WO2025128714A1 - Transcranial electrical stimulation for treatment of epilepsy - Google Patents

Abstract

In various aspects and embodiments, the present disclosure provides a method for treating epilepsy that comprises applying transcranial electric stimulation (tES). Given the role of particular regions of the brain in epilepsy, which can vary from patient to patient, a more personalized approach to tES targeting each patient's specific epileptogenic region(s) and/or propagation networks is disclosed herein, for reducing the frequency, duration, or intensity of seizures.

Description

TRANSCRANIAL ELECTRICAL STIMULATION FOR TREATMENT OF EPILEPSY

BACKGROUND

Epilepsy is a neurological disorder in which brain activity becomes abnormal and causes seizures. Epileptic seizures are typically caused by a group of neurons firing in an abnormal, excessive, and synchronized manner which can spread to other regions of the brain. Many symptoms of epilepsy can be controlled by medication; however, drug-resistant epilepsy is extremely difficult to control and treat. Accordingly, methods for treatment of epilepsy, including drug-resistant epilepsy are needed.

DESCRIPTION OF FIGURES

FIG. 1 illustrates an exemplary pipeline for preparing a head model according to embodiments of the disclosure where SEEG seizure data and clinical information are used to personalize a head model. Such head models can be prepared with an MRI image to model passive tissues and dMRI data to model connectivity, for example. Neural mass models (NMM) are created and positioned as nodes in various regions of the brain. The NMM nodes are interconnected with the subject’s unique information (the subject’s “connectome”). The resulting epileptogenic node model (termed “proto-NeT” in FIG. 1) can be further personalized by tuning the connectivity parameters to match to the subject’s SEEG and/or EEG data. These head models can be used to simulate the subject’s seizures and their propagation, and to select tDCS montages that reduce the seizure (e g., probability of seizure) or their propagation (e.g., propagation probability) according to the simulations.

FIGS. 2A - 2C illustrate a non-limiting schematic of group head models and group optimization methods. FIG. 2A show a schematic of assembling a template model. FIG. 2B show a schematic of assembling and using a template model to create a loss function for optimization of a montage. FIG. 2C show a schematic of creating individual models for a subject and assimilating them into a loss function for optimization of a montage.

DETAILED DESCRIPTION

In various aspects and embodiments, the present disclosure provides a treatment for epilepsy that comprises applying transcranial electrical stimulation (tES) such as transcranial direct current stimulation (tDCS). Transcranial electrical current stimulation (tES, sometimes also called tCS), includes both direct and alternating current variants known as tDCS and tACS, and is a non-invasive subthreshold neuromodulatory technique (see Nitsche, M A, and W Paulus (2000). Low intensity, controlled currents (typically ~1 mA but <4 mA) are applied through scalp electrodes in repeated sessions. This subtle but persistent modulation of neuronal activity is believed to lead to plastic effects deriving from Hebbian mechanisms. That is, tES induces concurrent and plastic effects from persistent (in time), mesoscale (in space), weak electric fields acting on brain networks. The recent evolution of tES has delivered multichannel systems using small electrodes much like EEG. Transcranial direct current stimulation (tDCS) is a non-invasive brain stimulation treatment that uses direct electrical currents to modulate the excitability of specific targets in the brain. A constant, low-intensity current is passed through electrodes placed on the scalp to modulate underlying neuronal activity. Given the role of particular regions of the brain in epilepsy, which can vary from patient to patient, a more personalized approach to tDCS targeting each patient’s specific epileptogenic region(s) and/or propagation networks may prove more efficacious than conventional techniques in reducing the frequency, duration or intensity of her/his seizures.

Epilepsy is a neurological disorder characterized by recurrent epileptic seizures. A seizure is a paroxysmal alteration of neurologic function caused by the excessive, hyper synchronous discharge of neurons in the brain. Epileptic seizures can vary from brief and nearly undetectable periods to long periods of vigorous shaking due to abnormal electrical activity in the brain. In epilepsy, seizures tend to recur and may have no immediate underlying cause. Most cases of epilepsy are idiopathic. In some cases, epilepsy is known to occur as the result of brain injury, stroke, brain tumors, infections of the brain, or birth defects through a process known as epileptogenesis. Epileptogenesis is the process whereby a previously normal brain is functionally altered and biased towards the generation of the abnormal paroxysmal electrical activity that defines chronic seizures.

The major goal of epilepsy treatment is to interrupt or reverse epileptogenesis. In instances of epileptogenesis, treatment options may be limited when the chronic seizures do not respond to medication. According to the present disclosure, neurostimulation, such as tDCS, provides a non-invasive neuromodulation therapy that can be an effective epilepsy treatment, including in cases of drug-resistant epilepsy. tDCS is a type of transcranial stimulation where the stimulation currents are held constant except for a period of ramp-up/down at the beginning and end of stimulation. tDCS produces effects on neuronal excitability by increasing or decreasing the strength of the electric field. tDCS generates weak electrical currents and electric fields measured in volts per meter that modulate neuronal activity in the brain. The multichannel stimulation is designed to generate an electric field on the areas of interest to selectively modify the excitability of neuronal populations. The component of the electric field orthogonal to the cortical surface is of primary interest because of the form factor and orientation of the most sensitive cells in the cortex (pyramidal cells).

The basic mechanism for transcranial electrical current stimulation (tCS or tES) may be through the coupling of electric fields to elongated form-factor neurons such as pyramidal cells. Physically, the external electric field forces the displacement of intracellular ions (which move to cancel the intracellular field), altering the neuron’s internal charge distribution and, as a result, modifying the transmembrane potential difference. For a long, straight fiber with a space constant (mm) in a homogenous electric field, the transmembrane potential difference is largest at the fiber termination, with a value that can be approximated by X • n^A, where n^A is the unit vector defining the fiber axis. This is an expected first-order result, with a spatial scale provided by the membrane space constant and directions by field and fiber orientation. See Ruffini G. et al. (2014), and computational modeling studies in Galan et al. (2022).

In an aspect, the present disclosure provides a method for treating epilepsy in a subject, comprising, providing a transcranial electrical stimulation (tES) electrode montage designed to modify the excitability of a plurality of target regions of the subject’s cortex and/or subcortical regions, wherein the target regions comprise an epileptogenic region and/or a propagation network. Targets can further include regions in the rest of the cortex and may include subcortical regions. The electrode montage is selected for the subject using a head model. The method further comprises applying the tES to the subject with the electrode montage. According to this aspect, the disclosure provides a use of the electrode montage for treating epilepsy in a subject. Also according to this aspect, the disclosure provides a method for determining an electrode montage for tES therapy for a subject (e g., an epilepsy subject as described further herein), where the electrode montage is designed using one or more head models as described herein.

