WO2024097867A2 - Systems and methods for optimizing electroencephalography recording sensitivity - Google Patents

Abstract

The present disclosure provides systems and methods for optimizing patient-specific electroencephalography recording sensitivity. In particular, the present disclosure provides novel systems and methods for optimizing the recording sensitivity of electrode configurations to minimize the number of implanted electrodes while maintaining effective recording coverage and localization.

Description

SYSTEMS AND METHODS FOR OPTIMIZING

ELECTROENCEPHALOGRAPHY RECORDING SENSITIVITY

GOVERNMENT FUNDING

[0001] This invention was made with Government support under Federal Grant No. F31NS124094 awarded by the National Institute of Neurological Disorders & Stroke (NIH/NINDS) and Federal Grant No. TR002553 awarded by the National Institutes of Health (Clinical and Translational Science Award). The Federal Government has certain rights to the invention.

RELATED APPLICATIONS

[0002] This application claims priority to and the benefit of U.S. Provisional Patent Application No. 63/422,091 filed November 3, 2022, which is incorporated herein by reference in its entirety for all purposes.

FIELD

[0003] The present disclosure provides systems and methods for optimizing patient-specific electroencephalography recording sensitivity. In particular, the present disclosure provides novel systems and methods for optimizing the recording sensitivity of electrode configurations to minimize the number of electrodes while maintaining effective recording coverage and localization.

BACKGROUND

[0004] More than 30% of the 65 million people with epilepsy worldwide do not benefit from pharmaceuticals. Surgical resection/ablation is the primary curative treatment for pharmacoresistant epilepsy but requires robust localization of the epileptogenic zone (EZ)-the minimum amount of neural tissue that needs to be removed to achieve seizure freedom. Noninvasive EZ localization is often insufficient and invasive monitoring is often required. Stereo-EEG (sEEG) is an EZ localization approach where up to 30 electrodes are implanted to record electrical activity from widespread areas of the brain, and sEEG has largely displaced other forms of invasive monitoring because of reduced complication rates. Implanting as few sEEG electrodes as necessary is critical because each additional electrode increases the risk of hemorrhage, while too few electrodes can lead to poor localization. Thus, there is a need for an approach that optimizes recording sensitivity in a manner that does not affect recording efficacy.

SUMMARY

[0005] Embodiments of the present disclosure include a method for visualizing electrode recording sensitivity. In accordance with these embodiments, the method includes obtaining simulated voltage recordings from a patient’s brain model using a plurality of electrodes present in the model; calculating a patient-specific map of recordable tissue based on the simulated voltage recordings from the plurality of electrodes; transforming the patient-specific map of recordable tissue onto a model of a human brain; and calculating a yield corresponding to at least one measurement of recording sensitivity from the model.

[0006] In some embodiments, the simulated voltage recordings are based on threshold values.

[0007] In some embodiments, the yield corresponding to the at least one measurement of recording sensitivity comprises at least one of a mean, a median, a sum, and/or a maximum measurement of recording sensitivity. In some embodiments, the at least one of a mean, a median, a sum, and/or a maximum measurement of recording sensitivity is obtained from a plurality of electrodes intersecting a defined point in the model.

[0008] In some embodiments, the model of the human brain is a standard human brain or scalp, or a patient-specific human brain or scalp.

[0009] In some embodiments, the method further comprises removing simulated voltage recordings from one or more confounding electrodes. In some embodiments, the one or more confounding electrodes comprise electrodes that intersect sulci, brain midline, ventricles, vasculature, and/or the other electrodes.

[0010] In some embodiments, the plurality of electrodes are compatible with electroencephalography (EEG), stereo-electroencephalography (sEEG), and/or magnetoencephalography (MEG).

[0011] In some embodiments, the method further comprises combining a plurality of patient-specific maps to form a generalized map of electrode recording sensitivity.

[0012] In some embodiments, the generalized map of electrode recording sensitivity is transformed onto a standard human brain or scalp, or a patient-specific human brain or scalp. [0013] In some embodiments, the patient has or is suspected of having a seizure disorder.

[0014] Embodiments, of the present disclosure also include a method for determining electrode implantation locations and trajectories. In accordance with these embodiments, the method includes obtaining simulated voltage recordings from a patient’s brain model using a plurality of electrodes present in the model; calculating a patient-specific map of recordable tissue based on the simulated voltage recordings from the plurality of electrodes; obtaining voltage recordings from at least one electrode positioned to record from a user-selected anatomical region-of-interest; and optimizing number, location, and/or implantation trajectory of each of the remainder of the plurality of electrodes to maximize a yield corresponding to at least one measurement of recording sensitivity in the region-of-interest.

[0015] In some embodiments, the simulated voltage recordings are based on threshold values.

[0016] In some embodiments, the method produces a patient-specific optimized electrode configuration.

[0017] In some embodiments, the yield corresponding to the at least one measurement of recording sensitivity comprises at least one of a mean, a median, a sum, a weighted sum, and/or a maximum measurement of recording sensitivity.

[0018] In some embodiments, the method produces an optimized electrode configuration that maximizes at least one of a mean, a median, a sum, a weighted sum, and/or a maximum measurement of recording sensitivity.

[0019] In some embodiments, the plurality of electrodes are compatible electroencephalography (EEG), stereo-electroencephalography (sEEG), and/or magnetoencephalography (MEG).

[0020] In some embodiments, the patient has or is suspected of having a seizure disorder.

[0021] In some embodiments, the number, location, and/or implantation trajectory of at least one of the remainder of the plurality of electrodes are determined manually by the user.

[0022] In some embodiments, a patient-specific map of recording sensitivity is updated in real-time as the user manipulates the number, location, and/or implantation trajectory of at least one of the plurality of electrodes.

[0023] In some embodiments, the location of the at least one electrode positioned to record from a user-selected anatomical region-of-interest is determined manually by the user, followed by optimizing the number, location, and/or implantation trajectory of each of the remainder of the plurality of electrodes.

[0024] In some embodiments, the number, location, and/or implantation trajectory of at least one of the remainder of the plurality of electrodes are determined by performing the method of claim 12 multiple times on one or more regions-of-interest. [0025] In some embodiments, the user assigns a significance value to multiple regions-of- interest.

BRIEF DESCRIPTION OF THE DRAWINGS

[0026] FIG. 1 : Pipeline for optimization of sEEG electrode implantation. T1 MRI, post-op CT, and DW-MRI imaging were used to generate 12 patient-specific finite element method (FEM) head models. A set of valid electrode trajectories, computed recording sensitivity, and optimized configurations were gathered for specific ROIs.

[0027] FIGS. 2A-2D: Recording sensitivity of sEEG electrode contacts. (A) Recording sensitivity as a function of distance between the center of the simulated epileptiform activity and the electrode contact for 10 representative contacts using a 500 pV signal detection threshold. (B) Area of recording sensitivity (colored cortex) for 10 contacts (pink dots) corresponding to each trace seen in A using a 500 pV signal detection threshold. (C) Median and interquartile range of recording sensitivity as a function of source to contact distance for 105 contacts in each of 12 patients (n = 1,260). (D) Median recording sensitivity as a function of source to contact distance for a range of source modeling parameters and signal detection thresholds for 105 contacts in each of 12 patients (n = 1,260). The source modeling parameters are 10 cm² and 0.465 nAm/mm² for panels A-C. The 20% recording sensitivity threshold is highlighted as a horizontal black line for (C)-(D).

[0028] FIGS. 3A-3G: Recording sensitivity (RS) of optimized and clinically implanted electrode configurations. (A)-(B) Recording strength (number of electrodes that can record each source) across patient 25 left temporal lobe (LTL) with RS (equation (1)) for clinically implanted (A) and optimized (B) temporal lobe configurations with 10 electrodes at 500 pV threshold. Colored cortex indicates recording strength in the ROI, grey area is outside the ROI, and contact points are shown as spheres. (C) Recording sensitivity as a function of number of implanted electrodes for patient 25 LTL with the optimized (solid) and implanted (dashed) configurations at multiple voltage thresholds. Stars correspond to RSsoo.u v of the configurations in (A) and (B). (D-E) Recording sensitivity at three thresholds for optimized (black) and clinically implanted (grey) electrode configurations using the same number of electrodes. All differences were significant (P < 0.02; two-tailed paired t test for Gaussian samples and Wilcoxon signed rank tests for non-Gaussian samples). (D) Temporal-lobe (TL) specific configurations (n = 8). (E) Configurations for full clinician-defined ROIs (n = 12). (F)-(G) Number of electrodes for clinically implanted (grey) and optimized (black) configurations with equivalent or improved recording sensitivity at three voltage thresholds. All differences were significant (P < 0.02; two-tailed paired t test for Gaussian samples and Wilcoxon signed rank tests for non-Gaussian samples). (F) Temporal-lobe specific configurations (n = 8). (G) Configurations for full clinician-defined ROIs (n = 12). The box plots in all figures indicate the median, interquartile range, maximum, and minimum values, with outside points indicating outliers. The individual tests used, P values, and test statistics are available in Tables 3 A and 3B.

