (MRI) data or computed tomography data of each patient. The structural neurological data and the further structural neurological data are obtained at different time points. As discussed herein, the structural neurological data can be obtained by magnetic resonance imaging, computed tomography, or positron emission tomography of each patient. These techniques are well known to those skilled in the art themselves - see, for example, Mangrum, Wells et al Duke Review of MRI Principles: Case Review Series E-book. Elsevier Health Sciences, 2018; and "Standardized low-resolution electromagnetic tomography (sLORETA): technical details" Methods Find Exp. Clin. Pharmacol. 2002:24 Suppl D:5-12; Pascual-Marqui RD et al.

[0024] The plurality of cortical regions can be at least 60 or at least 65. For example, 68. The cortical regions can be, for example, those provided by the Desikan-Killiany atlas (Desikan et al 2006).

[0025] The p-value can be determined for each pairwise correlation in multiple subjects and compared with a significance level, where only the p-values less than the significance level are used to generate the corresponding correlation matrix. When determining the pairwise correlation between structural nodes, the corresponding value of each structural node can be compared with a reference value, and its covariance can be determined. The significance level can be referred to as alpha ("α").

[0026] Comparing the first correlation matrix with the second correlation matrix can include comparing the number and / or density of inverse correlations in the first correlation matrix with the number and / or density of inverse correlations in the second correlation matrix. When making the comparison, groups of structural nodes corresponding to the same lobe can be identified, and the comparison made between the first correlation matrix and the second correlation matrix can utilize the same lobe.

[0027] Allocating the plurality of structural nodes corresponding to the cerebral cortical regions may also include defining groups containing structural nodes corresponding to homologous or non-homologous brain lobes. Comparing the first correlation matrix with the second correlation matrix may include comparing the number and / or density of correlations between different groups of structural nodes. In other words, comparing the first correlation matrix with the second correlation matrix may include comparing the correlations between pairs of non-homologous structural nodes.

[0028] In some examples, comparing the first correlation matrix with the second correlation matrix may include comparing the number and / or density of correlations between groups of structural nodes located in the frontal lobe (front nodes) and the parietal and occipital lobes (rear nodes), respectively. It has been found that in examples of effective neuropharmacological intervention, the number and / or density of inverse correlations between the front nodes and the rear nodes is reduced. Since it is assumed that inverse correlations indicate compensatory linkage formation, whereby atrophy in one node is associated with hypertrophy in functionally linked nodes, it should be understood that a reduction in the number and / or density of inverse correlations indicates a decrease in the number of compensatory linkages.

[0029] ***

[0030] Generally, the neurocognitive disease or cognitive disorder is a neurodegenerative disorder causing dementia, such as a tauopathy.

[0031] The patient may have been diagnosed with a neurocognitive disease, such as Alzheimer's disease or behavioral variant frontotemporal dementia. The disease may be mild or moderate Alzheimer's disease. The disease may be mild cognitive impairment. However, the findings of the inventors described herein are also applicable to other neurocognitive diseases.

[0032] Diagnostic criteria and treatments for tauopathies and other neurocognitive diseases are known in the art and are discussed, for example, in WO 2018 / 019823 and the references cited therein.

[0033] The disease may be behavioral variant frontotemporal dementia (bvFTD). Diagnostic criteria and treatments for bvFTD are discussed, for example, in WO 2018 / 041739 and the references cited therein.

[0034] As explained herein, the topology of perturbations in the structural network is different in these two pathologies (AD and bvFTD), and both are different from normal aging. It is global with respect to normal changes and is not limited to the frontotemporal and temporoparietal lobes in bvFTD and AD, respectively, and indicates an increase in both global correlation strength and the connectivity between non-homologous brain lobes defined by inverse correlations.

[0035] These changes appear to be adaptive, reflecting a coordinated increase in cortical thickness and surface area that compensates for corresponding damage in functionally linked nodes. This effect is more pronounced in the cortical thickness network in bvFTD and the surface area network in AD.

[0036] The inventors have observed that an important change differentiating the two forms of dementia from normal elderly controls is the emergence of a significant inverse correlation network between the forebrain and hindbrain regions that may be related to functional adaptation or compensation for damage caused by the pathology. Specifically, assuming that inverse correlation indicates compensatory linkage formation, whereby atrophy in one node is associated with hypertrophy in functionally linked nodes, it should be understood that a decrease in the number and / or density of inverse correlations indicates a reduction in the number of compensatory linkages.

[0037] Thus, if a neuropharmacological intervention is effective, it is expected to restore network organization to the state observed when using a normal (non-diseased) comparison group. If the disease is treated at an early enough stage, network organization can be restored to a state identical to that of normal controls. The method thus provides an objective means of differentiating disease-modifying treatment from symptomatic treatment: Symptomatic treatment focuses on abnormal network architecture and may actually focus on, for example, the risk of prion-like disease processes spreading to healthy brain regions. In contrast, disease-modifying drugs act in the opposite direction by normalizing function in regions affected by the pathology, reducing the need for compensatory input from relatively less damaged brain regions.

[0038] It will be appreciated from the disclosure herein that analysis of structure or network organization has particular utility in providing greater power in clinical trials, thereby allowing the use of fewer subjects and shorter treatment times. In particular, in diseases such as mild AD, mild cognitive impairment, and pre-mild cognitive impairment, clinical trial endpoints (cognition and function) may be relatively insensitive, thus requiring large numbers of subjects and / or longer time periods (see WO 2009 / 060191).

[0039] Thus, generally, the neuropharmacological intervention will be a pharmaceutical intervention.

[0040] The neuropharmacological intervention can be symptomatic treatment. Such compounds include acetylcholinesterase inhibitors (AChEIs) - these include tacrine, donepezil, rivastigmine, and galantamine. Another symptomatic treatment is memantine. These treatments are described in WO 2018 / 041739.

[0041] As explained above, the inventors have found an increase in compensatory networks (the number and / or density of non-homologous inverse correlations present in patient groups that have received such treatment).

[0042] The neuropharmacological intervention can be a disease-modifying drug rather than a symptomatic drug. These treatments can be differentiated, for example, based on what happens when a patient withdraws from active treatment. After an initial treatment period, a symptomatic agent postpones the symptoms of the disease without affecting the underlying disease process and does not change (or at least does not improve) the rate of longer-term decline. If, after withdrawal, the patient reverts to the state they would have been in without treatment, then the treatment is considered symptomatic (Cummings, J.L. (2006) Challenges to demonstrating disease-modifying effects in Alzheimer's disease clinical trials. Alzheimer's and Dementia, 2:263-271).

[0043] For example, a disease-modifying treatment can be an inhibitor of pathological protein aggregation, such as a 3,7-diaminophenothiazine (DAPTZ) compound. Such compounds are described in WO 2018 / 041739, WO 2007 / 110627, and WO 2012 / 107706. The latter describes colorless bis(hydrogen methanesulfonate) methylene blue, also known as leucomethane sulfonate methylene blue (LMTM; USAN name: hydrogentrimethylthioninium methanesulfonate).

[0044]

[0045] All of the content of these WO publications regarding the DAPTZ compounds they define is expressly incorporated by cross-reference.

[0046] Treatment with LMTM has been shown to reduce compensatory network-related (especially non-homologous forward and inverse correlations).

[0047] The neuropharmacological intervention can be a disease-modifying drug, and efficacy can be established by a decrease in the number and / or density of correlations between the forebrain regions and the hindbrain regions of the first correlation matrix and the second correlation matrix.

[0048] Therefore, it can be concluded that, in the case of an effective neuropharmacological intervention (such as a disease-modifying treatment), the number and / or density of inverse correlations between the anterior and posterior nodes decreases.

[0049] The present invention can also be used to identify functional adaptation or compensation for damage due to a pathology in a patient population, for example to study "cognitive reserve". The present invention can be used in combination with conventional diagnostic or prognostic measurements. These measurements include the Alzheimer's Disease Assessment Scale - Cognitive Subscale (ADAS-Cog); the National Institute of Neurological and Communicative Disorders and Stroke - Alzheimer's Disease and Related Disorders Association (NINCDS-ADRDA); the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV); and the Clinical Dementia Rating (CDR) scale.

[0050] As explained above, the method for determining a patient's response to a neuropharmacological intervention can in turn be used to evaluate different patient cohorts in a clinical trial of the neuropharmacological intervention. For example, the method can be used to determine the effectiveness of a neuropharmacological intervention in a group of patients. The method can be used to define patient groups based on their patient response (e.g., based on the determined correlation / anti-correlation). The patient groups can be identified with respect to their previous use of the neuropharmacological intervention and optionally selected for further treatment suitable for the patient response.

[0051] ***

[0052] In a second aspect, the present invention provides a method for determining the likelihood that a patient has one or more neurological disorders, the method comprising the steps of:

[0053] obtaining data indicative of electrical activity in the brain of the patient;

[0054] generating at least in part based on the obtained data a network comprising a plurality of nodes and directed connections between the nodes, wherein the network indicates the flow of electrical activity in the brain of the patient;

[0055] calculating for each node the difference between the number and / or strength of the connections entering the node and the number and / or strength of the connections leaving the node; and

[0056] using the calculated differences to determine the likelihood that the patient has one or more neurological disorders.

[0057] The inventors have shown that even very short use (e.g., analysis of brain EEG) can be used to potentially identify patients who are susceptible to one or more neurocognitive diseases (e.g., AD). Specifically, such individuals (patients or subjects, the terms being used interchangeably) may have a relatively large number of "sinks" or relatively strong sinks in the posterior lobe, and a relatively large number of "sources" or relatively strong sources in the temporal and / or frontal lobes. In ostensibly normal or preclinical subjects, in a preferred embodiment, the method can be more sensitive than psychometric tests commonly used to determine such risk.

[0058] ***

[0059] Optional features of the invention will now be set forth. These are applicable either individually or in any combination with any aspect of the invention.

[0060] The likelihood that a patient has one or more neurological disorders can be referred to as the patient's susceptibility to one or more neurological disorders. The method can include the step of defining the state of each node, whereby the node is defined as a sink or a source based on the calculated difference.

[0061] The network can be a renormalized partial directed coherence network. Any step of the method can be performed offline, i.e., without relying on the patient. For example, obtaining the data can be done by receiving previously recorded data from the patient via the network.

[0062] The data indicative of electrical activity in the brain can be electroencephalography data. The electroencephalography data can be beta-band electroencephalography data. The data indicative of electrical activity in the brain can also be magnetoencephalography data or functional magnetic resonance imaging data.

[0063] Determining the patient's susceptibility can be performed using a machine learning classifier. For example, a Markov model, a support vector machine, a random forest, or a neural network.

[0064] The method can include the step of generating a heat map at least in part based on the state of the nodes, the heat map indicating the location and / or intensity of the nodes defined as sinks and the nodes defined as sources in the patient's brain. Such a representation of the defined nodes can assist (e.g., ergonomically) in determining the patient's susceptibility.

[0065] When determining the patient's susceptibility, a comparison can be made between the number and / or intensity of sources in the parietal and / or occipital lobes and the number and / or intensity of sinks in the frontal and / or temporal lobes. It has been experimentally observed that patients susceptible to one or more neurodegenerative diseases (and particularly Alzheimer's disease) have relatively higher intensity sinks in the posterior lobe and relatively higher intensity sources in the temporal and / or frontal lobes.

[0066] The method may further include the step of using the states of the nodes to derive an indication of the degree of left-right asymmetry of the position and / or intensity of the nodes in the brain corresponding to the sinks and sources.

[0067] The neurological disorder may be a neurocognitive disease, which may be Alzheimer's disease.

[0068] The susceptibility of the patient to one or more neurological disorders can be determined by comparing the number and / or intensity of the nodes defined as sinks in the posterior lobe with a predetermined value, and / or by comparing the number and / or intensity of the nodes defined as sources in the temporal lobe and / or frontal lobe with a predetermined value. If the number and / or intensity of the nodes defined as sinks in the posterior lobe exceed the predetermined value and / or if the number and / or intensity of the nodes defined as sources in the temporal lobe and / or frontal lobe exceed the predetermined value (e.g., based on "control" subjects or subjects established to have a low risk, or reference data obtained therefrom (such as historical reference data)), then the patient can be determined to have a high susceptibility risk. In other words, and more generally, the determination of susceptibility can be based on whether the patient has more and / or stronger sources and / or sinks in one region of the brain relative to another region of the brain. For example, based on data from control subjects, if there are more and / or stronger sources in the temporal lobe and / or frontal lobe than expected, and / or if the patient has more and / or stronger sinks in the posterior lobe than expected, then the patient can be determined to be at risk of a neurodegenerative disorder. Such data from control subjects can have been established by longitudinal monitoring after a baseline assessment.

[0069] The inventors have further observed that, compared to the non-medicated group, one or more symptomatic treatments increase the activity exiting the frontal lobe.