In aspects and embodiments, there is provided a method for treating epilepsy in a subject, comprising: providing a tES electrode montage, such as a tDCS electrode montage, designed to modify excitability of a plurality of target regions of the subject’s brain, wherein the target regions comprise targets for reducing the excitability of one or more of an epileptogenic region and a propagation network, wherein the electrode montage is selected for the subject using a head model; and applying the tES with the electrode montage. In embodiments, the target regions comprise cortical targets and/or subcortical targets. In embodiments, the target regions further comprise one or more targets outside the epileptogenic region and/or propagation network.

In aspects and embodiments, there is provided a method for treating epilepsy in a subject, comprising: providing a tES electrode montage, such as a tDCS electrode montage, designed to modify excitability of a plurality of target regions of the subject’s cortex and/or subcortical regions, wherein the target regions comprise an epileptogenic region, a propagation network. The electrode montage is selected for the subject using a head model, and applied to the subject for tES (e.g., tDCS).

According to this aspect, there is also provided a method for designing an electrode montage for an epilepsy subject, the electrode montage designed to modify excitability of a plurality of target regions of the subject’s brain, wherein the target regions comprise targets for reducing the excitability of one or more of an epileptogenic region and a propagation network. According to this aspect, the electrode montage is selected for the subject using one or more head models (as described further below). In embodiments, the method comprises providing a head model and determining an electrode montage that reduces probability of seizure or probability of seizure propagation according to simulations using the model. In embodiments, the target regions comprise cortical targets and/or subcortical targets. In embodiments, the target regions further comprise one or more targets outside the epileptogenic region and/or propagation network.

In embodiments, the target regions comprise electric field specifications for stimulation. In embodiments, the target regions comprise, or consist essentially of, or consist of electric field specifications for treatment of epilepsy in the subject, and/or may comprise physiological targets. In embodiments, the target regions are selected to inhibit excitability of regions that consist essentially of or consist of an epileptogenic region and/or propagation network. In this context, the term “consist essentially of’ means that changes in the electric field outside of the targets regions (due to tES) are not considered physiologically significant.

As used herein, the term “electrode montage” refers to a configuration of electrodes comprising the number, location, and type of electrodes, as well as electrode current.

In embodiments, the target regions for reducing excitation comprise, or consist essentially of, or consist of an epileptogenic region and a propagation network. As used herein, the epileptogenic region is the region of the brain from which the patient’s habitual seizures arise. In some embodiments, the epileptogenic region comprises an area of the cortex that generates interictal spikes. In some embodiments, the epileptogenic region comprises an area of the cortex that initiates clinical seizures. In some embodiments, the epileptogenic region comprises a symptomatogenic region, where the area of the cortex, when activated, produces the initial ictal symptoms or signs of a seizure. In some embodiments, the epileptogenic region comprises a macroscopic lesion that is causative of the epileptic seizures.

In some embodiments, the epileptogenic region comprises a functional deficit region of the cortex that is not functioning normally in the interictal period. For example, the epileptogenic region can be defined as the area of cortex that is necessary and sufficient for initiating seizures and whose removal (or disconnection) is necessary for complete abolition of seizures.

As used herein, “a propagation network” refers to populations of neurons connected to the epileptogenic region that allow seizures to spread to wider areas of the brain. In some embodiments, a propagation network is a functionally and anatomically connected set of cortical and subcortical brain structures and regions in which activity in any one part affects activity in all the others.

In some embodiments, the target regions for reducing excitability comprise, consist essentially of, or consist of targets within an epileptogenic network, which are brain regions involved in the production and propagation of epileptic activities. In various embodiments, the electrode montage reduces excitability of the plurality of target regions (such as the epileptogenic region and a propagation network, or collectively the epileptogenic network), without substantial or significant excitation outside the target regions.

In some embodiments, the targets can be defined by a physician. In embodiments, the target regions are defined be stereoelectroencephalography (SEEG). SEEG involves the surgical implantation of electrodes (e.g., up to 20 electrodes) into the brain to record electrical activity and localize the seizure focus. The electrodes are placed at brain areas where seizures are suspected to initiate. In embodiments, the target regions are identified by one or more of SEEG, EEG, MEG, PET and MRI according to known techniques.

In embodiments, the head model is created from an image or scan of the subject’s head, such as with one or a combination of MRI, CT, DW-MRI, fMRI, fNIRS, PET, rs- fcMRI, EEG, MEG, SEEG, and SPECT. In embodiments, head models can be biophysical head models that model passive tissues, or in embodiments can be hybrid models created from a combination of imaging modalities to model the subject’s unique brain dynamics. In embodiments, a CT scan or other imaging techniques (MRI) are used to ascertain the precise location of SEEG electrodes, which in some embodiments define at least some target regions.

In embodiments, the head model is created using an MRI, and the model represents the geometry of the patient’s head and their passive electrical properties. In embodiments, these models are built from automatic segmentation of Tlw-MRIs, or Tlw and T2w-MRIs (preferably with full head coverage), into different tissues, including but not limited to: scalp, skull, cerebrospinal fluid (CSF, including the ventricles), grey matter (GM), and white matter (WM). Certain types of lesions can also be represented in the head model, such as corti sectomies, strokes, brain tumors, skull scar tissue, and titanium implanted plates. In embodiments, target regions can be mapped to the head model for montage determination.

In embodiments, currents, electrode locations, and electrode numbers are determined for montages using such a head model with electric field modeling. See Ruffini et al., 2014. In such embodiments, tissue boundaries are derived from MR images (scalp, skull, cerebrospinal fluid (CSF) including ventricles, Grey Matter, and White Matter), and the finite element method is used to calculate the electric potential in the head, subject to the appropriate boundary conditions. In embodiments, tissues are assumed to be uniform and possibly isotropic (although anisotropy can also be modeled), and values for their electric conductivity are taken from the literature or measured using techniques such as electrical impedance tomography (EIT) or Magnetic Resonance Electrical Impedance Tomography (MREIT).