[0029] FIGS. 4A-4D: Performance benefit of patient-specific optimization. (A)-(B) Average recording sensitivity across number of electrodes for optimized LTL (A) and LH (B) configurations (n = 12). Shading shows standard deviation. (C)-(D) Reduction in recording sensitivity for 11 transferred configurations compared to the patient-specific optimized configuration in each patient at a 500 p V threshold. The number of electrodes included in each configuration is the minimum number that yielded >75% recording sensitivity in the optimized configuration. Configurations were optimized for LTL ROIs (C) and LH ROIs (D). The box plots describe median, maximum, minimum, and interquartile range, with outside points indicating outliers.

[0030] FIGS. 5A-5C: Estimation of discernible recording threshold. (A) Example signals of sEEG background noise and an interictal spike. (B) Distribution of noise amplitude. (C) Distribution of interictal spike amplitude (log scale).

[0031] FIGS. 6A-6B: Optimized configurations have increased recording sensitivity. Increase in recording sensitivity between optimized configurations and clinically implanted configurations across all dipole moment densities and patch areas at three signal detection thresholds. (A) Configurations for temporal lobe ROIs (n = 8). (B) Configurations for full clinician-defined ROIs (n = 12). Both figures show standard box plots with median, interquartile range, maximum and minimum.

[0032] FIGS. 7A-7D: Optimized and clinically implanted configurations for 12 patients and three regions of interest (ROIs). Colored spheres represent electrode contacts, colored cortex indicates recording strength in the ROI, and grey area is outside the ROE Configurations were not able to record from the purple cortex. (A) Optimized configurations for LTL with six electrodes. (B) Optimized configurations for LH with 12 electrodes. C-D Optimized (C) and implanted (D) configurations for the clinician-defined region of interest with the number of electrodes noted.

[0033] FIGS. 8A-8H: Sensitivity of optimized configuration recording sensitivity to source model parameters and threshold priority order. A-B Recording sensitivity (RS) of LTL configurations (A) and LH configurations (B) averaged across 12 patients as a function of number of electrodes, using the mean source type (10 cm² and 0.465 nAm/mm²) for RS calculation and all nine source types as optimization parameters. The top series (0.465 nAm/mm², 10 cm², solid blue line) corresponds to the matched case of both optimization and RS calculation with the mean source type. C-D Percent error in RS due to uncertainty of source modeling parameters for all nine source types, for LTL configurations (C) and LH configurations (D), averaged across all 12 patients. Rows represent source parameters of analysis, and columns represent source parameters of optimization. All data in A-D correspond to a 500 pV discernible voltage threshold and a 500 pV-priority cost function. E-F Recording sensitivity (RS) for LTL (E) and LH (F) configurations averaged across all 12 patients as a function of number of electrodes, for all six possible threshold-priority orders, analyzed at 500 pV. G-H Average percent error in RS for 12 patients in LTL (G) and LH (H) configurations for six threshold-priority orders in configuration optimization (rows) and three thresholds for analysis (columns).

[0034] FIGS. 9A-9C: Area and mean radius of sources projected onto the inside of the skull. (A) 10.00 cm² patch (red) on the cortex surface (dark gray) with projected patch (black) on the inner skull surface (light gray) for patient 25. The projected patch has an area of 4.94 cm² and a mean radius of 1.30 cm. (B) Histograms of projected patch area for 21,163 patches across three cortex patch sizes. (C) Histograms of mean radius of projected patch for 21,163 patches across three cortex patch sizes. The average projected patch areas are 2.46, 3.85, and 7.09 cm² and the average mean radii are 1.14, 1.39, and l.83 cm for 6, 10, and 20 cm² patches, respectively.

[0035] FIGS. 10A-10B: (A) An example illustration of the expected recordable area across scalp insertion location. (B) Example average recordable area based on lateral distance from the midline (top), distance from the back of the head (middle), and distance from the base of the brain (bottom).

DETAILED DESCRIPTION

[0036] Section headings as used in this section and the entire disclosure herein are merely for organizational purposes and are not intended to be limiting.

[0037] All publications, patent applications, patents and other references mentioned herein are incorporated by reference in their entirety. 1. Definitions

[001] The otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art. In case of conflict, the present document, including definitions, will control. Preferred methods and materials are described below, although methods and materials similar or equivalent to those described herein can be used in practice or testing of the present disclosure. The phrase “in one embodiment” as used herein does not necessarily refer to the same embodiment, though it may. Furthermore, the phrase “in another embodiment” as used herein does not necessarily refer to a different embodiment, although it may. Thus, as described below, various embodiments of the invention may be readily combined, without departing from the scope or spirit of the invention. All publications, patent applications, patents and other references mentioned herein are incorporated by reference in their entirety. The materials, methods, and examples disclosed herein are illustrative only and not intended to be limiting.

[002] The terms “comprise(s),” “include(s),” “having,” “has,” “can,” “contain(s),” and variants thereof, as used herein, are intended to be open-ended transitional phrases, terms, or words that do not preclude the possibility of additional acts or structures. The singular forms “a,” “and” and “the” include plural references unless the context clearly dictates otherwise. The present disclosure also contemplates other embodiments “comprising,” “consisting of’ and “consisting essentially of,” the embodiments or elements presented herein, whether explicitly set forth or not.

[003] For the recitation of numeric ranges herein, each intervening number there between with the same degree of precision is explicitly contemplated. For example, for the range of 6- 9, the numbers 7 and 8 are contemplated in addition to 6 and 9, and for the range 6.0-7.0, the number 6.0, 6.1, 6.2, 6.3, 6.4, 6.5, 6.6, 6.7, 6.8, 6.9, and 7.0 are explicitly contemplated. Ranges can be expressed herein as from “about” one particular value, and/or to “about” another particular value. When such a range is expressed, another embodiment includes from the one particular value and/or to the other particular value. Similarly, when values are expressed as approximations, by use of the antecedent “about,” it will be understood that the particular value forms another embodiment. It will be further understood that the endpoints of each of the ranges are significant both in relation to the other endpoint, and independently of the other endpoint. It is also understood that there are a number of values disclosed herein, and that each value is also herein disclosed as “about” that particular value in addition to the value itself. For example, if the value “10” is disclosed, then “about 10” is also disclosed. It is also understood that when a value is disclosed that “less than or equal to” the value, “greater than or equal to the value” and possible ranges between values are also disclosed, as appropriately understood by the skilled artisan. For example, if the value “10” is disclosed the “less than or equal to 10” as well as “greater than or equal to 10” is also disclosed.

[004] It is also understood that the throughout the application, data is provided in a number of different formats, and that this data, represents endpoints and starting points, and ranges for any combination of the data points. For example, if a particular data point “10” and a particular data point 15 are disclosed, it is understood that greater than, greater than or equal to, less than, less than or equal to, and equal to 10 and 15 are considered disclosed as well as between 10 and 15. It is also understood that each unit between two particular units are also disclosed. For example, if 10 and 15 are disclosed, then 11, 12, 13, and 14 are also disclosed.

[005] “Comprising” is intended to mean that the compositions, methods, etc. include the recited elements, but do not exclude others. “Consisting essentially of “ when used to define compositions and methods, shall mean including the recited elements, but excluding other elements of any essential significance to the combination. Thus, a composition consisting essentially of the elements as defined herein would not exclude trace contaminants from the isolation and purification method and pharmaceutically acceptable carriers, such as phosphate buffered saline, preservatives, and the like. “Consisting of’ shall mean excluding more than trace elements of other ingredients and substantial method steps for administering the compositions provided and/or claimed in this disclosure. Embodiments defined by each of these transition terms are within the scope of this disclosure.

[006] A “control” is an alternative subject or sample used in an experiment for comparison purposes. A control can be “positive” or “negative.”

[007] “Optional” or “optionally” means that the subsequently described event or circumstance may or may not occur, and that the description includes instances where said event or circumstance occurs and instances where it does not.

[008] An “increase” can refer to any change that results in a greater amount of a symptom, disease, composition, condition or activity. An increase can be any individual, median, or average increase in a condition, symptom, activity, composition in a statistically significant amount. Thus, the increase can be a 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, or 100% increase so long as the increase is statistically significant.