[0070] The method according to this aspect of the invention can be used for assessing, testing or classifying the susceptibility of a subject to one or more neurological disorders for any purpose. For example, the score or other output of the test can be used to classify the mental state or disease state of the subject according to predetermined criteria.

[0071] ***

[0072] The subject can be any human subject. In one embodiment, the subject can be a subject suspected of having a neurocognitive disease or disorder as described herein (such as a neurodegenerative or vascular disease), or can be a subject not identified as being at risk.

[0073] In one embodiment, the method is for the purpose of early diagnosis or prediction of cognitive impairment (such as a neurocognitive disease) in the subject.

[0074] The disease can be mild to moderate Alzheimer's disease.

[0075] The disease can be mild cognitive impairment.

[0076] However, the discoveries of the inventors described herein are also applicable to other neurocognitive diseases. For example, the disease can be different dementias, such as vascular dementia.

[0077] The method can optionally be used to inform the subject of further diagnostic steps or interventions - for example, other methods based on imaging or invasive or non-invasive biomarker assessments, which are known in the art per se.

[0078] In some embodiments, the method can be used for the purpose of determining the risk of neurocognitive impairment in the subject. Optionally, the risk can additionally be calculated using other factors, such as age, lifestyle factors, and other measured physical or mental criteria. The risk can be classified as "high" or "low", or can be presented as a scale or spectrum.

[0079] ***

[0080] It will be clear from the disclosure herein that, in addition to assessing the likelihood of developing one or more neurological disorders, the method can also be used to assess the efficacy of disease-improving treatments in reducing the risk and / or treating the disease, i.e., to assess the efficacy of a drug in preventing or treating the disease or disorder. This can optionally be in the context of a clinical trial as described herein, for example, compared to a placebo or other normal control.

[0081] Specifically, the disclosure herein shows that the method of the invention (e.g., based on EEG technology) can provide an effective and sensitive measurement of the impact of the disease on the subject. This provides the opportunity to confirm the efficacy of disease-improving treatments in smaller subject groups (e.g., less than or equal to 200, 150, 100, or 50 in the treatment and comparison groups), over shorter intervals (e.g., less than or equal to 6, 5, 4, or 3 months), and in earlier or less severe diseases (e.g., prodromal AD, MCI, or even pre-MCI) compared to what may be achievable using currently available methods.

[0082] Thus, as discussed above, the method can be used in different patient cohorts in a clinical trial of neuropharmacological intervention. For example, the patient group (multiple patients) can be a treatment group that has been diagnosed with the disease (e.g., early disease) and is being treated with a putative disease-improving treatment, and a group treated with a placebo.

[0083] Accordingly, in another aspect, the method steps of the second aspect are for determining the disease state or severity of a patient, rather than for determining the likelihood that a patient has one or more neurological disorders. This state can in turn be monitored as part of clinical management or a clinical trial.

[0084] Accordingly, another aspect of the present invention provides a method for determining a patient's response to a neuropharmacological intervention for a neurological disorder, the method comprising the following steps prior to the neuropharmacological intervention:

[0085] (a) obtaining data indicative of the electrical activity in the brain of the patient;

[0086] (b) generating a network at least in part based on the obtained data, the network comprising a plurality of nodes and directed connections between the nodes, wherein the network indicates the flow of electrical activity in the brain of the patient;

[0087] (c) calculating, for each node, the difference between the number and / or strength of the connections entering the node and the number and / or strength of the connections leaving the node; and

[0088] (d) using the calculated difference to determine the patient's state with respect to the neurological disorder;

[0089] (e) repeating steps (a)-(d) after the neuropharmacological intervention to determine a further state of the patient with respect to the neurological disorder; and

[0090] (f) determining the patient's response to the neuropharmacological intervention based on the first state and the second state (e.g., by comparing the two).

[0091] Optionally, steps (e) and (f) are repeated, and subsequent states are used to determine the patient's response over time.

[0092] Accordingly, these methods (and the corresponding systems discussed below) of the second aspect and other aspects can be used for both clinical trials and clinical management. In terms of clinical management, a high degree of certainty (e.g., 70%, 80%, 90% or 95% probability) that the electrical activity in the brain of the patient (e.g., as evaluated using EEG) is abnormal in a "normal" person (i.e., currently undiagnosed) can be a strong indication to immediately initiate dementia drug treatment. EEG can also be used at intervals of, for example, 1, 2, 3, 4, 5 or 6 months to monitor the response to treatment. Conversely, those with a lower probability (e.g., 30%, 40%, 50%, 55% or 60%) of abnormal EEG can be more closely followed at intervals of once a month, once every two months or once every three months. Further tests by other means suitable for the disorder (as known in the art, e.g., amyloid or tau protein PET or CSF-based biomarker assessment) can optionally be used in combination with the method.

[0093] The optional features of the methods of the second aspect apply mutatis mutandis to this aspect.

[0094] ***

[0095] In a third aspect, the present invention provides a system for determining a patient's response to a neuropharmacological intervention, the system comprising:

[0096] a data acquisition component configured to obtain structural neurology data from a plurality of patients before a neuropharmacological intervention, the structural neurology data indicating the physical structure of a plurality of cortical regions;

[0097] a correlation matrix generation component configured to generate a first correlation matrix from the structural neurology data by:

[0098] assigning a plurality of structural nodes corresponding to brain cortical regions; and

[0099] determining pairwise correlations between structural node pairs based at least in part on the corresponding data of the structural neurology data;

[0100] wherein the data acquisition component is further configured to obtain further structural neurology data from the plurality of patients after the neuropharmacological intervention, the further structural neurology data indicating the physical structure of the plurality of cortical regions; and

[0101] the correlation matrix generation component is further configured to generate a second correlation matrix from the further structural neurology data by:

[0102] determining pairwise correlations between structural node pairs based at least in part on the corresponding data of the further structural neurology data;

[0103] The system further comprises:

[0104] a display component for presenting the first correlation matrix and the second correlation matrix; or

[0105] a comparison component for comparing the first correlation matrix with the second correlation matrix to determine the patient's response to the neuropharmacological intervention.

[0106] Optional features of the present invention will now be described. These are applicable either individually or in any combination with any aspect of the present invention.

[0107] The correlation matrix may mean generating a structural correlation network, which can then be represented by a matrix.

[0108] The measured or obtained physical structure may be cortical thickness and / or surface area. The values of cortical thickness and / or surface area may be the average values obtained from the structural neurological data. The structured neurological data may be collected from magnetic resonance imaging (MRI) data or computed tomography data of each patient. The structural neurological data and the further structural neurological data are obtained at different time points. As discussed herein, the structural neurological data may be obtained by magnetic resonance imaging, computed tomography, or positron emission tomography of each patient. These techniques are well known to those skilled in the art themselves - see, for example, Mangrum, Wells et al Duke Review of MRI Principles: Case Review Series E-book. Elsevier Health Sciences, 2018; and "Standardized low-resolution electromagnetic tomography (sLORETA): technical details" Methods Find Exp.Clin.Pharmacol.2002:24 Suppl D:5-12; Pascual-Marqui RD et al.

[0109] The plurality of cortical regions may be at least 60 or at least 65. For example, 68. The cortical regions may be, for example, those provided by the Desikan-Killiany atlas (Desikan et al 2006).

[0110] The display component may provide each of the first correlation matrix and the second correlation matrix on a display, wherein the correlation values in each correlation matrix are given in a color corresponding to the relative magnitude or intensity of the correlation.

[0111] The verification component can be configured to determine each pairwise-related p-value, compare each pairwise-related p-value, and can compare the p-value with a significance level. The correlation matrix generation component can be configured to use only p-values less than a corrected significance level when generating the correlation matrix. The significance level can be referred to as alpha (“α”).

[0112] The comparison component can be configured to compare the number and / or density of inverse correlations in the first correlation matrix with the number and / or density of inverse correlations in the second correlation matrix. When making the comparison, a group of structural nodes corresponding to the same lobe can be identified, and the comparison made between the first correlation matrix and the second correlation matrix can utilize the same lobe.

[0113] Assigning the plurality of structural nodes corresponding to the cerebral cortex regions can further include defining groups containing structural nodes corresponding to homologous or non-homologous lobes. The comparison component can be configured to compare the first correlation matrix with the second correlation matrix by comparing the number and / or density of correlations between different groups of structural nodes. In other words, comparing the first correlation matrix with the second correlation matrix can include comparing pairs of non-homologous structural nodes.

[0114] The comparison component can be configured to compare the first correlation matrix with the second correlation matrix by comparing the number and / or density of correlations between groups of structural nodes located in the frontal lobe (front nodes) and the parietal and occipital lobes (rear nodes), respectively. It has been found that in examples of effective neuropharmacological intervention, the number and / or density of inverse correlations between the front nodes and the rear nodes is reduced. Since it is assumed that inverse correlations indicate compensatory linkage formation, whereby atrophy in one node is associated with hypertrophy in functionally linked nodes, it should be understood that a reduction in the number and / or density of inverse correlations indicates a reduction in the number of compensatory linkages.

[0115] ***

[0116] In one embodiment, the patient response can be in the context of a clinical trial, such as a clinical trial for evaluating the efficacy of a drug for a neurocognitive disease. Thus, the group of patients (multiple patients) can be a treatment group that has been diagnosed with the disease, or can be a control (“normal”) group. Finally, the efficacy of the drug can be evaluated in whole or in part based on the patient group response determined according to the present invention.

[0117] As explained with respect to the first aspect, the neurocognitive disease will typically be a neurodegenerative disorder causing dementia, such as a tauopathy.

[0118] The patient may have been diagnosed with the neurocognitive disorder, such as Alzheimer's disease or behavioral variant frontotemporal dementia. The disorder may be mild or moderate Alzheimer's disease. The disorder may be mild cognitive impairment.

[0119] The diagnostic criteria and treatment of tauopathies and these disorders are discussed, for example, in WO 2018 / 019823 and the references cited therein.

[0120] The disorder may be behavioral variant frontotemporal dementia (bvFTD). The diagnostic criteria and treatment of bvFTD are discussed, for example, in WO 2018 / 041739 and the references cited therein.

[0121] As explained herein, the topology of the perturbations in the structural network is different in these two pathologies (AD and bvFTD), and both are different from normal aging. These changes appear to be adaptive, reflecting a coordinated increase in cortical thickness and surface area that compensates for the corresponding damage in functionally linked nodes.

[0122] Thus, if the neuropharmacological intervention is effective, it is expected to restore the network organization to the normal state. If the disorder is treated at an early enough stage, the network organization can be restored to the normal state indicating the halt or reversal of the disease state. Thus, the system provides an objective means of distinguishing disease-modifying treatment from symptomatic treatment as described above.

[0123] Generally, the neuropharmacological intervention will be a drug intervention.

[0124] The neuropharmacological intervention may be symptomatic treatment as described above.

[0125] For example, the disease-modifying treatment may be an inhibitor of pathological protein aggregation, such as the 3,7-diaminophenothiazine (DAPTZ) compound as described above.

[0126] ***

[0127] In a fourth aspect, the present invention provides a system for determining a patient's susceptibility to one or more neurological disorders, the system comprising:

[0128] a data acquisition component configured to obtain data indicative of electrical activity in the brain of the patient;

[0129] a network generation component configured to generate a network at least in part based on the obtained data, the network comprising a plurality of nodes and directed connections between the nodes, wherein the network indicates the flow of electrical activity in the brain of the patient;

[0130] A difference calculation component configured to calculate, for each node, the difference between the number and / or strength of connections entering the node and the number and / or strength of connections leaving the node; and any one of the following components:

[0131] A display component configured to display a representation of the calculated difference; or

[0132] A determination component configured to use the calculated difference to determine the patient's susceptibility to one or more neurological disorders.

[0133] As described above with respect to the second aspect, the inventors have shown that even very short use (e.g.) of brain EEG analysis can be used to potentially identify patients susceptible to one or more neurocognitive diseases (e.g., AD).

[0134] The system can be used for both clinical trials and clinical management.

[0135] Thus, in another aspect, there is provided a system as described above for determining a patient's response to a neuropharmacological intervention for a neurological disorder. In this aspect, the determination component system can be configured to use the calculated difference to determine the patient's status with respect to the neurological disorder.

[0136] The system can be used to determine the patient's further status after the neuropharmacological intervention and is optionally configured to determine the patient's response to the neuropharmacological intervention based on the first status and one or more subsequent statuses, as described above with respect to the corresponding method.

[0137] Other optional features of the present invention will now be set forth. These are applicable either individually or in any combination with any aspect of the present invention.

[0138] The system can include a state definition component configured to define a node as a sink or a source based on the calculated difference.

[0139] The network can be a renormalized partially directed coherence network. The system can operate "offline", i.e., without relying on the patient. For example, obtaining the data can be done by receiving data previously recorded from the patient via the network.

[0140] Data indicative of electrical activity in the brain can be electroencephalography data. The electroencephalography data can be beta-band electroencephalography data. Data indicative of electrical activity in the brain can also be magnetoencephalography data or functional magnetic resonance imaging data.