In embodiments, the head model comprises an electric field characteristic target map for the subject’s cortex, and includes the target regions and desired values for the electrical field at each of the target regions to modulate excitation of said target regions. In various embodiments, the head model further comprises a weight map for the cortical surface specifying a degree of relative importance for each of the target regions and the rest of the cortex for the purposes of modulating excitability.

As used herein, a “target map” defines desired values for the electric field at multiple spatial and temporal points for stimulation. Targets can be defined based on a coordinate system relative to the cortical surface, with target values for normal and/or tangential components of electric field to the cortex, or, more generally, by a spatiotemporal field in the brain.

The electrode montage can be identified using algorithms to optimize currents, for example, as well as the number and location of electrodes given appropriate constraints, such as the maximum current at any electrode and the maximum total injected current. For example, an electrode montage and stimulation parameters to be provided can be determined using a target map of a cortical surface specifying desired values for the electric field at each (spacetime) point. Further, determination of an electrode montage and stimulation parameters to be provided can employ a weight map providing the degree of relative importance of each location in the target map, and a set of constraints on the number of electrodes and their currents. In embodiments, the weighted target map of the cortical surface is generated by prioritizing the areas in the target map for optimization purposes. For example, a higher weight is given to those brain areas considered to be more important for the particular application of neurostimulation. In embodiments, the electrode montage comprises optimal currents and optimal number and locations for a plurality of electrodes to globally inhibit excitability of the target regions. In embodiments, the optimal currents and optimal number and locations for the plurality of electrodes is calculated under constraints regarding the total current injected into the brain by all electrodes at any time. In embodiments, the optimal currents and optimal number and locations for the plurality of electrodes is calculated under constraints regarding a maximal current at each electrode. In embodiments, the electrode montage reduces excitability of the plurality of target regions without substantial excitation outside the target regions. For example, the electrode montage may comprise optimal currents and optimal number and locations for a plurality of electrodes to globally inhibit excitability of at least an epileptogenic region and a propagation network.

In embodiments, placement of the electrodes uses the International 10-10 system. The International 10-10 system is a method for standardized placement of electrodes. The 10-10 system correlates external scalp locations with the underlying cortical areas. Electrode sites are identified with a letter to identify the lobe, or area of the brain. Regions of the brain are labeled as pre-frontal (Fp), frontal (F), temporal (T), parietal (P), occipital (O), and central (C). The International 10-10 system also identifies (z) sites; A “z” (zero) refers to an electrode placed on the midline sagittal plane of the skull, (Fpz, Fz, Cz, Oz) and is present mostly for reference/measurement points. The International 10-10 system uses even- numbered electrodes for the right side of the head, whereas odd numbers refer to electrodes placed on the left.

In embodiments, the calculation of stimulation parameters and electrode locations is performed under constraints regarding maximal electrode number, maximal or minimal current at each electrode, and the total current injected into the brain by all electrodes at any time. In embodiments, the calculations are performed under additional constraints including holding the current in an electrode at a constant fixed value.

In embodiments, the calculation of stimulation parameters using the biophysical head model (e.g., current intensity for tDCS) uses least squares. In embodiments, the present method comprises using constrained least squares to optimize current intensities. In embodiments, the calculation of optimal electrode locations and/or optimal electrode numbers employs a genetic or evolutionary algorithm. Exemplary algorithms are described, for example, in U.S. Patent No. 9,694,178, the entire disclosure of which is hereby incorporated by reference. The genetic algorithm can be based on the definition of a solution by a “DNA” binary string (in this case of dimension N-l) specifying the electrode locations and number, and stimulation parameters, and may employ as an optimization function the least squares error, i.e., the one with the best possible current configuration for the chosen electrode locations. Cross-over and mutation functions are defined to ensure that the offspring of solutions do not violate the constraint of maximal number of electrodes in the solution. Once a DNA string is specified (i.e., a particular montage), its fitness can be computed by inverting the solution for that particular montage. Solutions with more than the maximal number of electrodes desired are penalized strongly. The genetic algorithms with specifically designed fitness, cross-over and mutation functions, converge quickly and reliably to a solution.

In some embodiments, the head model is a template head model, which is either a single model from a different subject or a group-model. These embodiments are useful, for example, where an MRI of the subject is not available. For example, the head model can be a single model from a different subject and selected based on one or more criteria that match the subject, such as age, gender, and head measurements. Other criteria, such as ethnicity, may also be used. For example, in embodiments the template head model is from an individual that is similar in age to the subject, for example, ±25%, or ±20%, or ±15%, or ±10% of the age of the subject. In embodiments, the head model is selected at least according to age of the subject and one or more head measurements, such as one or more (or all) measurements selected from nasion-to-inion distance, tragus-to-tragus distance, and head coronal perimeter. These morphometric measurements help to ensure that the current intensity and electric field modeling according to the head model will be similar for the subject. In certain embodiments, these measurements are each within ±20%, ±15%, or ±10% between the selected head model and the subject.

In some embodiments, the head model is a group head model. In embodiments, a group model is a model where data from multiple sources and subjects is combined to represent an average head model. In embodiments, several head models are used to construct a loss function where each subject's head model contributes to the loss calculation. In embodiments, an MNI MRI template is used to create an average physical head model. In embodiments, an average head model can be customized for the pathology or a patient by selecting the appropriate data.

In embodiments, the group head model can be prepared from a group of template head models matched for the subject based on the one or more criteria (as described). In embodiments, the template head models for group optimization are selected at least according to age of the subject and one or more measurements selected from nasion-to-inion distance, tragus-to-tragus distance, and head coronal perimeter (as described). In embodiments, an optimization function determines a montage that minimizes the average of ERNI across all the subjects. In still other embodiments, multiple models may be prepared for the subject and used for group optimization according to this disclosure.

For example, a group-optimized solution can be produced by creating an ensemble of head models (such as biophysical models or hybrid brain models described below) and then using them to construct a cumulative loss function where the loss is computed through a weighted combination of the losses (i.e., match of the solution with desired optimization outcome) for each of the models. The loss function is a function of the losses of each model in the ensemble. Ensembles can be produced from head/hybrid models of different subjects or from models produced differently from the same data. For example, a plurality of models can be produced assuming different conductivity values, connectivity assumptions, or different target realizations, to reflect uncertainties and/or priors in the modeling pipeline. The ensemble for optimization itself may be personalized from available subject data (such as age, gender, ethnicity, clinical data such as SEEG, head measurements, body mass index, etc).