[009] A “decrease” can refer to any change that results in a smaller amount of a symptom, disease, composition, condition, or activity. A substance is also understood to decrease the genetic output of a gene when the genetic output of the gene product with the substance is less relative to the output of the gene product without the substance. Also, for example, a decrease can be a change in the symptoms of a disorder such that the symptoms are less than previously observed. A decrease can be any individual, median, or average decrease in a condition, symptom, activity, composition in a statistically significant amount. Thus, the decrease can be a 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 15, 20, 25, 30, 35, 40, 45, 50, 55, 60, 65, 70, 75, 80, 85, 90, 95, or 100% decrease so long as the decrease is statistically significant.

[010] “Inhibit,” “inhibiting,” and “inhibition” mean to decrease an activity, response, condition, disease, or other biological parameter. This can include but is not limited to the complete ablation of the activity, response, condition, or disease. This may also include, for example, a 10% reduction in the activity, response, condition, or disease as compared to the native or control level. Thus, the reduction can be a 10, 20, 30, 40, 50, 60, 70, 80, 90, 100%, or any amount of reduction in between as compared to native or control levels.

[OH] “Reduce” or other forms of the word, such as “reducing” or “reduction,” generally means a lowering of an event or characteristic. It is understood that this is typically in relation to some standard or expected value, in other words it is relative, but that it is not always necessary for the standard or relative value to be referred to.

[012] By “prevent” or other forms of the word, such as “preventing” or “prevention,” is meant to stop a particular event or characteristic, to stabilize or delay the development or progression of a particular event or characteristic, or to minimize the chances that a particular event or characteristic will occur. Prevent does not require comparison to a control as it is typically more absolute than, for example, reduce. As used herein, something could be reduced but not prevented, but something that is reduced could also be prevented. Likewise, something could be prevented but not reduced, but something that is prevented could also be reduced. It is understood that where reduce or prevent are used, unless specifically indicated otherwise, the use of the other word is also expressly disclosed.

[013] The term “subject” refers to any individual who is the target of administration or treatment. The subject can be a vertebrate, for example, a mammal. In one aspect, the subject can be human, non-human primate, bovine, equine, porcine, canine, or feline. The subject can also be a guinea pig, rat, hamster, rabbit, mouse, or mole. Thus, the subject can be a human or veterinary patient. The term “patient” refers to a subject under the treatment of a clinician (e.g., physician). [014] The term “therapeutically effective” refers to the amount of the composition used is of sufficient quantity to ameliorate one or more causes or symptoms of a disease or disorder. Such amelioration only requires a reduction or alteration, not necessarily elimination.

[015] The term “treatment” refers to the medical management of a patient with the intent to cure, ameliorate, stabilize, or prevent a disease, pathological condition, or disorder. This term includes active treatment, that is, treatment directed specifically toward the improvement of a disease, pathological condition, or disorder, and also includes causal treatment, that is, treatment directed toward removal of the cause of the associated disease, pathological condition, or disorder. In addition, this term includes palliative treatment, that is, treatment designed for the relief of symptoms rather than the curing of the disease, pathological condition, or disorder; preventative treatment, that is, treatment directed to minimizing or partially or completely inhibiting the development of the associated disease, pathological condition, or disorder; and supportive treatment, that is, treatment employed to supplement another specific therapy directed toward the improvement of the associated disease, pathological condition, or disorder.

[0038] Certain methods and materials are described below, although methods and materials similar or equivalent to those described herein can be used in practice or testing of the present disclosure. All publications, patent applications, patents and other references mentioned herein are incorporated by reference in their entirety. The materials, methods, and examples disclosed herein are illustrative only and not intended to be limiting.

2. Systems and Methods for Brain Source Localization

[0039] As described further herein, embodiments of the present disclosure include a novel approach to optimize the recording sensitivity of sEEG electrode configurations and thereby minimize the number of implanted electrodes while maintaining recording coverage and accurate localization. Automated planning algorithms for determining sEEG electrode implantation trajectories given a user-specified region of interest (ROI) have shown promise to increase grey matter sampling, decrease risk scores, and reduce planning time compared to manual planning. However, these algorithms did not account for the spatial extent of tissue that can be recorded (recording sensitivity), which is critical to EZ localization. Therefore, patientspecific head models were implemented to simulate the spatial distribution of voltages generated by spatially extended sources of epileptiform neural activity. The recording sensitivity of arbitrary sEEG configurations were quantified and an optimization method was developed to identify electrode trajectories that maximize the recording sensitivity of user- defined regions of interest while avoiding critical anatomy. This approach transforms sEEG planning from a manual ad hoc process to an intelligent mapping to improve EZ localization and patient safety.

[0040] As described further herein, optimized sEEG electrode configurations, based on patient-specific models, yielded substantial increases in recording sensitivity, and this may lead to more accurate delineation of the EZ, and, thereby, improved outcomes of surgical resection. Further, the optimized configurations had recording sensitivity equivalent to or greater with fewer electrodes than the clinically implanted configurations, and this may reduce the risk for complications, including intracranial hemorrhage.

[0041] The present disclosure describes the implementation and evaluation of an optimization algorithm to determine implantation trajectories to map any clinician-defined region of interest. Previous automated implantation trajectory planning algorithms were focused exclusively on finding “safe” trajectories that intersected a region of interest and did not incorporate measures of recording sensitivity. The optimized electrode trajectories described herein produced larger cortical coverage than manual trajectory planning, and thus have a greater probability of recording relevant epileptiform activity.

[0042] The recording sensitivity of optimized configurations transferred across patients was inferior to the sensitivity of patient-specific configurations. Differences in the recording sensitivity of individual contacts and differences in cortical geometry between patients influenced recording sensitivity more than general location of the electrodes. Therefore, standard configurations that perform well across most patients may not exist, and optimization of sEEG trajectory planning requires patient-specific modeling.

[0043] The methods described herein to calculate the recording sensitivity of any electrode configuration enable clinicians to visualize the coverage generated by their preoperative plan in comparison to the region of interest. They can then add, remove, and move electrodes to create configurations with high recording sensitivity within the region of interest. These methods transform the clinical use of sEEG from a discrete ad hoc sampling to an intelligent, data informed mapping of the region of interest.

[0044] The distances from an electrode contact that readily detect neural sources were similar to the spatial extents of neural sources (1.5 cm vs 1.39 cm for 10 cm² and 500 pV threshold). Therefore, for large recorded signals, a portion of the neural source is likely <5 mm from the electrode contact. However, clinically the size and center of the source are used to design resection plans and these source parameters cannot be intuitively determined because there is large variability in the recording sensitivity of sEEG electrode contacts as a function of distance from the contact. Sources near the electrode contact may generate smaller signals compared to those further away, based on the orientation of the sources within the active tissue and the location of the contact. Therefore, delineating the center of an epileptic source requires the use of sEEG source localization algorithms. Even though the recording sensitivity for large amplitude signals is limited, source analysis with these signals is useful and accurate for sEEG. Additionally, the recording sensitivity metric identifies the cortical regions that can be recorded by a set of sEEG electrodes, and this information can be used to constrain localization of the EZ, which will simplify the source localization problem and thereby improve accuracy.

[0045] Neural sources >200 p V were distinguishable from noise, while most epileptiform signals are only analyzed if they are >1000 pV. If only large signals are localized clinically, epileptic sources that generate smaller signals and are slightly farther from the electrodes may be missed. Recent spatiotemporal source localization algorithms using EEG reconstructed neural signals with a signal to noise ratio as low as 5 dB. Therefore, new source localization algorithms might be able to take advantage of lower amplitude signals (200 pV) to localize neural sources that were previously not visible to the clinician. Further, detecting lower amplitude signals allows greater recording sensitivity and fewer electrodes to be used in optimized electrode configurations.

[0046] The results of the present disclosure demonstrate clear performance benefits of optimization of electrode configurations, and future implementations could also include more accurate methods to avoid critical anatomy (e.g., vasculature, ventricles). The methods to generate sulcal surfaces selected large areas in the temporal lobes and likely overestimated the area around large blood vessels and thereby limited the search space of valid electrodes in the temporal lobes. While sulcal surfaces have been used for vasculature avoidance, vasculature surfaces can be extracted using CT angiography. Direct avoidance of vasculature could improve the safety profiles of trajectories and increase the search space of valid electrode implantation areas. Additionally, while all optimized configurations satisfied conservative risk metrics, future optimization methods could not only maximize recording sensitivity but also minimize risk of configurations.

[0047] In accordance with the above, embodiments of the present disclosure generally include a method for visualizing electrode recording sensitivity. In some embodiments, the method includes obtaining simulated voltage recordings from a patient’s brain model using a plurality of electrodes present in the model. In some embodiments, the method includes calculating a patient-specific map of recordable tissue based on the simulated voltage recordings from the plurality of electrodes. In some embodiments, the method includes transforming the patient-specific map of recordable tissue onto a model of a human brain. In some embodiments, the method includes calculating a yield corresponding to at least one measurement of recording sensitivity on the model. In some embodiments, the simulated voltage recordings are based on threshold values. As would be recognized by one of ordinary skill in the art based on the present disclosure, the method for visualizing electrode recording sensitivity can include one or more of the aforementioned steps in any order or arrangement. [0048] In some embodiments, the yield corresponds to at least one measurement of recording sensitivity. The measurement of recording sensitivity includes, but is not limited to, a mean, a median, a sum, and/or a maximum measurement of recording sensitivity. In some embodiments, the at least one of a mean, a median, a sum, and/or a maximum measurement of recording sensitivity is obtained from a plurality of electrodes intersecting a defined point on the model.