[0141] The determination component can be configured to use a machine learning classifier to determine the patient's susceptibility to one or more neurological disorders. For example, a Markov model, a support vector machine, a random forest, or a neural network.

[0142] The display member may be configured to present a heat map that indicates the location and / or intensity of nodes defined as sinks and nodes defined as sources within the brain. Such a representation of the defined nodes can assist (e.g., ergonomically) in determining the susceptibility of the patient.

[0143] The system may also include a heat map generation member configured to generate a heat map based at least in part on the states of the nodes, the heat map indicating the location and / or intensity of nodes defined as sinks and nodes defined as sources within the brain of the patient. Such a representation of the defined nodes can assist (e.g., ergonomically) in determining the susceptibility of the patient.

[0144] The determining member may compare the number and / or intensity of source points in the parietal and / or occipital lobes with the number and / or intensity of sink points in the compared frontal and / or temporal lobes. It has been experimentally observed that patients susceptible to one or more neurodegenerative diseases (and particularly Alzheimer's disease) have a relatively high number and / or intensity of sink points in the posterior lobes and a relatively high number and / or intensity of source points in the temporal and / or frontal lobes.

[0145] The system may also include an asymmetry map generation member configured to use the states of the nodes to derive an indication of the degree of left-right asymmetry of the location and / or density of the nodes in the brain corresponding to sink points and source points.

[0146] The neurological disorder may be a neurocognitive disease, which is optionally Alzheimer's disease.

[0147] The determining member may compare the number and / or intensity of nodes defined as sink points in the posterior lobes with a predetermined value, and / or compare the number and / or intensity of nodes defined as source points in the temporal and / or frontal lobes with a predetermined value. If the number and / or intensity of nodes defined as sink points in the posterior lobes exceeds the predetermined value and / or if the number and / or intensity of nodes defined as source points in the temporal and / or frontal lobes exceeds the predetermined value, then the determining member may determine the patient to have a high susceptibility risk. In other words, and more generally, the determination of susceptibility may be based on whether the patient has more and / or stronger source points and / or sink points in one region of the brain relative to another region of the brain. For example, if there are more and / or stronger source points in the temporal and / or frontal lobes than expected, and / or if the patient has more and / or stronger sink points in the posterior lobes than expected, then the patient may be determined to be at risk of a neurodegenerative disorder.

[0148] The inventors have further observed that, compared to the non-medicated group, one or more symptomatic treatments increase the activity exiting the frontal lobe.

[0149] The system of the invention according to this aspect can be used to evaluate, test or classify the susceptibility of a subject to one or more neurological disorders for any purpose. For example, the score or other output of the test can be used to classify the mental state or disease state of the subject according to a predetermined criterion.

[0150] The subject can be any human subject. In one embodiment, the subject can be a subject suspected of having a neurocognitive disease or disorder as described herein (e.g., a neurodegenerative or vascular disease of the nervous system), or can be a subject not identified as being at risk.

[0151] In one embodiment, the system is used for early diagnosis or prediction of cognitive impairment in the subject, such as a neurocognitive disease, as described above.

[0152] The system can optionally be used to inform the subject of further diagnostic steps or interventions - for example, based on other systems for imaging or invasive or non-invasive biomarker assessment, which are known in the art per se.

[0153] In some embodiments, the system can be used to determine the risk of a neurocognitive disorder in the subject. Optionally, the risk can be calculated using additional factors, such as age, lifestyle factors, and other measured physical or mental criteria. The risk can be classified as "high" or "low", or can be presented as a scale or spectrum.

[0154] Like the methods described herein, the system can be used in the context of a clinical trial to evaluate the efficacy of a neuropharmacological intervention. The system can be used to confirm the efficacy of a disease-improving treatment (e.g., LMTM) in a relatively small number of subjects (e.g., 50) over a relatively short time scale (e.g., 6 months) and at an early disease stage (e.g., mild cognitive impairment or possible pre-mild cognitive impairment).

[0155] Other aspects of the invention provide: a computer program comprising executable code that, when run on a computer, causes the computer to perform the method of the first or second aspect; a computer-readable medium storing a computer program comprising code that, when run on a computer, causes the computer to perform the method of the first or second aspect; and a computer system programmed to perform the method of the first or second aspect. For example, a computer system can be provided that includes: one or more processors configured to: perform the method of the first or second aspect. The system thus corresponds to the method of the first or second aspect. The system can further include: one or more computer-readable media operatively connected to the processor, the one or more media storing computer-executable instructions corresponding to the method of the first or second aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0156] Embodiments of the present invention will now be described by way of example with reference to the accompanying drawings, in which:

[0157] Figure 1 An example of the Desikan-Killiany brain atlas is shown;

[0158] Figure 2 An exemplary cortical surface area correlation matrix of a group of subjects diagnosed with behavioral variant frontotemporal dementia is shown, having pairwise correlations grouped by lobe;

[0159] Figures 3A to 3C Group-based cortical thickness correlation networks depicted as pairwise correlation matrices for the following groups are shown respectively: (i) a group of healthy elderly (HE) subjects, (ii) a group of bvFTD subjects, and (iii) a group of Alzheimer's disease (AD) subjects;

[0160] Figures 4A to 4C Group-based surface area correlation networks depicted as pairwise correlation matrices for the following groups are shown respectively: (i) a group of healthy elderly (HE) subjects, (ii) a group of bvFTD subjects, and (iii) a group of Alzheimer's disease (AD) subjects;

[0161] Figure 5 A plot of the average edge strength of the cortical thickness correlation network averaged over the lobes and compared between the HE, bvFTD, and AD groups is shown, where the upper plot is the network of positive correlations and the lower plot is the network of inverse correlations;

[0162] Figure 6 A plot of the average edge strength of the surface area correlation network averaged over the lobes and compared between the HE, bvFTD, and AD groups is shown, where the left plot is the network of inverse correlations and the right plot is the network of positive correlations;

[0163] Figure 7 A plot of the node degree of the cortical thickness correlation network averaged over the lobes and compared between the HE, bvFTD, and AD groups is shown, where the upper plot is the network of positive correlations and the lower plot is the network of inverse correlations;

[0164] Figure 8 A plot of the node inter-lobe participation index of the cortical thickness correlation network averaged over the lobes and compared between the HE, bvFTD, and AD groups for positive correlations is shown;

[0165] Figure 9 A plot of the node degree of the surface area correlation network averaged over the lobes and compared between the HE, bvFTD, and AD groups is shown, where the upper plot is the network of positive correlations and the lower plot is the network of inverse correlations;

[0166] Figure 10 Plots of the interlobar participation index of nodes of the surface area-related network averaged over the lobes and compared between the HE, bvFTD, and AD groups. The upper plots are positively correlated networks and the lower plots are inversely correlated networks;

[0167] Figure 11 Visualization in brain space of hubs in the cortical thickness network for the positively correlated nodes (in the upper plots) and inversely correlated nodes (in the lower plots) for the HE, bvFTD, and AD groups;

[0168] Figure 12 Visualization in brain space of hubs in the surface area network for the positively correlated nodes (in the upper plots) and inversely correlated nodes (in the lower plots) for the HE, bvFTD, and AD groups;

[0169] Figure 13 Visualization in brain space of the interaction between cortical thickness and the cortical surface area positive network for the HE, bvFTD, and HE groups;

[0170] Figure 14 Histograms of the retained edges in the cortical thickness- and surface area-related networks;

[0171] Figure 15 A plot showing the distribution of the modularity index (Q) in the regional cortical thickness-related networks generated on 100 surrogate datasets;

[0172] Figure 16 Binary correlation matrices of the cortical thickness networks (upper three plots) and surface area (lower three plots) for the HE, bvFTD, and AD groups, where white represents significant positive correlation and black represents significant inverse correlation;

[0173] Figures 17A to 17D Correlation matrices at baseline (i.e., week 01) according to the treatment status with symptomatic AD medications (cholinesterase inhibitors and / or memantine), where ach0 indicates no treatment and ach1 indicates the presence of such treatment;

[0174] Figures 18A to 18D Plots of the nodal degrees of non-homologous interlobar correlations at baseline nodal degrees according to the treatment status with symptomatic AD medications (acetylcholinesterase inhibitors and / or memantine);

[0175] Figure 19A and Figure 19B Correlation matrices of cortical thickness separated in time at baseline (week 01) and after 65 weeks (week 65) in patients receiving a combination of 8 mg / day of LMTM and symptomatic treatment;

[0176] Figures 20A to 20D Shows graphs of the positive and inverse non-homologous inter-lobular nodal degrees in cortical thickness (CT) and surface area (SA) at baseline and after 65 weeks in patients receiving the combination of 8 mg / day LMTM and symptomatic treatment;

[0177] Figure 21A and Figure 21B Shows cortical thickness-related matrices separated in time at baseline (Week 01) and after 65 weeks (Week 65) in patients receiving 8 mg / day LMTM as monotherapy (i.e., not in combination with symptomatic AD treatment);

[0178] Figure 22A and Figure 22B Shows graphs of the non-homologous inter-lobular nodal degrees of the cortical thickness-related network at baseline and after 65 weeks in patients receiving 8 mg / day LMTM as monotherapy (i.e., not in combination with symptomatic AD treatment);

[0179] Figure 23A and Figure 23B Shows surface area-related matrices separated in time at baseline (Week 01) and after 65 weeks (Week 65) in patients receiving 8 mg / day LMTM as monotherapy (i.e., not in combination with symptomatic AD treatment);

[0180] Figure 24A and Figure 24B Shows graphs of the non-homologous inter-lobular nodal degrees of the surface area-related network at baseline and after 65 weeks in patients receiving 8 mg / day LMTM as monotherapy (i.e., not in combination with symptomatic AD treatment);

[0181] Figures 25A to 25D Shows cortical thickness-related matrices separated in time at baseline and after 65 weeks of treatment with 8 mg / day LMTM as monotherapy for both AD (Clinical Dementia Rating 0.5, 1, and 2) and elderly control group (HE) to show the normalization of the post-treatment matrix;

[0182] Figures 26A to 26D Shows surface area-related matrices separated in time at baseline and after 65 weeks of treatment with 8 mg / day LMTM as monotherapy for both AD (Clinical Dementia Rating 0.5, 1, and 2) and elderly control group (HE) to show the normalization of the post-treatment matrix;

[0183] Figures 27A to 27DShows the cortical thickness correlation matrices separated in time for both AD (Clinical Dementia Rating 0.5) and healthy elderly controls (HE) at baseline and after 65 weeks of treatment with 8 mg / day LMTM as monotherapy, to show the normalization of the matrices after treatment;

[0184] Figures 28A to 28D Shows the surface area correlation matrices separated in time for both AD (Clinical Dementia Rating 0.5) and healthy elderly controls (HE) at baseline and after 65 weeks of treatment with 8 mg / day LMTM as monotherapy, to show the normalization of the matrices after treatment;

[0185] Figure 29 Shows examples of resting electroencephalography data;

[0186] Figure 30 Shows examples of directed networks derived from electroencephalography data;

[0187] Figure 31 Schematically shows the determination of node states based on the difference between incoming and outgoing electrical activity currents;

[0188] Figure 32 Shows a heatmap of the positions of net sinks (yellow / red) and net sources (blue) within the brains of a group of subjects;

[0189] Figure 33 Shows an indication of the Figure 32 Heatmap showing the asymmetry in the distribution of sources and sinks between the left and right sides of the heatmap in

[0190] Figure 34 Shows a heatmap of the positions of sources and sinks within the brains of a group of subjects diagnosed with Alzheimer's disease;

[0191] Figure 35 Shows a heatmap of the positions of sources and sinks within the brains of a group of subjects not diagnosed with Alzheimer's disease (i.e., paired volunteers);

[0192] Figure 36 Shows a heatmap of the positions of sources and sinks within the brains of subjects not diagnosed with Alzheimer's disease (i.e., paired volunteers);

[0193] Figure 37 Shows a heatmap of the positions of sources and sinks within the brains of subjects diagnosed with Alzheimer's disease;

[0194] Figure 38 Shows a heatmap of the positions of sources and sinks within the brains of a group of subjects determined to be at risk of dementia or cognitive decline, for example, due to having Alzheimer's disease;

[0195] Figure 39 Shows a heatmap of the positions of source and sink points in the brains of a group of subjects determined to be at no risk of Alzheimer's disease;

[0196] Figure 40 Is a box plot comparing source and sink points in EEG networks from frontal and posterior brain regions at the group level in subjects with and without AD risk;

[0197] Figure 41 Shows a comparison between the cortical thickness correlation matrix of the AD group (which shows an increase in the strength and number of significant inverse correlations between non-homologous brain lobes, left panel) and a heatmap of the positions of source and sink points in the brains of a group of subjects diagnosed with AD, which shows a correspondence between the increase in compensatory structural non-homologous inverse correlations in cortical thickness pointing to the posterior brain region and the increase in the strength and number of incoming connections to the posterior brain region as sink points shown by rPDC coherence analysis of resting state EEG;

[0198] Figure 42 Shows a comparison between the surface area correlation matrix of the HE group and a heatmap of the positions of source and sink points in the brains of a group of healthy elderly subjects, which shows a correspondence between the relative lack of compensatory structural non-homologous inverse correlations in cortical thickness pointing to the posterior brain region and the decrease in the number and strength of incoming connections to the posterior brain region as sink points shown by rPDC coherence analysis of resting state EEG;

[0199] Figure 43 Is a box plot showing the quantitative differentiation between mild AD and elderly controls;

[0200] Figure 44 Shows three heatmaps comparing medicated and unmedicated AD patients with paired volunteers at the group level; and

[0201] Figure 45 Is a group-level network box plot comparing medicated and unmedicated AD patients with paired volunteers. Detailed implementation

[0202] Aspects and embodiments of the present invention will now be discussed with reference to the accompanying drawings. Other aspects and embodiments will be apparent to those skilled in the art. All documents mentioned in this text are incorporated herein by reference.