In some embodiments, the method employs a head model that is a hybrid model prepared with MRI and one or more of SEEG, EEG, DW-MRI, DTI, PET and fMRI to further model the subject’s unique neural dynamics (e.g., connectivity). In embodiments, the head model is further informed with interictal SEEG/EEG data and/or ictal SEEG/EEG data.

In embodiments, the model comprises neural mass models (NMM) coupled to create a Brain Network Model (BNM). NMMs represent the average activity of populations of neurons. NMMs are mathematical representations of the dynamics of the average membrane potential and firing rate of a population of neurons. See, for example, Sanchez-Todo, R. et al. 2023. In other embodiments, the BNM is constructed using Wilson and Cowan models, or mean field models. The Wilson-Cowan model describes the evolution of excitatory and inhibitory activity in a synaptically coupled neuronal network. Mean-field models can reproduce local neural dynamics elicited by different cortical inputs and can accurately predict population-specific activity patterns. BNMs are created through an ensemble of neural population models (e.g., NMMs) and are informed by anatomical connectivity, which can be inferred from various neuroimaging techniques, such as diffusion MRI-based tractography. EEG, SEEG, and fMRI data can also be used.

In embodiments, a BNM is created using the following steps. In embodiments, structural data (such as from an Tlw-MRI and/or T2w-MRI) is obtained and used to prepare a biophysical head model that represents the geometry of the patient’s head and the passive electrical properties. The biophysical head model comprises segmentation of the different tissues: scalp, skull, cerebrospinal fluid, grey matter, and white matter. In embodiments, the biophysical head model also comprises lesions that are present. Non-limiting examples of lesions comprise one or more of corticectomy, stroke, brain tumor, skull scar tissue, and titanium implanted plate. Using measured data such as from DW-MRI (diffusion-weighted MRI), connectivity is inferred between different regions of the subject’s brain in the head model. In some embodiments, the connectivity of various regions is inferred from DW-MRI using DTI (diffusion tensor imaging). A BNM is created by placing in various regions (or nodes) of the head model, a mathematical model (e.g., NMM) to represent brain activity and connectivity at and between the various regions. The NMMs and their ensembles can be constructed to recapitulate, for example, measured EEG, SEEG, and/or fMRI activity, and to represent the epileptogenicity of each region. See, Lopez-Sola, et al. 2022 with regard to construction of NMMs. The model parameters can be tuned to model the actual measured data and the subject’s symptoms as closely as possible. In embodiments, the head model is based on MRI to determine biophysical structures, a CT scan/MRI to determine placement of SEEG electrodes, and SEEG data and DW-MRI/DTI to infer connectivity and construct NMMs and their ensembles (i.e., construct the BNM). In embodiments, the resulting BNM can be used to simulate the subject’s epilepsy seizures and/or their propagation. In embodiments, the subject’s connectivity parameters are tuned using SEEG and/or EEG data to simulate as closely as possible the subject’s resting interictal and/or ictal neural activity, which in embodiments can employ evolutionary algorithms or Bayesian optimization or deep learning optimization methods (see Lan, Qingfeng, et al. (2023), the contents of which are hereby incorporated in its entirety). In embodiments, subject data is first projected to a lower dimension latent space using deep learning techniques such as variational autoencoders to simplify the optimization process, avoiding arbitrary feature selection. In embodiments, the brain model comprises neural mass models, neural field models, or mean field models coupled to create a Brain Network Model.

In embodiments, the electrode montage is selected at least in-part to reduce seizure probability, intensity or and seizure spread (or probability of seizure propagation) in the Brain Network Model (BNM). In embodiments, the electrode montage is selected using a genetic or evolutionary algorithm (similar to that described in U.S. Patent No. 9,694,178), selected in part for its ability to inhibit the epileptic focus and/or reduce seizure spread according to the BNM, and may or may not further include selection based on the calculated electric field at target regions and/or the rest of the cortex.

In embodiments, FIG. 1 illustrates an exemplary pipeline for preparing a hybrid head model according to embodiments of the disclosure, where SEEG seizure data and clinical information is used to personalize a head model. Neural mass models (NMM) are created and positioned as nodes in various regions of the brain. Such head models can be prepared with an MRI image to model passive tissues and dMRI data to model connectivity, for example. The NMM nodes are interconnected with the subject’s unique information. The resulting epileptogenic node model can be further personalized by tuning the connectivity or other model parameters according to the subject’s measured SEEG and EEG data. These head models can be used to simulate the subject’s seizures and their propagation, and to select tDCS montages that reduce the seizure (or probability of seizure) or their propagation according to the simulations.

In embodiments, to assess the fitness of a particular montage according to this disclosure, a target map is first created to inhibit the epileptogenic and propagation regions (as described). A weight map is also created with larger weights in the epileptogenic regions and/or propagation regions compared to the rest of the regions in the brain. Electric field distribution is estimated for the montage (e.g., as described using the biophysical model), and the ERNI is calculated using the target and weight maps. In addition, the electric field distribution is used in the personalized brain network model to simulate its effects. From these simulations, metrics such as the seizure probability or the propagation probability outside of the epileptogenic regions can be extracted.

For example, in such embodiments, the method can comprise: preparing a target map comprising targets for reducing excitability of one or more of an epileptogenic region and a propagation network; preparing a weight map indicating the relative importance of the targets for reducing excitability of one or more of an epileptogenic region and a propagation network; estimating an electric field distribution for an electrode montage and calculating ERNI based on the target and weight maps; and simulating the effects of the electric field distribution using a brain network model on seizure probability or the propagation probability outside of the epileptogenic regions.

In embodiments, to find the optimal montage, an evolutionary algorithm is used to minimize the seizure probability and/or the seizure propagation probability while minimizing the ERNI. The evolutionary algorithm can be based on a parameter vector that defines the positions of the active electrodes and their currents (as described). Such parameter vector can be designed to ensure that the solutions being assessed do not violate constraints regarding the total injected current, maximum number of electrodes, and current conservation constraints. In addition, repair functions can be designed to ensure that the assayed montages do not violate the maximum current per electrode constraint. This evolutionary algorithm, with designed mutation and cross-over functions, can reliably converge to a solution.