[0049] In some embodiments, the model of the human brain is a standard human brain or scalp. In some embodiments, the model of the human brain is a patient-specific human brain or scalp. In some embodiments, the method further comprises combining a plurality of patientspecific maps to form a generalized map of electrode recording sensitivity. In some embodiments, the generalized map of electrode recording sensitivity is transformed onto a standard human brain or scalp. In some embodiments, the generalized map of electrode recording sensitivity is transformed onto a patient-specific human brain or scalp. In some embodiments, the patient has or is suspected of having a seizure disorder.

[0050] In some embodiments, the method further comprises removing simulated voltage recordings from one or more confounding electrodes. In some embodiments, the one or more confounding electrodes includes electrodes that intersect sulci. In some embodiments, the one or more confounding electrodes includes electrodes that intersect the brain midline. In some embodiments, the one or more confounding electrodes includes electrodes that intersect a ventricle(s). In some embodiments, the one or more confounding electrodes includes electrodes that intersect vasculature. In some embodiments, the one or more confounding electrodes includes electrodes that intersect other electrodes. In some embodiments, the plurality of electrodes are compatible with electroencephalography (EEG), stereo-electroencephalography (sEEG), and/or magnetoencephalography (MEG).

[0051] Embodiments of the present disclosure also include a method for determining electrode implantation locations and trajectories. In accordance with these embodiments, the method includes obtaining simulated voltage recordings from a patient’s brain model using a plurality of electrodes present in the model. In some embodiments, the method includes calculating a patient-specific map of recordable tissue based on the simulated voltage recordings from the plurality of electrodes. In some embodiments, the method includes obtaining voltage recordings from at least one electrode positioned to record from a user- selected anatomical region-of-interest. In some embodiments, the method includes optimizing number, location, and/or implantation trajectory of each of the remainder of the plurality of electrodes to maximize a yield corresponding to at least one measurement of recording sensitivity in the region-of-interest. In some embodiments, the simulated voltage recordings are based on threshold values. As would be recognized by one of ordinary skill in the art based on the present disclosure, the method for determining electrode implantation locations and trajectories can include one or more of the aforementioned steps in any order or arrangement.

[0052] In some embodiments, the yield corresponds to at least one measurement of recording sensitivity. The measurement of recording sensitivity includes, but is not limited to, a mean, a median, a sum, a weighted sum, and/or a maximum measurement of recording sensitivity. In some embodiments, the at least one of a mean, a median, a sum, and/or a maximum measurement of recording sensitivity is obtained from a plurality of electrodes intersecting a defined point on the model. In some embodiments, the plurality of electrodes are compatible electroencephalography (EEG), stereo-electroencephalography (sEEG), and/or magnetoencephalography (MEG).

[0053] In some embodiments, the method produces a patient-specific optimized electrode configuration. In some embodiments, the patient has or is suspected of having a seizure disorder. In some embodiments, the method produces an optimized electrode configuration that maximizes at least one of a mean, a median, a sum, a weighted sum, and/or a maximum measurement of recording sensitivity.

[0054] As described further herein, the method includes obtaining voltage recordings from an electrode positioned to record from a user-selected anatomical region-of-interest, and then optimizing the number, location, and/or implantation trajectory of the remainder of the plurality of electrodes to maximize a yield corresponding to at least one measurement of recording sensitivity in the region-of-interest. In some embodiments, the number, location, and/or implantation trajectory of one or more of the remainder of the plurality of electrodes are determined manually by the user. In some embodiments, the method includes updating the patient-specific map of recording sensitivity in real-time as the user manipulates the number, location, and/or implantation trajectory of at least one of the plurality of electrodes. In other embodiments, the location of the at least one electrode positioned to record from a user-selected anatomical region-of-interest is determined manually by the user, and then the optimization is conducted to determine the number, location, and/or implantation trajectory of each of the remainder of the plurality of electrodes. In some embodiments, the number, location, and/or implantation trajectory of at least one of the remainder of the plurality of electrodes are determined by performing the method described above multiple times on one or more regions- of-interest. In some embodiments, the user assigns a significance value to multiple regions-of- interest.

[0055] The systems and methods of the present disclosure can be implemented in hardware, software, firmware, or combinations of hardware, software and/or firmware. In some examples, the systems and methods described in this specification may be implemented using a non- transitory computer readable medium storing computer executable instructions that when executed by one or more processors of a computer cause the computer to perform operations. Computer readable media suitable for implementing the systems and methods described in this specification include non-transitory computer-readable media, such as disk memory devices, chip memory devices, programmable logic devices, random access memory (RAM), read only memory (ROM), optical read/write memory, cache memory, magnetic read/write memory, flash memory, and application-specific integrated circuits. In addition, a computer readable medium that implements a system or method described in this specification may be located on a single device or computing platform or may be distributed across multiple devices or computing platforms.

[0056] One skilled in the art will readily appreciate that the present disclosure is well adapted to carry out the objects and obtain the ends and advantages mentioned, as well as those inherent therein. The present disclosure described herein are presently representative of preferred embodiments, are exemplary, and are not intended as limitations on the scope of the present disclosure. Changes therein and other uses will occur to those skilled in the art which are encompassed within the spirit of the present disclosure as defined by the scope of the claims.

3. Materials and Methods

[0057] Semi-automated patient-specific volume conductor head modeling. The Duke University Health System IRB approved the use of clinical neuroimaging in this study to do secondary research on data collected as part of research study Pro00101171, and the participants whose neuroimaging was used provided written informed consent. From these data, 12 epilepsy patients over the age of 18 that had full sets of neuroimaging were selected. A semi-automated pipeline was developed to implement patient-specific head models combining patient-specific neuroimaging (T1 MRI, diffusion weighted MRI (DW MRI), and PostOp CT), implantation planning coordinates, and custom code (Fig. 1). Patient-specific head modeling was divided into four modules: 1) Geometry creation, 2) Defining the tissue electrical properties, 3) Electrode generation, and 4) Finite element model generation.

[0058] The model geometry was created using the T1 MRI. The skin surface was extracted using FSL (BET) and defined the white-gray matter boundary using freesurfer’s (surfer.nmr.mgh.harvard.edu/) recon-all function. The white-gray matter boundary mesh was manually smoothed by decimating and up-sampling the mesh in meshlab (meshlab.net).

[0059] The tissue electrical properties were defined by segmenting the T1 MRI into five different tissue types (Skin, Skull, CSF, White Matter, and Grey Matter) using FSL’s (fsl.fmrib.ox.ac.uk) analysis tools (FAST and BET). Diffusion tensors were then obtained from the DW-MRI using FSL and co-registered the tensors to T 1 space. Finally, the diffusion tensors were converted to conductivity tensors using the load preservation technique to define anisotropic conductivity tensors for each patient-specific model.

[0060] Electrode geometries were generated using the PostOp CT and a library of predefined electrode geometries created by PMT and ADTech. DEETO was used to localize the sEEG electrode contacts given the entry and target locations of each electrode along with the PostOp CT. The electrode contact locations were then coregistered to T 1 space and a line was fit to each electrode. If simulating the clinically implanted electrodes, a mesh electrode geometry was placed on each trajectory. For arbitrary trajectory simulations, the electrode geometries in the FEM were not included because changing the geometry for each simulation would require remeshing each simulation. Additionally, previous work has shown that the electrodes have minimal influence on the recorded potentials at distances >5 mm.

[0061] Additionally, all the processed components were combined into a finite element model in SCIRun v5.0 (SCI Institute, University of Utah, Salt Lake City, UT) with 19-22 million elements. SCIRun was used to solve for the voltages throughout the head model generated by dipole sources pointed orthogonally outward from the cortical surface at -40,000 cortical locations. Points at the base of the skin mesh were grounded to serve as a voltage reference in the FEM simulations. The simulated voltages were compiled for each source as columns into a leadfield matrix that defined the input-output relationship between neural sources and recorded voltages.

[0062] Extended source modelling. Simplified source models are required to simulate the voltage distribution generated by realistic epileptic sources. Dipoles are appropriate source representations of active neurons for sEEG, and extended dipole models are necessary to model patches of active cortex when recording within 1.5 cm. Three extended dipole models, or patch models, centered at every element on the cortical triangular surface mesh were defined. These were sets of adjacent dipoles on the cortical surface with surface areas of 6 cm², 10 cm², and 20 cm², which correspond to areas projected on the inside of the skull of 2.46 cm², 3.85 cm², and 7.09 cm² (FIG. 9). The spatial extents of each patch were quantified using the mean distances between the center of mass and the edges of each patch.