[0203] Figure 1An example of the Desikan-Killiany brain atlas is given. The Desikan-Killiany brain atlas divides the human cerebral cortex into gyrus-based target regions on MRI scans. Although 18 regions are shown in the figure, the full Desikan-Killiany brain atlas divides the human cortex into 68 target regions.

[0204] The subjects discussed in this document participated in three global Phase 3 clinical trials, which are now completed. Two of the clinical trials were conducted in mild to moderate AD (Gauthier et al., 2016; Wilcock et al., 2018), and the third was a large study from bvFTD (Feldman et al., 2016). Comparable data were obtained from well-characterized healthy elderly (HE) subjects who participated in the ongoing longitudinal study of the 1936 Aberdeen Birth Cohort (ABC36) (Murray et al., 2011). In the examples discussed in this article, there were a total of 628 subjects, 213 in each dementia group, and 202 healthy elderly subjects. The bvFTD patients were diagnosed with mild severity according to the international consensus criteria for bvFTD, with a Mini-Mental State Examination (MMSE) score of 20 - 30, inclusive of the end values. The AD patients were diagnosed with mild to moderate severity according to the criteria of the National Institute of Aging and the Alzheimer’s Association, defined as an MMSE score of 14 - 26 (inclusive of the end values) and a Clinical Dementia Rating (CDR) sum score of 1 or 2. They were drawn from the respective larger groups (N = 1132) to match the number of participants in the bvFTD group. The healthy elderly (HE) subjects were selected from the well-characterized 1936 Aberdeen Birth Cohort.

[0205] The multi-site source imaging dataset used to generate the correlation matrices discussed below is standard T1-weighted MRI images acquired in sequence using the same manufacturer's proprietary 3D T1. Data from the trial patients were combined to allow for overall between-group comparisons. The training scanners were limited to 1.5T and 3T (30%) field strengths from three manufacturers (Philips, GE, and Siemens). The MRI images in the ABC36 cohort were all acquired using the same (Philips) 3T scanner. The images were processed using an automated processing pipeline implemented in a manner known per se. In addition to volume-based image processing methods, the pipeline also produces surface-based regional measurements of cortical morphology such as thickness, local curvature, or surface area. An example of an automated processing pipeline suitable for the above method is FreeSurfer v5.3.0, which is available from the Athinoula A. Martinos Center for Biomedical Imaging at Massachusetts General Hospital.

[0206] Surface area was calculated from the imaging dataset using a triangulation of the gray / white matter interface and the white matter / cerebrospinal fluid boundary (referred to as the pial surface). Cortical thickness was calculated as the average of the distances from the white matter surface to the nearest point on the pial surface and back to the nearest point on the white matter surface. Cortical thickness and surface area of 68 cortical regions in both hemispheres were extracted based on the Desikan-Killiany atlas using a segmentation scheme known per se. A list of the regions and their lobe assignments is given in Table A.1 of Appendix A.

[0207] Figure 2 The cortical surface area correlation matrices for a group of subjects diagnosed with bvFTD are shown. Each matrix element represents the correlation strength ("edge strength") between 68 pairs of cortical surface areas from the Desikan-Killiany atlas. The intensity bar on the right indicates the correlation / edge strength. The sixty-eight cortical surface regions (network nodes) were sorted according to their affiliation with the frontal, temporal, parietal, and occipital lobes. Individual lobe regions are enclosed in squares and are sorted from top to bottom / left to right as: frontal, temporal, parietal, and occipital. Essentially, the correlation matrix represents a network constructed from partial correlations between 68 pairs of cortical thicknesses. Figures 3A to 3CRespectively showed the cortical surface area correlation matrices of healthy elderly, behavioral variant frontotemporal dementia (bvFTD) subjects, and Alzheimer's disease (AD) subjects. Significant differences were observable between healthy elderly and both bvFTD and AD subjects. Notably, within the lobes, the correlation strength in bvFTD and AD subjects increased significantly. In addition, the number of inverse correlations between non-homologous nodes increased. As can be observed, healthy elderly (HE) subjects had sparse correlations, and these were mostly positive correlations between homologous lobes. However, compared with the HE group, both bvFTD and AD had a significantly increased number of nodes linked by positive and inverse correlations. The increase in the number of correlations in both forms of dementia could be between the same lobes (homologous, mostly positive) or between different lobes (non-homologous, mostly negative). Generally, inter-lobe non-homologous inverse correlations were highly abnormal. Similarly, it was observable that bvFTD was particularly associated with a higher density of non-homologous inverse correlations in cortical thickness.

[0208] Figures 4A to 4C Respectively showed the surface area correlation matrices of healthy elderly, behavioral variant frontotemporal dementia (bvFTD) subjects, and Alzheimer's disease (AD) subjects. Significant differences were observable between healthy elderly and both bvFTD and AD subjects. In addition, it should be noted that AD was particularly associated with a higher density of non-homologous inverse correlations in surface area.

[0209] In these figures, zero entries corresponded to non-significant correlations. Significant network correlations were found to have both positive and negative values (for an explanation, see Figure 14 and Figure 16 ). Due to the obvious increase in the number of significant inverse correlations and the network correlation strength in the bvFTD and AD groups relative to the HE group, the sub-networks of significant positive and inverse correlations were considered separately. Given the obvious differences in the lobe network structure according to the diagnostic groups, an attempt was made to determine whether these differences could be quantified.

[0210] The networks represented as correlation matrices in these figures could be constructed by correlating the surface area or cortical thickness between all subjects within a specific diagnostic category (i.e., HE, bvFTD, and AD). Cortical regions (as defined by the Desikan-Killiany brain atlas) represented nodes, and the pairwise correlations between nodes represented graph edges, or the linkages / connections were constructed by correlating the SA or CT between all participants within each diagnostic category. Each correlation matrix was calculated based on an S x N array containing N regional CT / SA values from S subjects within each group. In this way, six N x N (e.g., 68 x 68) correlation matrices were obtained (one CT or SA structural correlation matrix for each study group). Matrix element e ij was between regions i and j (i, j = 1, 2,... N) (i.e., vector x i and xj Among them, they contain partial correlation values of regional measurements from the subjects within each group). After first removing the influence of all other regions m≠(i, j), and then adjusting x for the control variables (stored in a separate array S x C, where C represents the number of control variables) i and x j After both, the partial correlation is calculated as the linear Pearson correlation coefficient between x i and x j pairs. This means that before the correlation analysis, a linear regression is performed on each x i to remove the influence of age, gender, and mean CT (average cortical thickness of all regions) or total surface area (sum of the total surface areas). The autocorrelation (represented as the main matrix diagonnetwork measurement) is calculated on the lower triangular part of the matrix. The partial correlation e ij (i.e., the edge weight) can be calculated according to the following general equation:

[0211] e ij = ρ i ≡ corr(x i , x j |x c )

[0212] where x i，j represents the variable array, and x c represents any subset of the conditional variables. To achieve the partial correlation in this general form, the process starts with i, j, c = 1, 2, 3:

[0213]

[0214] Therefore, for any subset of c of the conditional variables:

[0215]

[0216] In some examples, to verify that the network only retains statistically significant correlations, the false discovery rate (FDR) procedure as described in Storey, 2002 is used to adjust the calculated correlation coefficients for multiple tests. The FDR procedure tests each calculated p-value (from the pairwise correlation calculation) against the corrected significance level. In this example, α = 0.05, and only p-values less than the adjusted significance level are accepted as truly significant. Those pairwise correlations that do not pass the FDR test can be set to zero; alternatively, all non-zero correlations (whether positive or negative) are retained (see Figure 14 , discussed in more detail below).

[0217] In this way, a 68×68 correlation matrix can be constructed for CT or SA in each clinical group, which represents the structural correlation network of surface area or cortical thickness. The matrix elements quantify the correlation strength between cortical regions for cortical thickness or surface area, and they do not substantially represent actual physical connections. In the case of structural correlation network analysis of neurodegenerative disorders, such correlations are considered to imply co-atrophy relationships (if positive) or inverse atrophy / hypertrophy relationships (if negative) between brain regions.

[0218] Regarding the structural correlation networks and / or matrices generated using the methods described above, it is useful to use the following measurements to compare the structural network properties of the three clinical groups: edge strength, node degree, within-module degree z-score of nodes, and participation index. Edge strength and node degree represent two basic network properties; they quantify the correlation strength between nodes and the number of pairwise correlations of each node, respectively. To evaluate whether the cortical lobes represent modules, two network measures that assess modularity in network interactions are utilized, namely the within-module degree z-score and the participation index. All measures (except node degree) are calculated on weighted graphs and are estimated as the average among the four brain lobes (described below). The measures are calculated on binary or weighted graphs (as discussed below). It is known from pure theoretical studies that the calculated network topology properties depend on the choice of threshold (van Wijk et al. 2010). In this document, a fixed threshold is selected for each group-based correlation matrix.

[0219] Node degree

[0220] Node degree k i Represents the number of significant correlations for each node in the network. Generally, the node degree is calculated from a binarized correlation matrix, where each significant correlation in the matrix is replaced by 1 (if it is significant) or by 0 (if it is not significant). An example of a binarized matrix is shown in Figure 16 . The binarized matrix can also be referred to as an adjacency matrix. In Figure 16 , the three upper graphs correspond to cortical thickness, and the three lower graphs correspond to surface area. Significant positive correlations are shown in white, while significant inverse correlations are shown in black.

[0221] The node degree of node i (i.e., the number of significant linkages connected to the node) can be calculated as follows:

[0222]

[0223] where N is the number of nodes, and a ij represents the connection between node i and j, which has a value of 1 when there is a direct connection between the nodes and 0 otherwise.

[0224] Modularity index

[0225] The node participation index and the within-module degree z-score evaluate the role of nodes according to the module. A network module (also known as community structure) represents a densely connected subgraph of the network, that is, a subset of nodes with denser internal network connections and sparser connections between them. It is useful to examine the modular organization such as the frontal, temporal, parietal, and occipital partitions of the cortical thickness or surface area network defined as modules. Since these lobe partitions of the cortical surface area are not necessarily modular in themselves, it may be necessary to first test whether the lobe partitions are inherently modular. In one example, this can be done by calculating the modularity index (Q) of the network according to each lobe. The modularity index quantifies the fraction of the observed within-module degree values relative to those expected when the connections are randomly distributed between the networks. Since the constructed cortical thickness and surface area networks contain both positive and negative edge intensities, an asymmetric generalization of the modularity quality function can be used. For example, as introduced in Rubinov and Sporns (2011):

[0226]

[0227] where if the strength ω of the pairwise correlation between cortical regions ij > 0, then is equal to the i,j-th element of the correlation matrix, that is, ω ij , otherwise it is equal to zero. Similarly, if ω ij >0, then is equal to -ω ij , otherwise it is equal to zero. The term represents the expected density of positive or negative connection weights, given a random null model that preserves the strength, where and when the i,j-th nodes are within the same module, the Kronecker delta function is equal to one, otherwise it is equal to zero. The performance of a given separation of the network into modules is tested by: applying a community detection function known in the art itself, while adopting a vector of node memberships, with a specific node as the initial community membership vector.

[0228] It is found that the organization of the cortical surface lobes into frontal, parietal, temporal, and occipital partitions is actually modular (see Appendix A). Therefore, the contribution of individual nodes to the lobe modules can then be calculated as the node participation index and the within-module z-score, which are called the inter-lobe node participation index and the intra-lobe node z-score.

[0229] Inter-lobe node participation index

[0230] Generally, the participation index p evaluates the inter-module connectivity. It can be considered as the ratio of the intra-lobe node edges to all other lobe modules in the network, where if a node has only linkages within its own module, then the node pi Tends to 0, and tends to 1 if the node has linkages only outside its own module. The weighted network participation is calculated by the following equation:

[0231]

[0232] where M is the module group and is the weighted number of linkages of the i-th node with all other nodes in module m - the inter-module degree, and is the total degree of the i-th node. In this document, the term inter-lobe participation is used for this network measure.