The method described herein is useful for treating various types of epilepsy. In embodiments, the epilepsy is focal epilepsy, generalized epilepsy, combination focal and generalized epilepsy, and unknown epilepsy. In embodiments, the epilepsy is focal epilepsy. In embodiments, focal epilepsy is idiopathic localization-related epilepsy, frontal lobe epilepsy, temporal lobe epilepsy, parietal lobe epilepsy, or occipital lobe epilepsy.

In embodiments, the epilepsy is drug-resistant epilepsy. In various embodiments the subject is resistant (e g., or has become non-responsive) to one or more drugs selected from Acetazolamide, Brivaracetam, Cannibadiol, Carbamazepine, Cenobamate, Clobazam, Clonazepam, Clorazepate, Corticotropin, Diazepam, Eslicarbazepine acetate, Ethotoin, Ethosuximide, Felbamate, Fenfluramine, Fosphenytoin, Gabapentin, Lacosamide, Lamotrigine, Levetiracetam, Lorazepam, Methsuximide, Midazolam, Oxcarbazepine, Perampanel, Phenobarbital, Phenytoin, Pregabalin, Primidone, Rufinamide, Stiripentol, Tiagabine, Topiramate, Valproate, Vigabatrin, and Zonisamide. In some embodiments, the subject has a moderate response to pharmaceutical intervention, but the drug does not completely control seizures, or the drug loses efficacy over time. In such embodiments, the therapy described herein can be used alongside pharmaceutical intervention.

In embodiments, the montage designed according to this disclosure comprises a cathode placed over or proximal to an epileptogenic region. In embodiments, the epileptogenic region is the region of the brain from which the patient’s seizures arise. In embodiments, the electrode montage targets the periphery of the epileptogenic region. In embodiments, the epileptogenic region is located in the mesial temporal lobe. In embodiments, the electrode montage disrupts the connectivity of the epileptogenic region.

In embodiments, the electrode montage comprises one or more cathodes and one or more anodes positioned to reduce excitability of the propagation network without substantial excitation outside of the target regions.

In embodiments, the electrode montage has at least 3 cathodes. In embodiments, the electrode montage has 4, 5, or 6 cathodes. In embodiments, the electrode montage has at least 6 electrodes or at least 7 electrodes. In embodiments, the electrode montage has no more than 8 electrodes, and optionally 4 electrodes. In embodiments, the electrode montage has from 2 to about 6 electrodes. In some embodiments, the electrode montage has a plurality of anodes (e.g., from 2 to 6, or from 2 to 4, anodes).

In embodiments, the maximal voltage of the electric field of the montage is about 23 volts per meter. In embodiments, the maximal voltage of the electric field is about 20 volts per meter, or about 21 volts per meter, or about 22 volts per meter, or about 23 volts per meter, or about 24 volts per meter, or no more than about 25 volts per meter.

In various embodiments, the current intensity of any cathode is at least about 0.01 milliamps (mA). In various embodiments, the current intensity of any cathode is at least about 0.10 milliamps (mA). In various embodiments, the current intensity of any cathode is at least about 1.00 milliamps (mA). In some embodiments, the current intensity of any cathode is at least about 1.25 mA or at least about 1.50 mA, or at least about 1.75 mA. In embodiments, the maximum current intensity of any cathode is about 2.00 mA.

In various embodiments, the method of this disclosure uses a total injected current intensity of at least about 2.00 mA. In some embodiments, the total injected current intensity is at least about 3.00 mA, or at least about 4.00 mA. In some embodiments, the total injected current intensity is no more than about 5.00 mA or no more than about 4.00 mA. In some embodiments, the total injected current intensity is half the sum of the absolute values of all the currents.

Generally, the tDCS is applied in multiple sessions. In some embodiments, the tDCS is applied for at least 5 sessions, at least 10 sessions, at least 15 sessions, or at least 20 sessions. Sessions can be administered continually. Sessions can be performed at a set frequency, or variable frequency. In embodiments, the frequency of the sessions is selected from (or varies within) the range of bimonthly (e.g., every other month) to three times daily. In some embodiments, sessions are performed at least once per week, such as once, twice, or three times per week (i.e., on average). In some embodiments, at least two sessions are performed on consecutive days. In some embodiments, sessions are performed on at least 2, 3, 4, or 5 consecutive days. In these or other embodiments, sessions can be performed multiple times (such as 2 or 3 times) on the same day. In some embodiments, sessions are performed twice daily or three times daily on average. In some embodiments, sessions are performed no more than three times on any day. In exemplary embodiments, sessions are performed about daily or about weekly. In some embodiments, sessions are performed about bimonthly, monthly, semimonthly, biweekly, weekly, or semiweekly, daily, or more than daily, and any number of periodic sessions there between.

In various embodiments, the tDCS is applied for a duration of at least about 5 minutes (i.e., per session), or at least about 10 minutes, or at least about 15 minutes, or at least about 20 minutes, or at least about 30 minutes, or at least about 45 minutes. In various embodiments, the tDCS is applied for a duration of no more than about 2 hours (i.e., per session), or for no more than about 1 hour. In some embodiments, the tDCS is applied for a duration of from about 10 minutes to about 1 hour. In some embodiments, the tDCS is applied for about 20 minutes to about 40 minutes. In embodiments, the tDCS is applied for a duration of about 15 minutes, about 20 minutes, about 30 minutes, or about 45 minutes.

Sessions can be performed for a definite period (such as one month, two months, six months, or one year), or can be performed indefinitely, or as needed to combat symptoms of epilepsy.

In embodiments, the present disclosure provides a tDCS system for treating epilepsy in a subject, comprising, an electrode montage that modifies the excitability of a plurality of target regions of the subject’s cortex, wherein the target regions comprise an epileptogenic region, a propagation network, and/or targets within the rest of the cortex. The system further comprises a head model and optionally an evolutionary algorithm as described herein, and can be applied in a method for treating epilepsy.

The tDCS methods and systems described herein can employ systems that are described in US 9,694,178, US 10,463,855, US 2020/029119, and US 2021/0031034, which are hereby incorporated by reference in their entireties.

As used herein, the term “about” means ±10% of a reference value, unless the context clearly requires otherwise.