[0063] The voltages generated by each patch model were calculated by finding the columns of the leadfield matrix corresponding to each dipole within a patch, and each column was sealed by the area of the corresponding triangle of the cortical mesh. Then, the columns were summed and scaled by an estimate of human neocortex dipole moment density (0.16 - 0.77 nA-m/mm²) to get the voltage. By concatenating the calculated distributions of voltages for all extended dipole models as columns, the patch leadfield matrix was assembled for all pairs of three patch areas and three dipole moment density values. The minimum, mean, and maximum values of 0.16, 0.465, and 0.77 nA-m/mm² were used to capture the range of dipole moment density values. The mean values of 10 cm² patch area and 0.465 nAm/mm² dipole moment density were used in all analyses to capture the best-estimated results.

[0064] Estimation of recording radius. The radius around a contact within which discernible neural signals could be recorded was estimated. Discernible voltage was defined as the amplitude of a signal spike that would consistently be discriminated from background noise in sEEG signals. Current clinical practice relies on identification of large interictal spikes typically >1 mV amplitude (FIG. 5). Interictal sEEG voltage data was analyzed from 12 patients to determine the distribution of noise and spike amplitude. The amplitude of the noise was determined by taking the standard deviation of the recorded voltages when there was no spiking activity and multiplied it by four (FIG. 5A). The noise distribution was roughly Gaussian, so four times the standard deviation will account for 99.99% of the data. Based on the standard deviation of the noise, interictal spikes >200 pV in magnitude (± four standard deviations) can be differentiated from noise (FIG. 5B). Using 120 clinician defined spikes across 10 patients, the maximum voltage across all the sEEG recordings were calculated using a common average referencing scheme. The 25 percentile maximum voltage was 548 p V and the mean maximum voltage was 958 pV. Therefore, voltage thresholds of 200 pV, 500 pV, and 1000 pV were used in all analyses (Table 1).

[0065] Table 1. Number of electrodes for >75% mean recording sensitivity for standard optimized configurations with standard deviation.

[0066] To estimate recording radius, 105 simulated contact locations were first selected from each hemisphere of each patient using three random points on the cortex surface from each of 35 cortical subregions based on the Desikan-Killiany atlas. The rows from the patch leadfield matrices corresponding to these locations were selected and one of the three discernible voltage thresholds was applied to determine the patches that were recordable by each contact location. The distance from the center element of every patch to each contact location was calculated and used to sort the patches into groups of radii from each contact with a 0.25 cm bin size. For each contact, there was at least one patch in every group of radii, and each patch was only a member of one group. The percent of patches within each group that were recordable by each contact (recording sensitivity) were calculated. This procedure was repeated for the same contact locations using all nine source types and all three discernible voltage thresholds and the median recording sensitivity across radii was calculated. The recording radius was determined as the maximum distance at which >50% of sources (median) had >20% recording sensitivity.

[0067] Quantification of configuration recording sensitivity. A visualization tool and metric were developed to quantify the recording sensitivity, or the extent of tissue in a region of interest that is recordable by a given electrode configuration. sEEG source localization requires multiple contacts to record discernible signals from a source, so the recordable patches for an electrode configuration were defined to be those that generated discernible signals at a minimum of two contacts on any single electrode. To find the set of recordable patches, all columns of the patch leadfield matrix corresponding to sources with central elements inside the region of interest were selected and a discernible recording threshold (200, 500, or 1000 pV) was applied. Then, for each electrode the rows corresponding to its 16 contacts were summed and concatenated, and a threshold of two was applied to obtain a logical matrix representing the recordable patches for every electrode.

[0068] An element of the cortex was recordable by a configuration if the patch model centered at that element was recordable, and the recording strength of the patch was the number of electrodes for which that patch was recordable. The recording sensitivity of an electrode configuration was visualized by plotting the recording strength of each element on the cortex on a 3D surface plot. Recording sensitivity, RSi,thr(ROI) , was quantified as the percentage of patches in a region of interest that were recordable by a configuration:

10069] RS _thr(RO!) = P^ ROI / P_total ROI (1),

[0070] where P,jhr(ROI) is the number of patch models in a ROI that are recordable by at least two contacts on any electrode of a configuration, i, at a certain threshold, thr, and is the total number of patch models with center elements inside the ROI.

is also equal to the total number of elements inside the ROI because one patch around each element was generated.

[0071] Optimization of electrode number and placement. Sets of valid electrode placements were generated and optimized electrode configurations were built for 12 patients given clinician-defined regions of interest (ROIs).

[0072] Generation of valid electrode set. To finely sample the search space of valid implantation area but keep the leadfield matrix less than 250 GB, a set of 95,000 valid electrode trajectories per patient was computed. Valid electrodes satisfied safety criteria of insertion location, maximum insertion angle, maximum trajectory length, and maximum distance to critical structures. Electrodes with 16 contacts each, 2 mm contact length, and 1.5 mm insulator length were molded based on PMT-2102-16-091 electrodes. Each contact was simplified to a single point recording location at its center, an approximation that was previously validated. Valid entry locations were defined by selecting all points on the patient scalp surface that were within 5 mm of a best-fit standard template of implantation scalp area. 800 insertion locations from this region and 366 insertion angles were sampled for each to evenly sample the implantation space with about 300,000 electrode lines (defined by insertion location and angle), assuming that most would be excluded. The implantation angle was kept at < 10 degrees from normal to the scalp. These lines were discretized into individual electrode trajectories by sampling insertion depth at multiples of 3.5 mm, keeping the total trajectory length less than 10 centimeters. Because the contact points were also 3.5 mm apart, most contact points for electrodes in the same line overlapped, allowing representation of a greater number of electrodes with fewer recording locations.

[0073] Electrodes were eliminated that intersected critical structures — sulci, the midline, and secondary skull locations. Angiogram imaging of blood vessels was not available and because sulci are often used to estimate locations of large vasculature and as critical structures themselves, the occurrence of intersections between electrodes and sulci surfaces were calculated to avoid areas where large blood vessels are likely to be. The sulci surfaces were generated for each hemisphere individually by taking the intersection of the cortex surface with a super-smoothed cortex surface (MeshLab, filter ’hc_laplacian_smoothing’ 100 times). The resulting surfaces were the bases of sulci. To calculate intersections, a bounding volume hierarchy method adapted from Sparks et al., was used where the minimum distance from the electrode to the surface was found using an efficient binary tree search. All electrodes that passed within 1.5 mm of the sulci surface or 4 mm of the skull (away from insertion location) or the midline were eliminated. Lines were then randomly eliminated until 95,000 valid electrode trajectories were left. Additionally, the occurrences of collisions (distance <4 mm)⁶ between every pair of electrodes was calculated by representing each electrode as a set of 128 points and conducting matrix distance calculations.

[0074] Optimization problem and cost function. A search algorithm was designed to find the best configurations for three ROIs per patient (left temporal lobe (LTL), clinician-defined ROI, and left hemisphere (LH)) given the sets of all valid electrode trajectories and occurrences of collisions between every pair. To quantify how well a certain configuration, i, can record from a ROI with threshold, thr, the following cost function, Ci.thr(ROI), was used to represent the number of patches with center elements in the ROI that are not recordable:

[0075] C_{i thr}(RO ) = P_total(ROI) - P_{i thr}(ROr), (2)

[0076] where P^thfROI) and P_tatai(ROI) are as defined in equation (1).

[0077] The generation of an electrode configuration was conceptualized as a tree traversal through an ‘N-ary’ tree, where each node represents a valid trajectory and each node has N-x children (i.e., the number of valid electrodes minus some number ‘x’ that intersect with electrodes in the current path). Each edge represents the addition of an electrode to a configuration, and each root-to-leaf path represents a full configuration where either there are no other valid electrode choices that would reduce the cost function or the full region of interest is recordable. The optimal configuration is the path with the lowest cost at a certain number of electrodes (level of tree).