[0233] Node intra-lobe degree z-score

[0234] Complementary to the inter-lobe participation index is the normalized intra-lobe degree z i , which evaluates the intra-lobe connectivity by means of the z-score, that is, by the normalized deviation of the inter-lobe degree of the nodes with the corresponding average degree distribution. Thus, the node intra-lobe z-score z i is larger for nodes with more intra-module connections relative to the inter-module average connectivity. For a network that preserves the relevant strength, the node intra-module degree z-score is calculated as follows:

[0235]

[0236] where as above, is the mean of the degree distribution within module m i and is the standard deviation of the degree distribution within module m i .

[0237] Node roles in network module organization

[0238] The node roles with modular lobe organization depend on its position in the z_p i parameter space. A node can have four possible roles in the network, which are assigned based on measures of node characteristics above the average. It is useful to consider two of these roles, namely the so-called connectors or global network hubs (which have high inter-lobe participation and high intra-lobe degree z-scores) and the so-called local hubs (which have high intra-lobe degree z-scores and low inter-lobe participation). Set the thresholds for high and low values of z i and p i to be above 1.5 and 0.05 respectively.

[0239] Statistical analysis

[0240] The statistical differences in the demographics and cognitive scores of the subjects were evaluated using one-way ANOVA or two-tailed t-tests. The one-sample Kolmogorov-Smirnov test was used to check the distribution normality of the data. The chi-square test was used to examine the distribution differences between men and women across groups. For unbalanced sample sizes, one-way ANOVA was used to test the statistical differences in global network-related strength according to the diagnostic groups (to account for unequal numbers of significant correlations between networks). The Kruskal-Wallis test (a non-parametric one-way ANOVA test) was used to compare node degree, within-lobe z-score z i and between-lobe participation index p i ,. Results were reported as significant at the p < 0.05 level.

[0241] Results

[0242] Table 1 below shows the demographics, cognition, and mean CT and SA for each group according to the clinical diagnosis. The ages of the 3 groups were significantly different, with AD patients being older than HE and bvFTD (p < 10 in all tests -4 ). Significant differences were also observed in the cognitive scores on the MMSE scale, with AD patients being the most impaired compared to HE subjects, and bvFTD patients being more impaired (p < 10 in all tests -4 ). The mean CT and total SA differed across groups.

[0243] Table 1

[0244]

[0245] Abbreviations: HE - healthy elderly, bvFTD - behavioral variant frontotemporal dementia, AD - Alzheimer's disease, M - male, F - female, MMSE - Mini-Mental State Examination, CT - cortical thickness, SA - surface area. Significant differences between groups: a - HE / bvFTD, b - HE / AD, c - bvFTD / AD (p < 0.05).

[0246] The differences between HE and the two patient groups were significant in terms of mean CT (p < 10 in both tests -4 ) and total SA (p < 0.003 in both tests), but the bvFTD and AD groups were not different from each other. The mean CT and total SA values averaged by lobe are given in Table A.3 of Appendix A. Thus, while AD and bvFTD differ in terms of the lobar distribution of pathology, age, and severity of cognitive impairment, neither the overall degree of cortical thinning nor the change in mean surface area provides a means to distinguish between these two disorders.

[0247] Lobar characteristics of the structural correlation networks

[0248] Since the definition of the network based on the relevant network organization depends on the choice of the threshold, it is useful to ensure that the networks defined in this paper are non-random in their global topology by calculating the density / sparsity value (k). If κ > 0.1, then the brain network is considered to show a non-random (small-world) topology, which is the case for all the networks considered here. It is also useful to ensure that the inverse correlations are not ignored after thresholding (see Figure 14 ). Thus, all the positive and inverse CT and SA correlation networks considered in this paper are non-random. For the total values of κ for CT and SA in these three groups, also see Table A.3.

[0249] The modularity index was used for the study to determine whether the cortical lobes as conventionally defined correspond to the network modules in the CT network. It was found that in the modularity index algorithm, only two homologous pairs in the CT network (the posterior cingulate gyrus and the precentral cortex) and two homologous pairs in the SA network (the posterior cingulate cortex and the right posterior bank of the paracentral gyrus and the superior temporal sulcus) were misassigned. Table A.2 (in Appendix A) gives the details of the algorithm input and output. In practice, it is generally recognized that a Q value higher than 0.3 is a good indicator of the presence of important modules in the network. To estimate the confidence interval of the Q value of the dataset, repeated calculations were performed for 100 CT matrices generated on alternative datasets. Each of the 100 alternative CT and SA matrices was generated as follows: 213 subjects were randomly selected from these three study cohorts, and the Q value was calculated on the correlation matrices obtained for CT and SA. The values of Q are shown in Figure 15 , which is a graph of the distribution of the modularity index Q in the regional CT networks generated on 100 alternative datasets. The central line indicates the mean, and the upper and lower lines indicate 1.5 standard deviations from the mean (Q = 0.36 ± 0.02), i.e., random, where the Q value is similar to that of the random graph. For the study group, the following values were obtained: Q HE = 0.49, Q bvFTD = 0.49 and Q AD = 0.45 (for the "positive" sub-network) and Q HE = 0.32, Q brFTD = 0.28 and Q AD = 0.29 (for the "negative" sub-network), and these values indicate a non-random modular topology organization in the frontal, temporal, parietal, and occipital partitions of the CT / SA correlation networks.

[0250] Therefore, it can be concluded that the cortical lobes as conventionally described correspond to the non-random modules in the CT network.

[0251] The average correlation strength of the CT and SA networks

[0252] Figure 5Shows the edge strength of each cortical thickness-related network averaged over the lobes and compared between the HE, bvFTD, and AD groups. Data for the networks with positive (upper panels) and inverse (lower panels) correlations are shown. Asterisks indicate significant differences between the three groups (*p < 0.05; **p < 0.01). As can be observed in Figure 5 , the mean correlation strength of CT showed significant differences between HE, bvFTD, and AD subjects in the frontal, temporal, parietal, and occipital lobes (for all tests, p < 10 -4 ). In the frontal, temporal, parietal, and occipital lobes, the mean correlation strength in bvFTD and AD subjects was higher than that in HE subjects (for all pairwise comparisons, p ≤ 0.003). In the frontal lobe (p < 10 -4 ) and temporal lobe (p = 0.005), the mean correlation strength in bvFTD was higher than that in AD.

[0253] The mean strength of the networks with inverse correlations in the CT network also differed between the frontal and temporal lobes, see the lower panel of Figure 5 . Similarly, in the frontal and temporal lobes, both the bvFTD and AD groups showed higher mean correlation strength than the HE group (in all tests, p ≤ 0.03), and in the frontal lobe, the mean inverse correlation strength of the bvFTD group was higher than that of the AD group (p = 0.003).

[0254] Figure 6 Showing the edge strength of each surface is the correlation network averaged over the frontal lobe of the brain and compared between the HE, bvFTD, and AD groups. Data for the networks with positive (right panels) and inverse (left panels) correlations are shown. Asterisks indicate significant differences between the three groups (*p < 0.05; **p < 0.01). The figure shows significant differences in the mean correlation strength between the SA networks. The diagnostic groups differed only in the frontal lobe, where the mean correlation strength of the AD group was lower than that of the HE group (p = 0.03). Similarly, when compared with the HE group, the inverse SA network correlations in the bvFTD and AD groups were significantly different in the frontal lobe and had lower mean correlation strength (in both tests, p ≤ 0.02). This is due to a larger number of correlations and a wider frequency distribution of strengths found in disease compared to the sparser networks with a narrower frequency distribution in healthy elderly subjects (see Figure 14 ).

[0255] Node metrics in the CT network

[0256] Node degree

[0257] Node degree quantifies the average number of significant positive correlations for each node. For the CT network, the node degree averaged over the frontal, temporal, parietal, and occipital lobes is shown in Figure 7Among them, node degrees were compared between the HE, bvFTD, and AD groups. Data for networks showing positive correlations (upper panel) and inverse correlations (lower panel) are presented. Asterisks indicate significant differences among the three groups (*p < 0.05; **p < 0.01).

[0258] There were significant differences among groups in the frontal, temporal, parietal, and occipital lobes (p ≤ 10 in all tests). -4 ) Compared with HE subjects, both bvFTD and AD subjects had higher node degrees in the frontal and temporal lobes (p < 0.006 for all tests). The bvFTD group had significantly higher node degrees than the AD group in the parietal and occipital lobes (p ≤ 0.02 for all tests). A similar pattern was found for the number of inverse correlations in the CT network in the frontal, temporal, parietal, and occipital lobes (p ≤ 0.02 in all tests). These differences were driven by a greater number of significant inverse correlations in the bvFTD and AD groups than in the HE group among all four lobes (p < 0.01 in all tests). There were no significant differences between the bvFTD and AD groups.

[0259] Inter-lobe participation index of nodes

[0260] Group differences were found in the inter-lobe participation index of nodes in CT. This index measures the degree of significant positive correlations with nodes in different lobes. This was significant for lobes located in the temporal, parietal, and occipital lobes (p < 0.03 for all tests). In the parietal lobe (p < 0.003 in both groups), temporal lobe (p = 0.01 in AD), and occipital lobe (p = 0.002 in bvFTD), these differences reflected higher index values relative to the HE group. This is shown in Figure 8 where the inter-lobe participation index of nodes in the cortical thickness-related network averaged over lobes was compared between the HE, bvFTD, and AD groups. Only data for positive correlations are shown in the figure. Asterisks indicate significant differences between groups (*p < 0.05; **p < 0.01). For inverse correlations in the CT network, the inter-lobe participation index comparisons were not significantly different in any lobe.

[0261] Node measures in the SA network

[0262] Node degree

[0263] For the frontal, temporal, parietal, and occipital lobes, node degree values in the SA network are shown in Figure 9 In the figure, the node degrees of the surface area-related network were averaged over lobes and compared between the HE, bvFTD, and AD groups. Data for networks showing positive correlations (upper panel) and inverse correlations (lower panel) are presented. Asterisks indicate significant differences among the three groups (*p < 0.05; **p < 0.01).

[0264] Among the frontal, temporal, parietal, and occipital lobes, the positive correlations differed between diagnostic groups (p ≤ 0.03). As with the CT network, in the frontal, temporal, and parietal lobes, the SA nodal degrees were higher in both the bvFTD and AD groups than in the HE group (in all tests, p < 10 -4 ). For the occipital lobe, the only significant difference was between the AD and HE groups (p = 0.04). Compared with the CT network, the nodal degree in the parietal lobe was also significantly higher in AD than in bvFTD (p = 0.004).

[0265] In the frontal, temporal, parietal, and occipital lobes, the inverse correlation SA network also showed significant group differences (in all tests, p ≤ 0.001). Similarly, in all four lobes, the nodal degrees were higher in both the bvFTD and AD groups than in the HE group (for all tests, p < 0.001). Compared with the CT inverse correlation network, in the frontal (p = 0.02) and parietal (p = 0.01) lobes, the nodal degree of the AD group was higher than that of the bvFTD group.

[0266] Inter-lobe participation index of nodes

[0267] Figure 10 Group differences in the inter-lobe participation index of nodes showing the SA network organization are shown. The figure shows the inter-lobe participation index of nodes averaged over the lobes and compared between the HE, bvFTD, and AD groups. Data for the positive correlation (upper panel) and inverse correlation (lower panel) networks are shown. Asterisks indicate significant differences between the three groups (*p < 0.05; **p < 0.01).

[0268] In all four lobes, for the positive SA correlation network, the index values were higher in both the bvFTD and AD groups than in the HE group (p < 10 -4 ). Compared with the CT correlation network, the inverse SA correlation network also showed significant differences in the frontal and parietal lobes (for both patient groups, p ≤ 0.04) and in the temporal lobe (for the AD group, p < 0.001) relative to the HE group.

[0269] Central points of the structural correlation network

[0270] Central points of the CT network

[0271] There are four possible combinations of the mean of the interlobar participation index (p high / low) and the within-lobe z-score (high / low). Here, only the case of high interlobar index and high within-lobe z-score is considered to focus on nodes with high centrality-like features. Tables A.4 to A.6 (see Appendix A) provide data on global and local network centralities. The other two combinations were examined, but no information was provided. In the positive CT-related network, the number and distribution of network centralities within high p-values and high z-values differed between the study groups. In the HE subjects, centralities were distributed throughout the cortex; each lobe had at least one centrality, with four centralities in the frontal lobe. The reorganization of centrality topology proceeded differently in these two disease groups. This is shown in Figure 11 the upper figure of

[0272] which is a visualization of the centralities of the cortical thickness network in brain space. In bvFTD, the number of centralities increased from 4 to 9 in the frontal lobe, decreased from 2 to 1 in the occipital lobe, and disappeared completely in the parietal and temporal lobes. In contrast, in AD, centralities were distributed almost equally among all four lobes. The number of centralities decreased in the frontal lobe (2 vs. 4), while the number increased in the temporal and occipital lobes relative to the HE group (1 vs. 3 and 2 vs. 3, respectively). A complete list of nodes with centrality-like properties and lobe locations in the CT network is provided in Table A.4 (see Appendix A).