EXAMPLES

Example 1: Biophysical Head Model

A physics pipeline leverages personalized biophysical head models, built from structural neuroimaging data of a patient’s head, to determine the ideal locations for the electrodes and their currents in order to decrease the cortical excitability in an epileptogenic focus region identified by the physician. Biophysical head models are 3D volume conductor models that accurately represent the geometry of the patient’s head and their passive electrical properties. These models are built from automatic segmentation of Tlw-MRIs, or Tlw and T2w-MRIs, into different tissues: scalp, skull, cerebrospinal fluid (CSF, including the ventricles), grey matter (GM), and white matter (WM). Certain types of lesions can also be represented in the head model, such as corticectomies, strokes, certain types of brain tumors, skull scar tissue and titanium implanted plates (provided details about the geometry of the plate is known). The inclusion of lesions is, in most cases, not automatic and requires a manual segmentation performed by a trained operator.

Geometric representations of the electrodes in the positions present in a tES headcap are then modeled on the scalp of the patient. The finite element numerical technique is then used to calculate the lead-field matrix, i.e, a matrix that allows for the calculation of the electric field distribution in any structure of the head model induced by any electrode montage. A montage includes the currents assigned to each electrode. This calculation requires that the electrical conductivity of the healthy/lesioned tissues and electrodes be set, and this is done by using reference values, which can be obtained from a literature review of different modeling studies that employ these types of head models.

The montage optimization step that follows requires the identification of the epileptogenic focus region by the physician. This target region can be identified directly in the brain of the personalized head model, or by mapping it from a template head located in a reference coordinate space. In most applications, the target region to inhibit is identified in the GM surface (the interface between the CSF and the GM, also known as the cortical surface), based on the assumption that the neuromodulatory effects of tDCS are predicted by the direction and intensity of the component of the E-field normal (En) to this surface. Under this assumption (lambda-E model), an En directed out of the GM surface, results in a hyperpolarization of the soma of pyramidal cells in the cortex, thus resulting in a decreased cortical excitability.

The montage optimization method, such as the Stimweaver algorithm, then determines the ideal montage by employing numerical methods to minimize the weighted difference (ERNI, error with respect to no intervention) between the En distribution (in the GM surface) induced by the montage and a target En distribution. The latter assigns a strong inhibitory En field in the region identified by the physician. A map of weights is also used to determine the relative importance of the different regions in the target map. Stimweaver restricts the montage to a maximum number of electrodes (which depends on the ones available to the stimulator, depending on the model of the stimulator that is available) using a genetic algorithm. Furthermore, the total injected current and maximum current per electrode (in absolute value) are also restricted to the values deemed safe based on available evidence using tDCS. The Stimweaver algorithm needs to evaluate several candidate montages until it converges, thus taking advantage of the lead-field matrix formulation to calculate them in a computationally efficient manner. This algorithm can also be configured to run in a manner that generates a negligible En-field in the target, but still using at least one electrode in the scalp with a current that is sufficient to produce scalp sensations. This montage is deemed the actisham montage, and it is a model driven approach to perform a sham stimulation, that is superior to traditional sham methods in blinding the patient to the type of stimulation being performed.

This pipeline can also be generalized to other neurological diseases. In more general cases, the target of stimulation is marked for excitation or, in cases where networks of regions are involved, a combination of inhibition and excitation.

In cases where obtaining a patient-specific MRI is not viable, template head models can be employed as described in Example 3. An alternative approach is to use group- optimized montages, which employ a loss function for optimization resulting from the combination of several biophysical head models, with anatomical properties falling within the range of the ones of the subject being processed. For group-optimized approaches, Stimweaver is employed to minimize the average of ERNI across all the subjects or models. See Example 3.

Example 2: Physical and Physiological Hybrid Model

A hybrid model combines physical and physiological models for montage optimization. The physical modeling framework is equivalent to that explained in Example 1. That is, biophysical head models are produced from MRI images, and simulations are run to estimate the electric field distribution in the brain generated by a montage. If an MRI is not available, it is possible to use template head models as described.

The physiological models used in this Example are patient-specific (i.e. personalized) brain network models. These models consist of nodes or elements that represent the different brain regions and edges or connections representing anatomical or functional links between them. The strength of these connections is usually referred to as the connectome and is generally derived from empirical data. There are various means to calculate the connectome: (1) from Diffusion weighted MRI data and tractography (technique to track the fibers that connect the different regions); (2) using a mathematical analysis of the temporal correlation between regions in data obtained by resting state fMRI, task based fMRI, EEG, MEG and SEEG; using Ising models in combination with one of the following resting state fMRI, task based fMRI, EEG, MEG and SEEG (see Ruffini, G. et al. (2023), the contents of which are hereby incorporated in its entirety); and from template or group connectivity values (databases).

The local dynamics in each of the brain regions can be modeled by using either neural mass models, neural field models, or mean-field models. These models have a series of parameters that can be personalized for each patient. The personalization is done through assimilation of data from different modalities (EEG, SEEG, fMRI, etc.). To do so, a loss function that evaluates the fitness of the model with the data is selected and the model parameters are adjusted to minimize the value of the loss function. The loss function evaluation may require the generation of synthetic data using the brain network model. This data assimilation process may also include information provided by the clinicians, for example the identification of the epileptogenic and the propagation zones.

Once a personalized brain network model is in place, it can be used to generate an optimized stimulation montage. In some aspects the process is equivalent to that in the physics pipeline. That is, an optimization algorithm tries different montages to find the one that minimizes a loss function (a mathematical map from montage configuration space to the real numbers). The optimization process can also have constraints on the number of electrodes to use, the available electrode locations, the maximum current values per electrode, the maximum total injected current or the minimum current per electrode.

During the montage optimization process, to evaluate a montage, the electric field distribution needs to be estimated, as in the physics model described in Example 1. The key difference with the physics model is that the electric field distribution is then used in the personalized brain network model to simulate its effects. From these types of simulations, a value of the loss function is calculated to assess the fitness of a montage. The loss function calculation may combine results obtained from the brain network model (e.g. seizure probability or seizure propagation probability) with results obtained from the biophysical head model (e.g. average field in brain areas of importance or fitness with a weight map and a target map).

Example 3: Group Models and Group Optimizations

FIGs. 2A-2C show a non-limiting example of group head models and group optimization of a montage. A group model is a model where data from multiple sources and subjects is combined to represent the average patient (which may be a biophysical model or a hybrid model. As an example, an MNI MRI template can be used to create an average physical head model. The creation of an average head model can be customized for a pathology or even a patient by selecting the appropriate data. One can use such a group model to create a “template-based solution”.