[0078] Next-best search algorithm. For any configuration of three or more electrodes, it is computationally intractable to find the true best configuration because the number of paths grows exponentially with the number of electrodes. Thus, a next -best iterative tree search was conducted to find solution configurations. A single optimization trial required the selection of many parameters — source area, source strength, ROI, and a threshold-priority order, which assigns the three thresholds (200, 500, 1000 pV) primary, secondary, and tertiary importance. At each level of the search tree, the electrode was picked to decrease the cost function maximally at the priority threshold. Ties between electrodes were broken using the cost functions at the second then third priority thresholds and finally a random choice, if necessary. The next -best search algorithm is not guaranteed to produce the optimal electrode configuration because the mapping problem does not have optimal substructure. However, given that finding the true best configuration is computationally intractable, this approach identifies a good option. The next-best search was conducted for each patient’s LTL, LH, and clinician-defined ROI, using all six permutations of cost functions (all permutations of 200 pV, 500 pV, and 1000 pV for priority order) and all nine source types (all choices of three patch area and three dipole moment density parameters). Electrodes were continually added until 31 electrodes were added or no remaining electrodes could improve the cost function. Because the algorithm was iterative, the optimized configuration of any number of electrodes, X between 1 and 31 , was the set of the first X electrodes in the configuration. The LTL surfaces were manually defined based on cortex geometry. The ROI surfaces were manually defined by selecting targeted areas of the cortex defined in clinician notes (Table 2) and subregions that did not contain electrode trajectories were excluded. The optimized configurations were analyzed using the minimum cost function for each threshold of interest.

[0079] Table 2. sEEG patient population specifications.

[0080] Quantification of clinically implanted configuration recording sensitivity. The recording sensitivities (equation (1)) of the optimized configurations were compared to those of the clinically implanted sets. The order of implanted electrodes were defined for a certain case by sorting them in the best possible order to maximally decrease the cost function (increase the recording sensitivity) at each addition.

[0081] Eight patients had clinician-defined ROIs that included, but were not limited to, one of the temporal lobes. For these patients, two versions of the implanted configuration were defined, one including only those electrodes (4 to 10) inside the temporal lobe ROI and the other including all electrodes (12 to 15) in a broad ROI. The broad ROIs for three of these patients included secondary areas on the other hemisphere, and these areas and the corresponding electrodes from consideration were excluded. An electrode in the temporal lobe configuration was included if at least one contact was within 3 mm of the ROI surface and at least half of the contacts were within 10 cm of the ROI. The recording sensitivity between the optimized configurations for the temporal lobe ROIs and the broad clinician-defined ROIs were compared to recording sensitivities of the implanted configurations. The remaining four patients had ROIs outside of or not focused on a temporal lobe. For these, the recording sensitivity of the full implanted configurations of 11 to 16 electrodes for the clinician-defined ROI was computed. [0082] Transfer of patient-specific configurations to other patients. To determine the importance of patient-specific configuration generation, the optimized configurations for the LH and LTL ROIs for each patient were transferred into all other patients. FSL’s FLIRT function with six degrees of freedom was used and a mutual information cost function was used to map configurations into other patients. Limiting the transformations to simple rotation and translation ensured consistent electrode geometries. A 10 cm² patch area, 0.465 nAm/mm² dipole moment density, and a 500 pV-priority cost function was used to generate optimized configurations. After transferring each configuration to all other patients, the set of all contact locations for the optimized and transferred configurations was found and the leadfield matrix at these locations was recomputed. The percent recording sensitivity was calculated for the LTL and LH configurations in all patients and the percent error between the optimized and transferred cases was found for each configuration. For each configuration, recording sensitivity values at the minimum number of electrodes that gave >75% recording sensitivity of the ROI in the matched-patient case were compared. The validity of all transferred configurations was tested by calculating the minimum distance of electrodes to patient sulci surfaces. Any configuration with an electrode that passed within 2.5 mm of a sulci surface was considered invalid, but all electrodes were included for analysis to understand the best-case recording sensitivity.

[0083] Statistical analysis. Statistical analyses were conducted on MATLAB using built-in functions. The data are presented as median with interquartile range (FIG. 2), standard box plots (FIGS. 3A-3G, FIGS. 4C-4D, and FIGS. 6A-6B) (median, interquartile range, maximum, and minimum), mean +/- SD (FIGS. 4A-4B), and standard histograms (FIGS. 9A-9C) based on the characteristics of the datasets. P < 0.05 was considered to be statistically significant. Shapiro-Wilk normality tests were conducted on each set of recording sensitivity and electrode number data (FIG. 3). If the optimized and implanted data were both Gaussian for one threshold and one ROI, the difference between the two was compared using a two-tailed paired t test, and if not, using a Wilcoxon signed rank test (TL: n = 8; clinician-ROI: n = 12).

[0084] Sensitivity Analysis. Sensitivity analysis was performed to assess the robustness of the optimization methods described herein to uncertainty in source parameters. Since the true source parameters of epileptic sources are unknown, experiments were conducted to determine the recording sensitivity error associated with using the wrong source type. Optimized configurations were computed for all nine source types, 12 patients, and the LTL and LH ROIs. The recording sensitivity for each configuration was analyzed with all nine source types (one matched case and eight “crossed” cases). The error due to source uncertainty was quantified by calculating the percent error in recording sensitivity between the crossed and matched cases for every source type. The minimum number of electrodes in each configuration was used such that the matched recording sensitivity was >75%. A 500 pV-priority cost function was used in all optimization cases, a 500 pV threshold for analysis, and the percent error across all 12 patients was averaged.

[0085] Sensitivity analysis was also performed to quantify the dependence of the optimization results on threshold-priority ordering of the cost function. Using the mean source type, optimized configurations were found for all six permutations of threshold-priority ordering of the cost functions, all 12 patients, and the LTL and LH ROIs. The recording sensitivity was compared for each configuration with all three thresholds (200, 500, 1000 pV), giving two matched cases (first priority threshold matches analysis threshold) and four crossed cases (first priority threshold does not match analysis threshold). The minimum number of electrodes was used in each configuration such that the maximum recording sensitivity across cases was >75%, and the percent error in recording sensitivity between the best case and all other threshold-priority cases was quantified. The error across all 12 patients was averaged.

[0086] The projected area and radii of the source models was quantified to compare more directly to recordable radius (FIG. 1). For a single patient, the area and mean radius of the sources projected onto the inside of the skull were calculated for all patches of three source areas (6 cm², 10 cm², and 20 cm²). All patches whose maximum distance to the skull was >3 cm were excluded because patches far from the skull (i.e., in the midline and the insula) did not have a clear projected area (18832 of 39995 sources). For the remaining sources, the projected patch was found by first selecting the closest point on the skull surface to each of the vertices of the cortex patch. To smooth the boundary and fill in holes, the patch vertices were dilated and eroded three times each. The faces whose vertices were all included in the set were then selected and the corresponding area was obtained. The mean radius between the centroid and all boundary points were calculated in the projected patch.

[0087] A two-factor ANOVA was conducted to analyze the recording radius across cortical subregions and patients. The maximum recording radius (largest binned radius with >20% recording sensitivity) was calculated for each of the 105 patches for each patient using the median source type (10 cm² area and 0.465 nAm/mm² dipole moment density) and all three voltage thresholds (200 pV, 500 pV, and 1000 pV). The patches were grouped based on the subregion of their central face according to the Desikan-Killiany atlas and paired t-tests with Bonferroni corrections were used for post hoc testing.

[0088] Table 3A. Test statistics for FIGS. 3D-3E. Recording sensitivity.

[0089] Table 3B. Test statistics for FIGS. 3F-3G. Number of electrodes.

4. Examples

[0090] Stereo-EEG (sEEG) is a minimally invasive technique used to localize the origin of epileptic activity (the epileptogenic zone) in patients with drug-resistant epilepsy. However, current sEEG trajectory planning methods are agnostic to the spatial recording sensitivity of the implanted electrodes. sEEG involves implanting 10-25 electrodes into the brain to record from extended neural populations. Clinicians use these recordings to localize the epileptogenic zone, a target for surgical resection to treat epilepsy. Currently, sEEG planning and analysis is largely conducted manually, and computation tools have the potential to improve the success rates of sEEG localization and surgical outcomes. For example, PCT/US2022/043430, which is incorporated herein by reference in its entirety and for all purposes, discloses four computational tools to automate, visualize, and optimize sEEG implantation and source reconstruction useful for the identification of the location of the epileptogenic zone. Embodiments of the present disclosure expand on these computation tools and includes systems and methods for the visualization of recording sensitivity of an electrode, in a patientspecific manner; systems and methods pertaining to an interactive visualization tool that determines and displays the recording sensitivity of any clinician-defined electrode location; and systems and methods pertaining to a dual-selection implantation trajectory planning technique that combines clinician-selected electrode locations and trajectories and electrode location and implantations trajectories determined by optimization.

[0091] As described further herein, image-based patient-specific computational models were used to design optimized sEEG electrode configurations. Patient-specific optimized electrode configurations exhibited substantially higher recording sensitivity than clinically implanted configurations, and this may lead to more accurate delineation of the epileptogenic zone. The optimized configurations also achieved as good as or better recording sensitivity with fewer electrodes compared to clinically implanted configurations, and this may reduce the risk for complications including intracranial hemorrhage. This approach improves localization of the epileptogenic zone by transforming the clinical use of sEEG from a discrete ad hoc sampling to an intelligent mapping of the regions of interest.