[0272] In all three groups, nodes with centrality-like properties in the inverse correlation CT matrix were present only in the frontal and temporal lobes, and their topological distribution differed between groups. See Figure 11 the lower figure of

[0273] SA network centralities

[0274] Centralities in the positive correlation SA network are shown in Figure 12 the upper figure of

[0275] Table A.5 (see Appendix A) provides a list of nodes and lobe locations classified according to the interlobar participation index and the within-lobe z-score. Visual comparison of centrality topology between groups showed that in all diagnostic groups, there were more nodes with centrality-like properties in the left hemisphere. However, HE subjects had only one SA centrality (left insula), while the two disease groups had more centralities in each lobe. The number of SA centralities in the AD group was twice that of bvFTD (14 vs. 7). Surprisingly, the SA centralities in the bvFTD group were more numerous in the temporal lobe than in the frontal lobe (4 vs. 1), while in AD subjects, the frontal centralities were more numerous than the temporal centralities (6 vs. 4). Compared with 1 centrality in bvFTD subjects, AD subjects had 3 centralities in the parietal lobe.

[0275] In all three groups, central points in the inverse correlation SA network were only present in the frontal or temporal lobes. However, the HE group had a central point in the parietal lobe (precuneus), and bvFTD had two central points (inferior parietal lobule and paracentral gyrus) (see Table A.5 in Appendix A). Interestingly, most of the inverse correlation SA central points in AD were found in the frontal lobe.

[0276] Cortical thickness - cortical surface area coupling topology

[0277] The coupling strength between CT and SA nodes was calculated by element - by - element multiplication of the corresponding CT and SA correlation matrices. Figure 13 Shows the CT / SA coupling strength visualized in brain space. It can be observed that in HE subjects, inter - hemispheric homolog pairs showed coupled CT / SA correlations. In contrast, the CT / SA couplings in the AD and bvFTD groups were very similar to each other and different from the HE group. Both the bvFTD and AD groups showed more couplings between non - homologous nodes in the ipsilateral and contralateral hemispheres. Between the bvFTD and AD groups, the inter - lobar correlations were also significantly different. In the bvFTD group, most of the inter - lobar CT / SA correlations were due to fronto - temporal interactions. In AD, most of the inter - lobar CT / SA couplings were due to fronto - parietal interactions. A list of central points of the CT / SA coupling topology is given in Table A.6 of Appendix A.

[0278] Discussion

[0279] The baseline structural connectivity networks of subjects clinically diagnosed with bvFTD or AD have been examined in three large global clinical trials and compared to healthy elderly subjects in a well-characterized birth cohort. For each group, networks were constructed from partial correlations between 68 x 68 cortical surface regions (nodes) in terms of their thickness and surface area. The approach taken allowed for a systematic analysis of both positive and inverse network correlations in three clinical contexts. The methods and data discussed in this paper represent the first systematic comparative analysis of cortical thickness and surface area in a large subject population. Due to the need for comparable numbers in the three groups, the overall study size was determined based on the number of available bvFTD subjects. Since this is a rare disease, the bvFTD component of the study had to be global, with patients entering the study from 70 trial sites in 13 countries. 213 patients were included in the study, representing the largest MRI scan dataset of available bvFTD subjects to date. To match this, 213 patients were randomly selected from a much larger group of 1131 AD patients from 116 sites in 12 countries in study TRx-237-005 and 128 sites in 16 countries in study TRx-237-015 (accessible, for example, from the US National Library of Medicine). 202 normal elderly subjects were from a well-characterized birth cohort in which longitudinal studies had been conducted. Thus, the findings reported are robust and can be considered to represent an international group meeting accepted diagnostic criteria.

[0280] Networks according to the modularity of brain lobes

[0281] It has been shown that structural connectivity in the frontal, temporal, parietal, and occipital partitions of the cortical surface is inherently modular for both cortical thickness and surface area networks. That is, the results confirm that the standard lobar partitions of the cortex share a common network modularity property such that they differ from what would be expected in comparable random networks. The modules of highly clustered networks confer so-called "small-world" network properties and are thought to provide an optimal balance between local specialization and global integration. The results from healthy elderly subjects are comparable to previous work in smaller and younger healthy groups, revealing a potential modular architecture in regional thickness-related networks. The results also show that the inherent lobar modularity persists in both bvFTD and AD, indicating preservation of the overall lobar architecture of the network in the presence of neurodegenerative changes. As further discussed below, this contrasts with the hub-like organization of networks that change in a disease-specific manner.

[0282] Similarities and differences between AD and bvFTD subjects relative to healthy elderly subjects

[0283] The morphological correlation networks were found to be different from those of healthy elderly subjects in a highly significant manner in both patient groups (bvFTD and AD). Compared with healthy elderly subjects, both groups showed a significant increase in the overall correlation strength of the thickness and surface area networks. This effect was more significant in the cortical thickness networks of all lobes for both positive and inverse correlations. This contrasts with the significantly lower correlation strength of frontal lobe surface area in AD relative to normal and the directional similar differences in bvFTD. This may be due to a larger number and broader frequency distribution of correlations in the diseases compared with the sparser networks with a narrower frequency distribution in healthy elderly subjects. In addition to the increased overall correlation strength, the number of intra-lobe positive and inverse correlations measured by node degree was higher in both demented groups in all lobes than in healthy elderly controls. The number of inter-lobe positive correlations in thickness measured by the inter-lobe participation index was also higher in all lobes. The number of intra-lobe and inter-lobe surface area positive correlations was also greater in both bvFTD and AD than in healthy elderly subjects in the frontal, temporal, and parietal lobes. In terms of the correlations in the coupling between cortical thickness and surface area, both disease groups also differed from healthy elderly subjects. Thus, the two diseases are characterized by an overall increase in the strength and degree of structural correlations that are present both locally within lobes and globally between lobes.

[0284] The similarity between the two disorders in terms of the significant increase in the overall strength and degree of structural correlations might seem to call into question the clinical differences between bvFTD and AD, based on which the subjects were classified in the study. In fact, there were no differences between the two disorders in terms of overall cortical thickness and surface area. However, there were multiple important network differences between the two disorders. In the cortical thickness network, in the frontal and temporal lobes, the overall positive correlation strength was greater in bvFTD than in AD, and in the frontal lobe, the inverse correlation strength was also greater in bvFTD than in AD. In the parietal and occipital lobes, the number of significant intra-lobe positive correlations was higher in bvFTD than in AD. Conversely, in the frontal and parietal lobes, the number of positive and inverse intra-lobe correlations was greater in AD than in bvFTD. Most of the inverse correlations in cortical thickness and surface area were related to the inter-hemispheric non-homologous fronto-temporal in bvFTD and the fronto-parietal in AD.

[0285] The hub-like tissue of the relevant networks also differs significantly between these two disorders. While network connection hubs are thought to provide network integration, local hubs, in contrast, provide network differentiation. It has been proposed that hubs provide resilience to injury in neurodegenerative disorders. Alternatively, it has been proposed that hubs represent sites with specific vulnerabilities. Therefore, there is interest in studying how hubs change in the context of neurodegenerative diseases. bvFTD is characterized by an increase in the number of cortical thickness hubs in the frontal lobe and a decrease or elimination of hubs in the temporal, parietal, and occipital lobes. In contrast, AD is characterized by a distribution of hubs across all lobes compared to bvFTD, a decrease in the number of hubs in the frontal cortex, and an increase in hubs in the temporal and occipital lobes. In the positive correlation networks of surface area, the overall number of hubs in AD subjects is twice that of bvFTD, and the topology of these hubs is different. Thus, overall, AD is characterized by a much more distributed pattern of hubs in both thickness and surface area degeneration networks compared to bvFTD. In contrast, in bvFTD, the hub-like tissue is much more localized. It has been controversial that bvFTD is a clinical syndrome with focal but heterogeneous atrophy centered on hubs. Identifying the insular region as one of the inverse network hubs (for the CT network, in both the bvFTD and AD groups) is consistent with recent unexpected findings from diffusion MRI that hub-like fiber connectivity of the insula is increased in bvFTD. On the other hand, the hubs in the healthy elderly group are highly connected in a homologous manner within and between lobes and are not otherwise interlinked. The differences in hub-like tissue between AD and bvFTD indicate differences in the hierarchy of nodal vulnerability and differences in the organization of different compensatory network adaptations in these two disorders. Thus, unlike the preservation of lobe modularity in neurodegenerative diseases, there is no preserved hub-like tissue, meaning that pre-existing hubs are not an inherent structural property of cortical network organization.

[0286] Although AD is also characterized by changes in cortical thickness, these changes are not as pronounced overall as in bvFTD, while changes in surface area are more prominent in AD, indicating coordinated changes in the number of adjacent affected columns. These differences are consistent with the pathology in bvFTD that affects interneurons and astrocytes with more restricted connectivity. The surface area-related predominance in AD is consistent with the pathology that predominantly affects the principal cell-mediated long-range corticocortical projection system. bvFTD differs from AD in several important respects: there is no cholinergic deficit in bvFTD, treatment with acetylcholinesterase inhibitors or memantine has no therapeutic benefit, bvFTD is characterized by prominent astrocytic pathology, the affected neurons in the neocortex are mainly in layers II and VI (in AD, the pyramidal cells in layers III and V are mainly affected) and the spiny interneurons in the hippocampal dentate gyrus (in AD, the affected neurons are in CA 1-4 rather than the dentate gyrus), and bvFTD is characterized by elevated glutamate levels in the neocortex, but AD is not. However, neither of these disorders provides a simple explanation for the different distribution patterns of the relevant structural changes described herein.

[0287] Global features and significance of network changes in dementia

[0288] The overall situation that emerges in these two disease groups studied is that, for both positive and inverse correlations, network architecture changes in a coordinated manner across the whole brain. This is surprising because the neurodegenerative processes in these two disorders are generally thought to be anatomically restricted to the frontal and temporal lobes (in the case of bvFTD) and the temporal and parietal lobes (in AD). Instead, network analysis indicates that there are changes in cortical thickness and surface area networks that globally affect all lobes in both disorders, but the anatomical topology of these changes is different. It is known that both tau protein and TDP-43 aggregation pathologies spread in a prion-like manner, whereby the pathology in an affected neuronal population can initiate pathology in connected but previously unaffected neuronal populations. Thus, positive correlations may in part reflect the spread of pathology in existing normal networks, whereby existing functional networks are affected or spared together. Alternatively, such correlations may represent functional dependencies, such that the loss of function of one member in a dyadic relationship results in a parallel loss of function in the dyad, typically with the affected nodes being functionally synchronous. This explanation is consistent with previous work on cortical thickness correlations in healthy adults, where positive correlations were found to be consistent with diffusion-based axonal connections.

[0289] The work discussed herein highlights for the first time the importance of the inverse correlation network. It should be noted that since the inverse correlations observed in two neurodegenerative disorders mainly reflect inter-lobar non-homologous associations, they cannot be detected using only lobe-based analysis methods. In particular, the emergence of these non-homologous inter-lobar inverse correlations and their increased strength represent the clearest overall difference between neurodegenerative diseases and normal aging. In contrast, the brains of normal aging are characterized by significantly weaker homologous positive correlations. An attractive hypothesis is that when some nodes are functionally impaired, other still unaffected nodes compensate, thus emphasizing the non-homologous associations in the disease. This would imply that the major reorganizations observed in the structural network may be partially adaptive. Structural plasticity has been confirmed in other cases and functional compensation is known to occur in focal diseases.

[0290] The work discussed herein represents the first comparative study of related structural network abnormalities in bvFTD and AD relative to healthy aging. These correlations stem from both positive and inverse coupled changes in cortical thickness and surface area in these two pathologies, which are quite different from those observed in normal elderly subjects. The changes observed in the disease are global and are not restricted to the frontotemporal and temporoparietal lobes in bvFTD and AD, respectively. Instead, they seem to represent structural adaptations to different neurodegenerations in these two pathologies. Moreover, all related networks show a rather unique hub-like organization, which is different from the normal situation and also between these two forms of dementia. Different from the lobe organization of the networks that remain unchanged in the disease, the hub-like organization varies with the underlying pathology. This means that the hub-like organization is not a fixed feature of the brain and attempts to explain the disease in terms of hubs may be insufficient. The differences documented between AD and bvFTD confirm that the clinical differences between these two dementia groups correspond to systematic differences in the underlying network structure of the cortex. The topological differences in the hub-like organization of thickness and surface area and the underlying positive and inverse correlation networks can provide a basis for developing analytical tools to assist in differential diagnosis in these two pathologies, which may be difficult to distinguish by clinical criteria alone.

[0291] Use of correlation matrices in determining the response of patient groups to neuropharmacological interventions

[0292] The methods discussed above have been used to determine the response of patient groups to neuropharmacological interventions.