Another approach is group optimization. A group-optimized solution is produced by first creating an ensemble of head models (such as hybrid brain models) and then using them to construct a cumulative loss function where the loss is computed through a weighted combination of the losses (i.e., match of the solution with desired optimization outcome) for each of the models. More generally, the loss function is a function of the losses of each model in the ensemble. Ensembles can be produced from head/hybrid models of different subjects or from models produced differently from the same data (e.g., assuming different conductivity values or different target realizations) to reflect uncertainties and/or priors in the modeling pipeline. The ensemble for optimization itself may be personalized from available subject data (such as age, gender, ethnicity, clinical data, head measurements, body mass index, etc).

EQUIVALENTS

While the invention has been described in connection with specific embodiments thereof, it will be understood that it is capable of further modifications and this application is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains and as may be applied to the essential features hereinbefore set forth and as follows in the scope of the appended claims. Those skilled in the art will recognize, or be able to ascertain, using no more than routine experimentation, numerous equivalents to the specific embodiments described specifically herein. Such equivalents are intended to be encompassed in the scope of the following claims. INCORPORATION BY REFERENCE

All patents and publications referenced herein are hereby incorporated by reference in their entireties. Exemplary publications are listed below in the “REFERENCES” section and throughout the above disclosure.

The publications discussed herein are provided solely for their disclosure prior to the filing date of the present application. Nothing herein is to be construed as an admission that the present invention is not entitled to antedate such publication by virtue of prior invention.

REFERENCES

E Ruffini G., et al., Optimization of multifocal transcranial current stimulation for weighted cortical

from realistic of electric fields. NeuroImage

(2014).

2. Miranda PC., et al., The electric field in the cortex during transcranial current stimulation NeuroImage Vol 30 (2013).

3. Ruffini, G., et al., Targeting brain networks with multichannel transcranial current stimulation

Curr. Opin. Biomed. Eng. 8, 70-77. (2018).

4. Lopez-Sola E., et al., A personalizable autonomous neural mass model of epileptic seizures 10.1101/2021.12.24.474090 (2021).

5. Nitsche, M A, and W Paulus., Excitability changes induced in the human motor cortex by weak transcranial direct current stimulation The Journal of physiology vol 527

Pt 3, Pt 3 (2000): 633-9. doi: 10.1111/j. l469-7793.2000.t01-l-00633.x

6. Lan, Q., et al., Learning to Optimize for Reinforcement Learning. arXiv preprint arXiv:2302.01470 (2023),

7. Ruffini, G. et al., LSD-induced increase of Ising temperature and algorithmic of brain dynamics PLoS computational biology vol 19,2 el010811. 3 Feb.

2023, doi: 10.1371/journal.pcbi.1010811

Claims

CLAIMS What is claimed is:

1. A method for treating epilepsy in a subject, comprising: providing a transcranial electrical stimulation (tES) electrode montage designed to modify excitability of a plurality of target regions of the subject’s cortex, wherein the target regions comprise targets for reducing excitability of one or more of an epileptogenic region and a propagation network, wherein the electrode montage is selected for the subject using a head model; and applying the tES with the electrode montage.

2. A method for designing an electrode montage for an epilepsy subject, the electrode montage designed to modify excitability of a plurality of target regions of the subject’s brain, wherein the target regions comprise an epileptogenic region and/or a propagation network, the method comprising providing a head model and determining an electrode montage that reduces probability of seizure or probability of seizure propagation according to simulations using the model.

3. The method of claim 1 or 2, wherein the target region comprises cortical targets and/or subcortical targets.

4. The method of any one of claims 1 to 3, wherein the target regions further comprise one or more targets outside the epileptogenic region and/or propagation network.

5. The method of any one of claims 1 to 4, wherein the electrode montage modifies excitability of the plurality of target regions without substantial excitation outside the target regions.

6. The method of any one of claims 1 to 5, wherein the target regions comprise electric field specification for stimulation.

7. The method of any one of claims 1 to 6, wherein the target regions comprise physiological targets comprising an epileptogenic region and/or propagation network.

8. The method of any one of claims 1 to 7, wherein the tES comprises transcranial direct current stimulation (tDCS).

9. The method of any one of claims 1 to 8, wherein the head model is created from one or a plurality of images or scans selected from MRI, CT, DW-MRI, fMRI, fNIRS, PET, rs- fcMRI, EEG, MEG, SEEG, and SPECT.

10. The method of any one of claims 1 to 9, wherein the target regions are identified by one or more of SEEG, EEG, MEG, PET and MRI.

11. The method of any one of claims 1 to 10, wherein the head model comprises: an electric field characteristic target map for the subject’s cortex, said target map including the target regions and desired values for the electrical field at each of the target regions to modulate excitation of said target regions; and a weight map for the cortical surface specifying a degree of relative importance for each of the target regions and the rest of the cortex for the purposes of modulating excitability.

12. The method of claim 11, wherein the electrode montage comprises optimal currents and optimal number and locations for a plurality of electrodes to globally inhibit excitability of at least an epileptogenic region and a propagation network.

13. The method of claim 12, wherein the optimal currents and optimal number and locations for the plurality of electrodes is calculated under constraints regarding the total current injected into the brain by all electrodes at any time.

14. The method of claim 12 or 13, wherein the optimal currents and optimal number and locations for the plurality of electrodes is calculated under constraints regarding a maximal current at each electrode.

15. The method of any one of claims 11 to 14, wherein the head model is prepared from an image or scan of the subject’s brain, and optionally where the image or scan is an MRI, and optionally comprises a Tlw-MRI and/or a T2w-MRI.

16. The method of claim 15, wherein the MRI image or scan comprises a Tlw-MRI with full head coverage.

17. The method of claim 15 or 16, wherein the MRI image or scan comprises a T2w- MRI with full head coverage.

18. The method of any one of claims 1 to 17, wherein the head model comprises one or more of a cortical resection geometry, skull defect(s), and/or implants of the subject’s head.

19. The method of any one of claims 1 to 14, wherein the head model is a template head model, which is either a single model from a different subject or a group head model.

20. The method of claim 19, wherein the head model is a single model from a different subject, and selected based on one or more criteria selected from: age, gender, and head measurements.

21. The method of claim 20, wherein the head model is selected at least according to age of the subject and one or more head measurements selected from nasion-to-inion distance, tragus-to-tragus distance, and head coronal perimeter.

22. The method of claim 19, wherein the template head model is a group head model.

23. The method of claim 22, wherein the group head model is prepared from a group of template head models matched for the subject based on one or more criteria selected from: age, gender, and head measurements.