[0092] It will be readily apparent to those skilled in the art that other suitable modifications and adaptations of the methods of the present disclosure described herein are readily applicable and appreciable, and may be made using suitable equivalents without departing from the scope of the present disclosure or the aspects and embodiments disclosed herein. Having now described the present disclosure in detail, the same will be more clearly understood by reference to the following examples, which are merely intended only to illustrate some aspects and embodiments of the disclosure, and should not be viewed as limiting to the scope of the disclosure. The disclosures of all journal references, U.S. patents, and publications referred to herein are hereby incorporated by reference in their entireties.

[0093] The present disclosure has multiple aspects, illustrated by the following non-limiting examples.

Example 1

[0094] Clinical strategies for sEEG electrode implantation maximize the number of electrodes placed in a region of interest (ROI) that avoid vasculature and critical structures. This approach assumes that recordable signals are inherently localized and are generated by neural activity <5 mm away from a contact. However, the true spatial extent of recording sensitivity is unknown. Optimization of sEEG electrode trajectories to maximize recording sensitivity requires a robust understanding of how the complex source geometry and electrical conductivities of the brain influence the spatial extent of recording sensitivity. Therefore, the spatial extent of recording sensitivity was first quantified from simulated sEEG electrodes throughout the brain. Subsequently, optimized sEEG electrode configurations were designed that had substantially higher recording sensitivity than clinically implanted configurations, and the optimized configurations required fewer implanted electrodes to achieve as good as or better recording sensitivity than the clinically implanted configurations.

[0095] sEEG contacts record discernible signals from realistic epileptiform neural activity up to 1.5 cm away. To determine the distance that sEEG contacts can record discernible signals from neural (epileptiform) sources, a semi-automated patient-specific head modeling pipeline was developed to simulate the voltages generated by cortical dipole sources throughout the brain (FIG. 1). Epileptiform sources were simulated generating interictal spikes as patches of active cortex with 10 cm² area²³ and 0.465 nAm/mm² source strength at the peak of the spike. Sources with origins as far as 1.5 cm from the recording contacts generated readily detectable voltages (>500 pV at 20% sensitivity threshold) (FIG. 2C), but patches of active cortex are spatially extended with an average radius of 1.39 cm for 10 cm² patches (FIG. 1). Therefore, intracranial electrode contacts generally record readily detectable signals from neural sources where a portion of the source is coincident with the recording electrode. However, there was large variability across contacts in recording sensitivity (percent of sources that are recordable) (equation (1)) as a function of distance from the contact, and in some cases, larger signals were seen at farther distances from the contact compared to closer distances (FIGS. 2A-2C). Therefore, there is a complex relationship between contact location and recording sensitivity that necessitates patient-specific extended source modeling to determine accurately the recording sensitivity of each electrode contact.

[0096] There is uncertainty in the area and strength of the source used to represent epileptiform activity, and the recording sensitivity was quantified across a range of physiologically relevant source areas (6 cm², 10 cm², and 20 cm²) and strengths (0.16 nAm/mm², 0.465 nAm/mm², and 0.77 nAm/mm²). Similar trends of recording sensitivity were observed as a function of distance across source areas and strengths, but with substantial differences in scale where the strongest source (20 cm² and 0.77 nAm/mm²) generated signals >200 pV up to 3.5 cm away while the weakest source (6 cm² and 0.16 nAm/mm²) never achieved >20% sensitivity for signals >200 pV (FIG. 2D).

Example 2

[0097] Optimized sEEG configurations yielded higher recording sensitivity than clinically implanted configurations. Because sEEG contacts can record appreciable signals from neural sources located cms from the recording site, there is likely overlap in the regions recorded by multiple electrodes, and optimization can determine a set of sEEG trajectories with high recording sensitivity. sEEG configurations were optimized to record from the left temporal lobe (LTL), left hemisphere (LH), and clinician-defined ROIs for 12 patients implanted with sEEG electrodes (Table 2). The optimization algorithm iteratively identified electrode trajectories that minimized the number of regions (cortical patch sources) in the ROI that were not yet recordable.

[0098] Optimized configurations exhibited far broader coverage of the ROI compared to the clinically implanted configurations using the same number of electrodes (FIGS. 3A-3B). Optimized configurations had significantly higher recording sensitivities (2-47% higher) than implanted configurations in all patients for all voltage thresholds in both the LTL and in clinician-defined ROIs (P < 0.02) (FIGS. 3D-3E). Further, optimized configurations required significantly fewer electrodes (0-11 fewer) (P < 0.02) than implanted configurations to achieve the same or better recording sensitivity (FIGS. 3F-3G). Finally, the recording sensitivity of optimized configurations was higher than implanted configurations across all source areas, source strengths, signal detection thresholds, and ROIs (FIG. 6).

Example 3

[0099] Patient-specific optimization is necessary. Optimized electrode configurations were more sensitive than those implanted clinically, and experiments were conducted to determine whether patient-specific optimization was necessary or if there were consistent optimal configurations to map the LTL and LH across patients. Patient-specific electrode configurations yielded comparable recording sensitivities as a function of the number of implanted electrodes across all subjects (FIGS. 4A-4B, Table 1). However, the sets of optimized electrode trajectories were not consistent across patients (FIG. 7). To quantify the benefit of patient-specific optimization, each optimized configuration was transferred into all other patients and compared the recording sensitivities of the patient-specific and transferred configurations. The transferred configurations had on average 30.5% lower recording sensitivity than the patient-specific configurations for LTL ROIs (FIG. 4C) and 20.1% lower recording sensitivity for LH ROIs (FIG. 4D). Further, all of the transferred configurations had invalid electrodes that intersected sulci and performance would decline further if these were removed. Therefore, standard optimal configurations did not exist, and patient-specific optimization was required to maximize the recording sensitivity of sEEG electrode configurations. Example 4

[0100] Assuming that the mean source type was ground truth, the error in optimizing was computed with all other source types. There was a maximum average error of 24% for the LTL (FIGS. 8A and 8C, boxed row) and 27% for the LH (FIGS. 4B and 4D, boxed row). Assuming that optimization was run using the mean source type, the error associated with the ground truth source type being any of the other source types was also quantified. There was a maximum average error of 38% for the LTL (FIG. 8C, boxed column) and 58% for the LH (FIG. 8C, boxed column). However, the difference in dipole moment density dominated the errors, and the mean source type optimizations (boxed columns) had < 17% error excluding 0.16 nAm/mm² cases.

[0101] The error in recording sensitivity between threshold-priority order cases with the same first priority threshold was small (<3.3%) (FIGS. 8E-8H). Using 500 pV-priority order in optimization, there was <17% error in recording sensitivity for the LTL and <25% for the LH when compared to optimizations with different first priority thresholds (FIGS. 8G-8H). The only large error was for 200 pV -priority thresholds analyzed at 1000 pV for LH configurations (41%) (FIG. 8H). Therefore, the choice of first priority threshold is very important for appropriate optimization.

[0102] For the projected patch analysis, it was found that the 10 cm² sources had an average projected area of 3.85 cm² and mean radius of 1.39 cm (FIG. 9). The average projected area was -40% the surface area of source models (6 cm², 10cm², 20cm²).

[0103] For the 200 pV voltage threshold, there was a significant difference in the maximum recording radius across cortex subregions (2-factor ANOVA; P < 0.001, F = 2.15). For the other thresholds (500 pV and 1000 pV), there was no significant difference in the maximum recording radius across cortex subregions (2-factor ANOVA; P = 0.093, F = 1.34). For all three thresholds, there was a significant difference in the maximum recording radius across patients (P < 0.0001). Test statistics were included for all tests in Table 4.

[0104] Table 4. Two-factor ANOVA statistics for maximum recording radius across cortex subregion and patients for the median source type on all three discernible voltage thresholds. Bold values show significance (P < 0.05).

Client Ref. No. DU7982PCT Atty. Docket No. DUKE-42488.601

[0105] Although there was a significant difference in maximum recordable radius for the 200 p V threshold across cortex subregions, post hoc testing revealed no significant differences between any pairs of temporal subregions (transverse temporal, superior temporal, middle temporal, inferior temporal, fusiform, temporal pole, para-hippocampal, and entorhinal cortex) (Tables 5 and 6). All significant pairs (4 of 595) included the inferior parietal region.

[0106] Table 5. T-values matrix for post hoc t-tests for maximum recording radius between temporal regions for all patients with a 200 pV threshold. No pairs are significant. Lateral temporal regions are shaded gray.

Client Ref. No. DU7982PCT

Atty. Docket No. DUKE-42488.601

|0107] Table 6. P-value matrix for post hoc t-tests for maximum recording radius between temporal regions for all patients with a 200 pV threshold. No pairs are significant.