[0293] Figures 17A to 17D Correlation matrices are depicted for the following two patient groups: those being treated with symptomatic AD medications (cholinesterase inhibitors and / or memantine, ach1 in the figure caption) and those not undergoing such treatment (ach0 in the figure caption). The Clinical Dementia Rating (CDR) scores of the subjects range from 0.5, 1, or 2. Figure 17Ais the cortical thickness correlation matrix at baseline (i.e., week 0) for 96 subjects diagnosed with AD who were not receiving one or more symptomatic treatments. In contrast, Figure 17B is the cortical thickness correlation matrix at baseline for 445 subjects diagnosed with AD who were receiving one or more symptomatic treatments. Figure 17C is the surface area correlation matrix at baseline for 96 subjects diagnosed with AD who were not receiving one or more symptomatic treatments, and Figure 17D is the surface area correlation matrix at baseline for 445 subjects diagnosed with AD who were receiving one or more symptomatic treatments.

[0294] As can be observed from Figures 17A to 17D , compared to untreated patients ( Figure 17A and Figure 17C ), one or more symptomatic treatments for AD induced a significant increase in interlobar non-homologous inverse correlation networks ( Figure 17B and Figure 17D , blue). This was particularly significant for the surface area networks.

[0295] These connections represent inverse correlations whereby a decrease in the volume or surface area of affected regions in specific nodes (usually located in the posterior brain) is statistically significantly associated with linked nodes in which there is a corresponding increase in volume or surface area. As discussed above, the presence of these non-homologous inverse correlations indicates a neurodegenerative disease and most likely represents frontal compensation for posterior dysfunction arising from the pathology. Symptomatic AD treatment induces an increase in these non-homologous compensatory linkages.

[0296] Figures 18A to 18D are graphs of non-homologous interlobar nodal degrees (as discussed above) for cortical thickness - positive correlation, cortical thickness - inverse correlation, surface area - positive correlation, and surface area - inverse correlation, respectively. As can be observed from these graphs, the number of significant non-homologous interlobar compensatory inverse correlations increased significantly with symptomatic AD treatment.

[0297] Figure 19A and Figure 19B show cortical thickness correlation matrices based on structural neurology data separated in time. Figure 19A is the cortical thickness correlation matrix at week 0 (i.e., at baseline) for a group of 445 AD-diagnosed patients being treated with symptomatic AD treatment. Figure 19Bis the cortical thickness correlation matrix of 445 AD-diagnosed patients in the same group at week 65. During the intervention period, the group was also treated with the tau protein aggregation inhibitor methylthioninium chloride (LMTM; USAN name: methylthioninium chloride monohydrate) at a dose of 8 mg / day (administered twice daily at 4 mg each time from then on). As can be observed, in patients receiving symptomatic treatment for AD, the effect of LMTM on the structure-related network is minimal.

[0298] Figures 20A to 20D is a graph of the inter-lobe node degrees between week 0 and week 65 in the ach1 group (while undergoing symptomatic AD treatment), respectively for cortical thickness - positive correlation, cortical thickness - negative correlation, surface area - positive correlation, and surface area - negative correlation. As can be observed, over 65 weeks, the overall effect of LMTM as an add-on on the brain network-related structure is minimal. Importantly, note that this is an in-cohort analysis, whereby for the changes occurring after 65 weeks of treatment with LMTM, the patients at baseline served as their own controls.

[0299] Figure 21A and Figure 21B shows the cortical thickness correlation matrix based on structurally neurological data separated in time. Figure 21A is the cortical thickness correlation matrix of a group of 96 AD-diagnosed patients taking LMTM at a dose of 8 mg / day as monotherapy at week 0 (i.e., at baseline). Figure 21B is the cortical thickness correlation matrix of the same group of 96 AD-diagnosed patients at week 65. The 96 patients in this cohort did not receive one or more symptomatic AD treatments in combination with LMTM. As can be observed, LMTM as monotherapy significantly reduced the thickness-related (both intra-lobe (positive) and inter-lobe compensatory (negative) correlations). This is an in-cohort analysis, whereby for the changes occurring after 65 weeks of treatment with LMTM, the patients at baseline served as their own controls.

[0300] Figure 22A and Figure 22B is a graph of the inter-lobe node degrees between week 0 and week 65 in the ach0 group, for cortical thickness - positive correlation and cortical thickness - negative correlation. The graph indicates a highly significant effect of LMTM at 8 mg / day as monotherapy on the number of inter-lobe correlations in the AD group. A significant reduction in the number of positive and negative non-homologous cortical thickness correlations was observed after 65 weeks. This may be due to the normalization of neuronal function in the posterior brain, whereby LMTM reduces pathology and alleviates neuronal dysfunction arising from the pathology, thus reducing the need for compensatory input from the unaffected or less affected frontal regions of the brain.

[0301] Figure 23A and Figure 23B shows a surface area thickness correlation matrix based on structurally neurological data separated in time. Figure 23A is the surface area correlation matrix at week 0 (i.e., baseline) for a group of 96 AD-diagnosed patients who had been continuously taking LMTM at a dose of 8 mg / day as monotherapy. Figure 23B is the surface area correlation matrix at week 65 for the same group of 96 AD-diagnosed patients. None of the 96 patients in this cohort were undergoing one or more symptomatic AD treatments concurrently. As can be observed, LMTM as monotherapy significantly reduced surface area correlations (both intra-lobar (positive) and inter-lobar compensatory (negative)). This is an in-cohort analysis, whereby the patients at baseline were used as their own controls for changes occurring after 65 weeks of treatment with LMTM.

[0302] Figure 24A and Figure 24B are graphs of inter-lobar node degrees between week 0 and week 65 in the ach0 group, for surface area - positive and surface area - negative correlations. The graphs indicate a significant effect of LMTM at 8 mg / day as monotherapy on the number of inter-lobar correlations in the AD group. Notably, there was a significant reduction in the number of positive and inverse / compensatory surface area correlations after 65 weeks.

[0303] Figures 25A to 25D shows a cortical thickness correlation matrix comparing the AD group (CDR 0.5, 1, or 2) of 96 patients at baseline and week 65 with a healthy elderly control group of 202 subjects. As can be observed, LMTM at 8 mg / day as monotherapy made the cortical thickness network closer to normal.

[0304] Figures 26A to 26D shows a surface area correlation matrix comparing the AD group (CDR 0.5, 1, or 2) of 96 patients at baseline and week 65 with a healthy elderly control group of 202 subjects. As can be observed, LMTM at 8 mg / day as monotherapy normalized the surface area network.

[0305] Figures 27A to 27D shows a cortical thickness correlation matrix comparing the AD group (CDR only 0.5) of 54 patients at baseline and week 65 with a healthy elderly control group of 202 subjects. As can be observed, LMTM at 8 mg / day as monotherapy reduced the number of inverse / compensatory non-homologous correlations to become equivalent to normal elderly controls.

[0306] Figures 28A to 28DShows the surface area - related matrix comparing between the AD group (CDR only 0.5) of 54 patients at baseline and week 65 and 202 subjects in the healthy elderly control group. As can be observed, 8 mg / day of LMTM as monotherapy reduces the number of inverse / compensatory non - homologous correlations to equal or lower than that of normal elderly controls.

[0307] In summary, the structure - related network analysis discussed above reveals the emergence of highly abnormal inverse correlations between non - homologous lobes in AD and bvFTD. It is hypothesized that these represent compensatory inputs from frontal brain regions that are not or less affected by the disease. Symptomatic treatment and LMTM act in fundamentally different ways on the structure - related network in AD. Symptomatic treatment induces a significant increase in compensatory networks. LMTM as monotherapy reduces the need for these compensatory networks by reducing the primary pathology and thus allowing affected neurons to function more normally. These results confirm that the abnormal non - homologous inverse correlations observed in neurodegenerative diseases such as AD are adaptive, as they can be reversed or attenuated by disease - modifying treatments rather than by symptomatic AD treatment. The effects were observed within the cohort before / after analysis, in which subjects at baseline were used as their own controls for changes occurring after 65 weeks of treatment with 8 mg / day of LMTM as monotherapy. These analyses are far more sensitive to treatment effects than gross whole - brain or lobar volume analyses. Furthermore, as will be discussed below, the results observed in the structure - related network are consistent with the functional effects observed by the renormalized partial directed coherence electroencephalography analysis technique.

[0308] Structure / function correlations using electroencephalography (EEG)

[0309] The renormalized partial directed coherence (rPDC) network method for EEG data allows indication of the direction and strength of intracerebral electrical activity to be studied using network methods. This is discussed, for example, in WO 2017 / 118733 (the entire content of which is incorporated herein by reference). Figure 29 Shows an example of the raw EEG data, and Figure 30 Shows an exemplary rPDC network obtained from the collected EEG data.

[0310] As Figure 30 The resulting network shown in contains multiple nodes indicating approximate locations within the brain (the figure is drawn in a schematic top - down view of the head, and the triangle at the top indicates the nose). The positions of the nodes are determined by placing electrodes on the scalp surface, the electrodes being used to obtain the EEG data as shown in Figure 29 the EEG data. The directed connections between the nodes indicate the flow of electrical activity within the brain from one node to another.

[0311] By counting the number of directed connections entering and leaving a given node and / or measuring their relative strength, a node can be defined as a sink (and there are more and / or stronger incoming connections than outgoing connections) or a source (and there are more and / or stronger outgoing connections than incoming connections). This is schematically shown in Figure 31 where the number / strength of incoming directed connections is subtracted from the number / strength of outgoing directed connections. Thus, in the extreme case, if the difference is negative, the node acts as a net source, and if the difference is positive, the node acts as a net sink. More generally, as can be observed in the figure, as shown in Figure 40 a lower value indicates more / stronger outgoing connections, and a higher value indicates more / stronger incoming connections.

[0312] After deriving the difference between the incoming and outgoing connections for all nodes, a heatmap can then be provided to indicate the location and strength of sinks and sources within the patient's brain. This can include the step of defining each node as a sink or a source. An example of such a heatmap is shown in Figure 32 In this example, the blue area (arrow A) indicates more outgoing connections and thus contains more source nodes, while the red / yellow area (arrow B) indicates more incoming connections and thus contains more sink nodes. This type of heatmap can be referred to as a "brainprint".

[0313] Figure 33 illustrates Figure 32 the visualization of asymmetry in the heatmap of

[0314] where the number of sources and sinks on either side is compared. A higher difference in sources and sinks between the left and right sides of the heatmap is shown in yellow (arrow A), while a lower difference is shown in black (arrow B).

[0315]

[0316] MMSE - Mini - Mental State Examination; ADAS - Cog - Alzheimer's Disease Assessment Scale - Cognitive Subscale

[0317] As can be observed, the diagnosed subjects were significantly more severely cognitively impaired on the MMSE and ADAS - Cog psychometric scales and also scored higher on the overall Clinical Dementia Rating (CDR) scale. In other respects, there were no differences in age or gender distribution.

[0318] Figure 34Shows a heatmap visualizing the locations of the brain's sinks and sources in a group of subjects diagnosed at baseline. Arrow A indicates the blue area containing more / stronger sources, and arrow B indicates the red area containing more / stronger sinks. Figure 35 Shows a heatmap visualizing the locations of the brain's sinks and sources in a group of paired volunteers. When comparing the two images, it becomes clear that AD patients have significantly stronger sources (i.e., more / stronger outgoing connections, shown in blue) in their frontal lobes and significantly stronger sinks (i.e., more / stronger incoming connections, shown in red / orange) in their posterior parietal, temporal, and occipital lobes compared to paired volunteers.

[0319] A machine learning classifier is trained based on the dataset provided by the 329 subjects discussed above. In each case, rPDC networks are prepared using β-band EEG data from 100 seconds of brain activity during the eyes-closed resting state. Then, the machine learning classifier classifies all 329 subjects as either AD or paired volunteers (PV) with 95% accuracy. Additionally, the probability that a subject has AD can be estimated using the machine learning classifier, thus allowing for more than just a binary decision. For example, the heatmap shown in Figure 36 The subject in has AD. The patient is known to have AD through clinical diagnosis. The machine learning classifier estimates with 99% probability that the patient has AD and thus correctly classifies the subject. Figure 37 Is another example of a heatmap from a subject known to have AD through clinical diagnosis. In this case, the machine learning classifier estimates that the patient has a 63% probability of having AD and a 37% probability of not having AD. This information can be used to determine a patient's susceptibility to AD where no clinical diagnosis has yet been made. Additionally, specific distribution patterns of abnormal sink regions indicating potential dysfunction can be correlated with specific patterns from clinical tests for future, more detailed neuropsychological testing and clinical evaluation. For example, Figure 36 The case illustrated in may have a form of dementia other than AD but is classified as having AD in this article.

[0320] Psychometric testing of a seemingly healthy cohort shows a downward cognitive trajectory on the Hopkins Verbal Learning Test over 18 months in a subset of the subjects. The characteristics of the cohort are as follows. As can be observed, there is no difference in cognitive scores at baseline on the MMSE scale between those subjects found to be at risk of decline and those found not to be at risk of decline.