24. The method of claim 23, wherein the template head models are selected at least according to age of the subject and one or more measurements selected from nasion-to-inion distance, tragus-to-tragus distance, and head coronal perimeter.

25. The method of claim 19, wherein the group head model is created from a plurality of models created from the subject’s neuroimaging data.

26. The method of any one of claims 1 to 25, wherein the head model is a hybrid model prepared with MRI and one or more of SEEG, EEG, DW-MRI, DTI, and fMRI.

27. The method of claim 26, wherein the head model is prepared with interictal EEG data and ictal SEEG data.

28. The method of claim 26 or 27, wherein the brain model comprises neural mass models, neural field models, or mean field models coupled to create a Brain Network Model.

29. The method of claim 28, wherein epilepsy seizures and/or their propagation, or probabilities thereof, are simulated using the Brain Network Model.

30. The method of any one of claims 26 to 29, wherein the hybrid model is prepared based on neuroimaging and empirical data measured for the subject.

31. The method of claim 30, wherein a plurality of hybrid models are prepared for the subject for group optimization.

32. The method of any one of claims 26 to 29, wherein hybrid models are prepared from neuroimaging and empirical data from different subjects for group optimization.

33. The method of any one of claims 28 to 32, wherein the subject’s connectivity parameters are tuned using one or more of SEEG, EEG, DW-MRI, DTI, and fMRI data, optionally wherein SEEG and/or EEG data is used to reproduce the subject’s resting interictal EEG and optionally ictal EEG.

34. The method of any one of claims 26 to 33, wherein the electrode montage is selected to reduce probability of seizure or seizure spread in the Brain Network Model.

35. The method of claim 34, comprising: preparing a target map comprising targets for reducing excitability of one or more of an epileptogenic region and a propagation network; preparing a weight map indicating the relative importance of the targets for reducing excitability of one or more of an epileptogenic region and a propagation network; estimating an electric field distribution for an electrode montage, and calculating ERNI based on the target and weight maps; simulating the effects of the electric field distribution using a brain network model on seizure probability or the propagation probability outside of the epileptogenic regions.

36. The method of claim 35, comprising using an evolutionary algorithm to minimize the seizure probability and/or the seizure propagation probability while minimizing the ERNI.

37. The method of claim 36, wherein the evolutionary algorithm is based on a parameter vector that defines the positions of active electrodes and their currents.

38. The method of any one of claims 1 to 37, wherein the epilepsy is focal epilepsy, generalized epilepsy, combination focal and generalized epilepsy, and unknown epilepsy.

39. The method of claim 38, wherein epilepsy is focal epilepsy.

40. The method of claim 39, wherein focal epilepsy is idiopathic localization-related epilepsy, frontal lobe epilepsy, temporal lobe epilepsy, parietal lobe epilepsy, and occipital lobe epilepsy.

41. The method of any one of claims 38 to 40, wherein the epilepsy is drug resistant epilepsy.

42. The method of any one of claims 1 to 41, wherein a cathode is placed over or proximal to an epileptogenic region.

43. The method of claim 42, wherein the electrode montage targets the periphery of the epileptogenic region.

44. The method of claim 42 or 43, wherein the epileptogenic region is located in the mesial temporal lobe.

45. The method of any one of claims 1 to 44, wherein the electrode montage disrupts the connectivity of the epileptogenic region.

46. The method of any one of claims 1 to 45, wherein one or more cathodes and one or more anodes are positioned to reduce excitability of the propagation network without substantial excitation outside of the target regions.

47. The method of claim 46, wherein the electrode montage has at least 3 cathodes.

48. The method of claim 47, wherein the electrode montage has 4, 5, or 6 cathodes.

49. The method of claim 47, wherein the electrode montage has at least 6 electrodes.

50. The method of claim 49, wherein the electrode montage has at least 7 electrodes.

51. The method of claim 46, wherein the electrode montage has no more than 8 electrodes, and optionally 4 electrodes.

52. The method of claim 46, wherein the electrode montage has from 2 to about 6 anodes

53. The method of any one of claims 1 to 52, wherein the maximal voltage of the electric field of the montage is about 23 volts per meter.

54. The method of any one of claims 1 to 53, wherein the current intensity of any cathode is at least about 0.01 milliamps (mA), or at least about 0.10 mA, or at least about 1.00 mA.

55. The method of claim 54, wherein the current intensity of any cathode is at least about 1.25 mA.

56. The method of claim 55, wherein the current intensity of any cathode is at least about 1.50 mA.

57. The method of claim 55 or 56, wherein the maximum current intensity of any cathode is about 2.00 mA.

58. The method of any one of claims 53 to 57, wherein the total injected current intensity is at least about 2.00 mA.

59. The method of claim 58, wherein the total injected current intensity is at least about 3.00 mA.

60. The method of claim 58 or 59, wherein the total injected current intensity is no more than about 5.00 mA.

61. The method of claim 60, wherein the total injected current intensity is no more than about 4.00 mA.

62. The method of any one of claims 1 to 61, wherein the tDCS stimulation comprises at least one session.

63. The method of claim 62, wherein the tDCS stimulation comprises more than one session, with the frequency of the sessions being from bimonthly to three times daily.

64. The method of claim 63, wherein sessions occur at a frequency of bimonthly, monthly, semimonthly, biweekly, weekly, semiweekly, daily, or twice daily, or a combination thereof.

65. The method of any one of claims 62 to 64, wherein the tDCS stimulation is applied for a duration of at least 5 minutes.

66. The method of claim 65, wherein the tDCS stimulation is applied for a duration of at least 10 minutes, or at least 15 minutes, or at least 20 minutes, or at least 25 minutes, or at least 30 minutes.

67. The method of claim 66, wherein the tDCS is applied for a duration of no more than 2 hours.

68. The method of claim 67, wherein the tDCS is applied for a duration of no more than 1 hour.

69. The method of any one of claims 66 to 68, wherein the tDCS is applied for a duration of from about 10 minutes to about 1 hour.

70. The method of claim 69, wherein the tDCS is applied for a duration of from about 20 minutes to about 40 minutes.

71. The method of claim 70, wherein the tDCS is applied for a duration of about 15 minutes, about 20 minutes, about 30 minutes, or about 45 minutes.

72. A tDCS system for use in treating epilepsy according to a method of any one of claims 1 to 71.