Example 5

[0108] One aspect of the present disclosure includes visualization of sEEG electrode recording sensitivity based on insertion location (FIGS. 10A-10B). Current clinical practice relies on the intuition of a clinician in determining the best electrode trajectories to record neural signals arising from a specific anatomical region of interest (e.g., temporal lobe). Combining patient-specific head models and calculations of the region of tissue from which a specific electrode can detect neural signals (i.e., the recording sensitivity of an electrode), embodiments of the present disclosure include the development a metric for the visualization of the expected region of recording sensitivity at each valid potential electrode insertion location on the skin.

[0109] Two embodiments of this method were developed: one for patient-specific analysis and one for generalized guidelines across patients. For each patient, a set of thousands of electrodes were generated at each of about 700 insertion locations across the valid skin implantation area by varying insertion angle and depth. All electrodes that intersected the sulci surface or crossed the midline were then excluded. For all remaining (i.e., valid) electrode insertion points and electrodes, the area of recordable tissue was calculated (i.e., electrode sensitivity) and averaged them to obtain the patient-specific expected recordable area for each insertion location. The patient-specific insertion locations were then transformed onto a standard scalp surface and the maps of many patients were combined to obtain the generalized map. Clinicians can visualize the expected area map for any specific patient and the generalized map to determine which implantation locations will lead to more informative sEEG electrode locations for epileptogenic zone delineation.

Example 6

[0110] In another aspect, the present disclosure provides an interactive visualization tool of the tissue recordable by a set of sEEG electrodes. This interactive visualization tool determines and displays the recording sensitivity of any clinician-defined sEEG electrode location. Each electrode location can be dynamically specified by the user by dragging and dropping the electrode within a visualization of patient specific head anatomy. At each location, the resulting recording sensitivity (the region of tissue from which a specific electrode can detect neural signals) is updated and displayed. A user can manipulate a single electrode at a time by translating or rotating the electrode until the electrode is in their desired location. The method iterates through the new locations of the electrode contacts and calculates the voltages at each contact using a patient specific finite element model to determine the cortex that can be recorded. Rather than receiving a single fixed visualization of the recording sensitivity of a fixed set of sEEG electrode locations, this tool provides an interactive platform to update recording sensitivity maps in real time, as electrodes are dynamically repositioned, throughout the sEEG planning process. This allows the user to fine-tune electrode locations and trajectories to avoid critical structures and target specific cortical substructures.

[0111] Additionally, this tool can be combined with the automated sEEG electrode position optimization tool to enable users to manipulate electrode trajectories based on clinical judgment, especially of difficult to define safety criteria. The optimization of electrode position would be used first to determine an optimal number and position of electrodes based on the clinician defined region of interest and safety criteria. Subsequently, the user can manually manipulate the positions and trajectories of the electrodes based on their preferences and clinical judgement. This method provides rapid updated maps of recording sensitivity as electrodes are dynamically repositioned.

[0112]

Example 7

[0113] In another aspect, the present disclosure provides a dual-selection implantation trajectory planning method. In one example, the present disclosure provides an electrode implantation trajectory planning method that combines clinician-selected electrode locations and trajectories and electrode location and implantations trajectories determined by optimization, to record from clinician-determined anatomical regions of interest. Currently, clinicians choose electrode locations by targeting specific cortical substructures, maximizing the number of electrodes in a region of interest, and avoiding major vasculature, sulci, and ventricles. This process relies on intuition and clinical judgement to estimate the area of tissue from which each electrode can record neural signals (i.e., the electrode recording sensitivity). Embodiments of the present disclosure include the development of a method to quantify the extent of recordable tissue for any electrode and identify electrode configurations, including number of electrodes, electrode position, and electrode implantation trajectory to optimally record from regions of interest (i.e., maximize the electrode recoding sensitivity in the clinician identified anatomical region of interest).

[0114] The new dual-selection process allows the user to specify or designate one or more electrode locations, for example to target a user-determined specific anatomical region of interest. The location of the user specified electrode(s) are then used as inputs to the optimization method. The optimization method then determines the number, locations, and implantation trajectories of additional electrodes to map optimally the remainder of the region of interest with the minimum number of electrodes. This method allows users to combine their own specified electrode locations and trajectories and calculated optimal electrode positions and trajectories to optimally map the regions of interest.

10115] It is understood that the foregoing detailed description and accompanying examples are merely illustrative and are not to be taken as limitations upon the scope of the disclosure, which is defined solely by the appended claims and their equivalents. Various changes and modifications to the disclosed embodiments will be apparent to those skilled in the art.

Claims

CLAIMS What is claimed is:

1. A method for visualizing electrode recording sensitivity, the method comprising: obtaining simulated voltage recordings from a patient’s brain model using a plurality of electrodes present in the model; calculating a patient-specific map of recordable tissue based on the simulated voltage recordings from the plurality of electrodes; transforming the patient-specific map of recordable tissue onto a model of a human brain; and calculating a yield corresponding to at least one measurement of recording sensitivity on the model.

2. The method of claim 1, wherein the simulated voltage recordings are based on threshold values.

3. The method of claim 1 or claim 2, wherein the yield corresponding to the at least one measurement of recording sensitivity comprises at least one of a mean, a median, a sum, and/or a maximum measurement of recording sensitivity.

4. The method of claim 3, wherein the at least one of a mean, a median, a sum, and/or a maximum measurement of recording sensitivity is obtained from a plurality of electrodes intersecting a defined point on the model.

5. The method of any one of claims 1 to 4, wherein the model of the human brain is a standard human brain or scalp, or a patient-specific human brain or scalp.

6. The method of any one of claims 1 to 5, further comprising removing simulated voltage recordings from one or more confounding electrodes.

7. The method of claim 6, wherein the one or more confounding electrodes comprise electrodes that intersect sulci, brain midline, ventricles, vasculature, and/or the other electrodes.

8. The method of any one of claims 1 to 7, wherein the plurality of electrodes are compatible with electroencephalography (EEG), stereo-electroencephalography (sEEG), and/or magnetoencephalography (MEG).

9. The method of any one of claims 1 to 8, wherein the method further comprises combining a plurality of patient-specific maps to form a generalized map of electrode recording sensitivity.

10. The method of claim 9, wherein the generalized map of electrode recording sensitivity is transformed onto a standard human brain or scalp, or a patient-specific human brain or scalp.

11. The method of any one of claims 1 to 10, wherein the patient has or is suspected of having a seizure disorder.

12. A method for determining electrode implantation locations and trajectories, the method comprising: obtaining simulated voltage recordings from a patient’s brain model using a plurality of electrodes present in the model; calculating a patient-specific map of recordable tissue based on the simulated voltage recordings from the plurality of electrodes; obtaining voltage recordings from at least one electrode positioned to record from a user-selected anatomical region-of-interest; and optimizing number, location, and/or implantation trajectory of each of the remainder of the plurality of electrodes to maximize a yield corresponding to at least one measurement of recording sensitivity in the region-of-interest.

13. The method of claim 12, wherein the simulated voltage recordings are based on threshold values.

14. The method of claim 12 or claim 13, wherein the method produces a patient-specific optimized electrode configuration.

15. The method of any one of claims 12 to 14, wherein the yield corresponding to the at least one measurement of recording sensitivity comprises at least one of a mean, a median, a sum, a weighted sum, and/or a maximum measurement of recording sensitivity.

16. The method of any one of claims 12 to 15, wherein the method produces an optimized electrode configuration that maximizes at least one of a mean, a median, a sum, a weighted sum, and/or a maximum measurement of recording sensitivity.

17. The method of any one of claims 12 to 16, wherein the plurality of electrodes are compatible electroencephalography (EEG), stereo-electroencephalography (sEEG), and/or magnetoencephalography (MEG).

18. The method of any one of claims 12 to 17, wherein the patient has or is suspected of having a seizure disorder.

19. The method of any one of claims 12 to 18, wherein the number, location, and/or implantation trajectory of at least one of the remainder of the plurality of electrodes are determined manually by the user.

20. The method of any one of claims 12 to 19, wherein a patient-specific map of recording sensitivity is updated in real-time as the user manipulates the number, location, and/or implantation trajectory of at least one of the plurality of electrodes.

21. The method of any one of claims 12 to 20, wherein the location of the at least one electrode positioned to record from a user-selected anatomical region-of-interest is determined manually by the user, followed by optimizing the number, location, and/or implantation trajectory of each of the remainder of the plurality of electrodes.

22. The method of any one of claims 12 to 21, wherein the number, location, and/or implantation trajectory of at least one of the remainder of the plurality of electrodes are determined by performing the method of claim 12 multiple times on one or more regions-of- interest.

23. The method of any one of claims 12 to 21, wherein the user assigns a significance value to multiple regions-of-interest.