[0321]

[0322] The heatmap of the at-risk subject group is shown in Figure 38while the heatmap of the risk-free subject group is shown in Figure 39 Since both groups were drawn from a superficially healthy cohort, the differences here are not as obvious as those between the AD and PV groups above. Figure 38 showing more / stronger sinks in the posterior brain regions, manifested as stronger red / orange in the heatmap. Figure 40 is a box plot comparing sources and sinks in the EEG networks from the frontal and posterior brain regions at the group level. As can be observed from this figure, the at-risk group is characterized by increased activity exiting the frontal cortex and increased activity entering the posterior brain regions. At baseline, EEG recordings were made before any measurable decline on the Hopkins Word Learning Task. Thus, apparently normal subjects at risk of decline over the next 18 months may have been identified at baseline based on the heatmap of their brain activity non-invasively obtained through EEG analysis.

[0323] As shown above, there are significant differences in the networks between the diagnosed subjects and the paired volunteers. These differences are highly significant at the group level. As will be appreciated, the accuracy level of the first version of the machine learning classifier is higher than that of conventional superficial clinical assessments and gives the probability of having AD at the individual subject level, which can be used for decision-making in further clinical management.

[0324] Figure 41 shows a comparison between the cortical thickness correlation matrix of the ach0 AD group at week 0 (discussed above) and the heatmap of the group-level diagnosed subjects for comparison. As can be observed, there is a large amount of non-homologous inverse correlation between the frontal lobe and the posterior parietal and occipital brain regions. The heatmap of the diagnosed subject group shows the same phenomenon in terms of brain connectivity as measured by EEG. Both the structural and EEG methods show the same pattern of increased frontal to posterior activity. Figure 42 shows a comparison between the cortical thickness correlation matrix of the healthy elderly group at week 0 (as discussed above) and the heatmap of the group-level paired volunteers for comparison. As can be observed, there is no non-homologous inverse correlation between the frontal lobe and the posterior parietal and occipital brain regions, which matches the absence of increased anterior to posterior electrical activity on EEG.

[0325] Figure 41 The increase in the non-homologous inter-lobe compensatory inverse correlation network observed in

[0326] Figure 43 shows a box plot demonstrating the quantitative discrimination between mild AD and elderly controls. As can be observed, in the beta band, AD subjects have more activity exiting the frontal cortex and more activity entering the posterior cortex.

[0327] Figure 44Three heatmaps are shown. From left to right, they are: a group of diagnosed AD subjects who have been treated with symptomatic medications (medication (med)), a group of diagnosed AD subjects who have not been treated with symptomatic medications (no medication (nonMed)), and a group of paired volunteers. Figure 45 Is a box plot comparing the group-level networks of the medicated, non-medicated, and paired volunteer groups. The data is from a pilot study that included 53 diagnosed subjects (DS), 15 of whom received standard medication treatment and 38 of whom did not. The characteristics of the two groups are shown in the table below. Although the non-medicated group was significantly younger, there were no differences between the two groups in cognitive scores as measured by MMSE or in gender distribution.

[0328] Medication No Medication Age 75.2(7.6) 67.5(9.5)*** Gender 8F:7M 23F:15M MMSE 23.84(1.95)(N＝13) 23.37(2.41)

[0329] ***p < 0.005

[0330] There were also no statistically significant differences in the ADAS-Cog or CDR scales.

[0331] As can be observed in Figure 44 and Figure 45 both groups of AD subjects had more activity exiting the frontal cortex in the beta band compared to the paired volunteers. Additionally, it was observed that one or more symptomatic treatments increased the activity exiting the frontal lobe compared to the non-medicated group. This is shown in the box plot of Figure 45 The medicated group had significantly more electrical activity exiting the frontal cortex. In the posterior brain regions, one or more symptomatic treatments reduced the need for supportive incoming electrical activity.

[0332] The frontal lobe showed the same phenomenon via EEG as Figures 17A to 17D and Figures 18A to 18D shown by the structural analysis of the relevant networks in. Figure 44 Shows that differences detectable by MRI structural analysis at the group level can also be detected by EEG. It should be noted that while the structural analysis of the network differences between patients receiving and not receiving one or more symptomatic treatments indicates an increase in non-homologous inter-lobe connectivity pointing to the posterior brain regions, the EEG analysis shows less incoming activity pointing to the posterior regions. It is currently hypothesized that in other frequency bands, one or more symptomatic treatments increase the activity entering the posterior brain regions.

[0333] In addition to the described structural components and user interactions, the systems and methods of the above embodiments can be implemented in a computer system (specifically, in computer hardware or in computer software).

[0334] The term "computer system" includes the hardware, software, and data storage devices used to embody the system or execute the method according to the above embodiments. For example, a computer system may include a central processing unit (CPU), input components, output components, and a data memory. Preferably, the computer system has a monitor to provide a visual output display. The data memory may include RAM, disk drives, or other computer-readable media. The computer system may include multiple computing devices that are connected via a network and capable of communicating with each other via the network.

[0335] The method of the above embodiments may be provided as a computer program or as a computer program product or a computer-readable medium carrying a computer program, which when run on a computer is configured to execute one or more of the above methods.

[0336] The term "computer-readable medium" includes, but is not limited to, any one or more non-transitory media that can be directly read and accessed by a computer or computer system. The media may include, but is not limited to, magnetic storage media such as floppy disks, hard disk storage media, and magnetic tapes; optical storage media such as optical discs or CD-ROMs; electrical storage media such as memories including RAM, ROM, and flash memories; and mixtures and combinations of the above storage media such as magnetic / optical storage media.

[0337] Although the present invention has been described in connection with the above exemplary embodiments, many equivalent modifications and changes will be apparent to those skilled in the art from the present disclosure. Accordingly, the above exemplary embodiments of the present invention are considered illustrative rather than restrictive. Various changes may be made to the embodiments without departing from the spirit and scope of the present invention.

[0338] In particular, although the method of the above embodiments has been described as being implemented on the system of the embodiments, the method and system of the present invention need not be implemented in combination with each other, but may be implemented separately on alternative systems or using alternative methods.

[0339] Appendix A

[0340] Table A.1 - Cortical surfaces of the frontal, temporal, parietal, or occipital regions according to the Desikan-Killiany atlas (DKA). The cortical regions (nodes) of each structural correlation matrix are sorted throughout the document according to the following list:

[0341]

[0342] Table A.2 - Assignment of nodes to the frontal, temporal, parietal, or occipital lobes by the algorithm and by DKA cortical segmentation. *For the cortical thickness network, nodes are misassigned to lobes; and + For the surface area network, the nodes are those that are misassigned

[0343]

[0344]

[0345] Abbreviations: F - frontal lobe, T - temporal lobe, P - parietal lobe,

[0346] O - occipital lobe, L - left, R - right.

[0347] Table A.3 - Mean cortical thickness (CT) and total surface area (SA) averaged across four brain lobes in each study group

[0348]

[0349] Abbreviations: HE - healthy elderly, bvFTD - behavioral variant frontotemporal dementia,

[0350] AD - Alzheimer's disease

[0351] Table A.4 Central points of CT network frontal, temporal, parietal, and occipital lobe modular organization in HE, bvFTD, and AD. Central points are ranked according to their inter - lobe participation index (p) and intra - lobe z - score (z). High p / high z scores indicate so - called integrative regions (i.e., nodes that interact between all brain lobes), and regions with low p / high z are so - called local central points (i.e., nodes that interact within their own module / lobe).

[0352] a) Central points of the positive sub - network

[0353]

[0354] b) Central points of the negative sub - network

[0355]

[0356]

[0357] Abbreviations: HE - healthy elderly, bvFTD - behavioral variant frontotemporal dementia, AD - Alzheimer's disease, F - frontal lobe, T - temporal lobe, P - parietal lobe, O - occipital lobe

[0358] Table A.5 Central points of SA network frontal, temporal, parietal, and occipital lobe modular organization in HE, bvFTD, and AD. Central points are ranked according to their inter - lobe participation index (p) and intra - lobe z - score (z). High p / high z scores indicate so - called integrative nodes (i.e., nodes that interact between all brain lobes), and regions with low p / high z are so - called local central points (i.e., nodes that interact within their own module / lobe).

[0359] a) Central points of the positive subnetworks

[0360]

[0361]

[0362] b) Central points of the negative subnetworks

[0363]

[0364] Abbreviations: HE - Healthy elderly, bvFTD - Behavioral variant frontotemporal dementia, AD - Alzheimer's disease, F - Frontal lobe, T - Temporal lobe, P - Parietal lobe, O - Occipital lobe.

[0365] Table A.6 Central points of the CT - SA coupling networks in the frontal, temporal, parietal, and occipital lobes of HE, bvFTD, and AD. Regions are ranked according to their inter - lobe participation index (p) and intra - lobe z - score (z). High p / z scores indicate so - called integrative regions (which interact between all lobes)

[0366] a) Central points of the positive subnetworks

[0367]

[0368]

[0369] Abbreviations: HE - Healthy elderly, bvFTD - Behavioral variant frontotemporal dementia, AD - Alzheimer's disease, F - Frontal lobe, T - Temporal lobe, P - Parietal lobe, O - Occipital lobe.

[0370] References

[0371] Gauthier,S.et al.“Efficacy and safety of tau - aggregation inhibitortherapy in patients with mild or moderate Alzheimer’s disease:a randomised,controlled,double - blind,parallel - arm,phase 3trial”,The Lancet 388,2873 - 2884(2016)

[0372] Wilcock,G.K.et al“Potential of low dose leuco-methylthioninium bis(hydromethanesulphonate)(lmtm)monotherapy for treatment of mild Alzheimer’sdisease:Cohort analysis as modified primary outcome in a phase iii clinicaltrial.Journal of Alzheimer’s disease 61,635-657(2018)

[0373] Feldman,H.et al“Aphase 3trial of the tau and tdp-43aggregationinhibitor,leuco-methylthioninium bis(hydromethanesulfonate)(lmtm),forbehavioural variant frontotemporal dementia(bvFTD)”Journal of Neurochemistry138,255(2016)

[0374] Murray,A.D.et al“The balance between cognitive reserve and brainimaging biomarkers of cerebrovascular and Alzheimer’s diseases”Brain 134,3687-3696(2011)

[0375] Storey,J.D.“A direct approach to false discovery rates”Journal of theRoyal Statistical Society:Series B(Statistical Methodology)64,479-498(2002)

[0376] Van Wijk,B.C.,Stam,C.J.&Daffertshofer,A.“Comparing brain networks of different size and connectivity density using graph theory.”PLoS One 5,e13701(2010)

[0377] Rubinov M,Sporns O.“Weight-conserving characterization of complex functional brain networks”,Neuroimage,56(4):2068-79(2011)

[0378] All of the references mentioned above are hereby incorporated by reference.

Claims

1. A method for determining the likelihood that a patient has one or more neurological disorders, the method comprising the steps of: obtaining data indicative of electrical activity in the brain of the patient; generating, at least in part based on the obtained data, a network comprising a plurality of nodes and directed connections between the nodes, wherein the network indicates the flow of electrical activity in the brain of the patient; calculating, for each node, the difference between the number and / or strength of the connections entering the node and the number and / or strength of the connections leaving the node; and using the calculated differences to determine the likelihood that the patient has one or more neurological disorders.

2. The method according to claim 1, wherein the network is a renormalized partial directed coherence network.

3. The method according to claim 1 or claim 2, wherein the data indicative of electrical activity in the brain is electroencephalography data.

4. The method according to claim 3, wherein the electroencephalography data is beta-band electroencephalography data.

5. The method according to any one of claims 1-4, wherein determining the susceptibility of the patient is performed using a machine learning classifier.

6. The method according to any one of claims 1-5, the method further comprising the step of generating a heat map, at least in part based on the state of the nodes, the heat map indicating the location and / or strength of the nodes defined as sinks and the nodes defined as sources in the brain of the patient.

7. The method according to any one of claims 1-6, the method further comprising the step of using the state of the nodes to derive an indication of the degree of left-right asymmetry of the location and / or strength of the nodes in the brain corresponding to the sinks and sources.

8. The method according to any one of claims 1-7, wherein the neurological disorder is a neurocognitive disease, optionally Alzheimer's disease.

9. The method according to any one of claims 1-8, wherein the susceptibility of the patient to one or more neurological disorders is determined by comparing the number and / or strength of the nodes defined as sinks in the posterior lobe with a predetermined value, and / or by comparing the number and / or strength of the nodes defined as sources in the temporal lobe and / or frontal lobe with a predetermined value.

10. The method according to claim 9, wherein if the number and / or strength of the nodes defined as sinks in the posterior lobe exceeds a predetermined value and / or if the number and / or strength of the nodes defined as sources in the temporal lobe and / or frontal lobe exceeds a predetermined value, then it is determined that the patient has a high susceptibility risk.

Images