Validating surgical plans using virtual resections via dynamic network brain models

Abstract

Techniques for predicting an epileptogenic zone resection surgery outcome for an epilepsy patient are presented. The techniques may include: receiving an identification of a brain location; obtaining a patient brain imaging signal, where the patient brain imaging signal represents interictal activity of the patient's brain; fitting a first dynamical network model to the patient brain imaging signal; calculating a first sink index from the first dynamical network model; constructing, using the first dynamical network model, a synthetic brain imaging signal, where the synthetic brain imaging signal corresponds to a resection at the brain location; fitting a second dynamical network model to the synthetic brain imaging signal; calculating a second sink index from the second dynamical network model; and providing, based on the first sink index distribution and the second sink index distribution, an outcome prediction for resecting the patient's brain at the brain location.

Description

VALIDATING SURGICAL PLANS USING VIRTUAL RESECTIONS VIA DYNAMIC NETWORK BRAIN MODELSCross-Reference to Related Applications

[0001] This application claims the benefit of U.S. Provisional Patent Application No 63 / 624,410, filed on January 24, 2024, the disclosure of which is incorporated herein by reference.Field

[0002] This disclosure relates generally to surgery for epilepsy.Background

[0003] Epilepsy is a neurological disorder marked by sudden recurrent episodes of abnormal electrical activity in the brain, known as seizures. More than 30% of epilepsy patients are diagnosed as epileptic drug resistant, when anti-epileptic drugs are not effective in suppressing seizures. Surgery is a hopeful alternative for seizure freedom, but can only be achieved through complete resection or disconnection of the epileptogenic zone (EZ). Unfortunately, surgical success rates vary between 30-70% because no clinically validated biological markers of the EZ exist. Furthermore, once clinicians review the data and localize the EZ, they make a surgical plan, but at the present time there is no tool that allows a clinician to test whether their surgical plan is likely to succeed.Summary

[0004] According to various embodiments, a method of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient is presented.The method includes: receiving an identification of a brain location; obtaining a patientbrain imaging signal, wherein the patient brain imaging signal represents interictal activity of the patient’s brain; fitting a first dynamical network model to the patient brain imaging signal; calculating a first sink index from the first dynamical network model; constructing, using the first dynamical network model, a synthetic brain imaging signal, wherein the synthetic brain imaging signal corresponds to a resection at the brain location; fitting a second dynamical network model to the synthetic brain imaging signal; calculating a second sink index from the second dynamical network model; and providing, based on the first sink index and the second sink index, an outcome prediction for resecting the patient’s brain at the brain location.

[0005] Various optional features of the above method embodiments include the following. The method may include: repeating, for a plurality of brain locations, the receiving, the constructing, the fitting the second dynamical model, the calculating, and the providing, whereby a plurality of outcome predictions are obtained; and surgically resecting the patient’s brain based on the plurality of outcome predictions. The providing the outcome prediction for resecting the patient’s brain at the brain location may include: determining a first measure of a first statistical distribution of the first sink index and a second measure of a second statistical distribution of the second sink index, wherein the outcome prediction is based on first measure and the second measure. The first statistical distribution of the first sink index may include a histogram, wherein the first measure comprises a measure of normality, and the second statistical distribution of the second sink index may include a histogram, wherein second first measure comprises a measure of normality. The brain imaging signal may include an Electroencephalogram (EEG) signal of the patient’s brain. The brain imaging signal may include a functional Magnetic Resonance Imaging (fMRI) Blood-Oxygen-Level-Dependent (BOLD) signal of the patient’s brain. The brainimaging signal may include a Magnetoencephalography (MEG) signal of the patient’s brain. The fitting the first dynamical network model to the patient brain imaging signal may include determining a sequence of linear time-invariant dynamical network models. The constructing the synthetic brain imaging signal may include: removing channel information from segments of the patient brain imaging signal, whereby a plurality of modified segments of the patient brain imaging signal are produced; and applying each of the sequence of linear time-invariant dynamical network models to a respective segment of the plurality of modified segments of the patient brain imaging signal. The receiving the identification of the brain location may include: displaying a representation of the patient’s brain on a screen; and obtaining an input representing the identification of the brain location.

[0006] According to various embodiments, a system for predicting an epileptogenic zone resection surgery outcome for an epilepsy patient is presented. The system includes a non-transitory computer readable medium comprising instructions, and at least one electronic processor that executes the instructions, to perform operations comprising: receiving an identification of a brain location; obtaining a patient brain imaging signal, wherein the patient brain imaging signal represents interictal activity of the patient’s brain; fitting a first dynamical network model to the patient brain imaging signal; calculating a first sink index from the first dynamical network model; constructing, using the first dynamical network model, a synthetic brain imaging signal, wherein the synthetic brain imaging signal corresponds to a resection at the brain location; fitting a second dynamical network model to the synthetic brain imaging signal; calculating a second sink index from the second dynamical network model; providing, based on the first sink index and the second sink index, an outcome prediction for resecting the patient’s brain at the brain location.

[0007] Various optional features of the above system embodiments include the following. The operations may further comprise: repeating, for a plurality of brain locations, the receiving, the constructing, the fitting the second dynamical model, the calculating, and the providing, whereby a plurality of outcome predictions are obtained; and surgically resecting the patient’s brain based on the plurality of outcome predictions. The providing the outcome prediction for resecting the patient’s brain at the brain location may include: determining a first measure of a first statistical distribution of the first sink index and a second measure of a second statistical distribution of the second sink index, wherein the outcome prediction is based on first measure and the second measure. The first statistical distribution of the first sink index may include a histogram, wherein the first measure comprises a measure of normality, and the second statistical distribution of the second sink index comprises a histogram, wherein second first measure comprises a measure of normality. The brain imaging signal may include an Electroencephalogram (EEG) signal of the patient’s brain. The brain imaging signal may include a functional Magnetic Resonance Imaging (fMRI) Blood-Oxygen-Level-Dependent (BOLD) signal of the patient’s brain. The brain imaging signal may include a Magnetoencephalography (MEG) signal of the patient’s brain. The fitting the first dynamical network model to the patient brain imaging signal may include determining a sequence of linear time-invariant dynamical network models. The constructing the synthetic brain imaging signal may include: removing channel information from segments of the patient brain imaging signal, whereby a plurality of modified segments of the patient brain imaging signal are produced; and applying each of the sequence of linear time-invariant dynamical network models to a respective segment of the plurality of modified segments of the patient brain imaging signal. The receiving the identification of the brain location may include: displaying arepresentation of the patient’s brain on a screen; and obtaining an input representing the identification of the brain location.

[0008] Combinations, (including multiple dependent combinations) of the above-described elements and those within the specification have been contemplated by the inventors and may be made, except where otherwise indicated or where contradictory.Brief Description of the Drawings

[0009] Various features of the examples can be more fully appreciated, as the same become better understood with reference to the following detailed description of the examples when considered in connection with the accompanying figures, in which:

[0010] Figs. 1A and 1 B present a flow diagram for a virtual resection method according to various embodiments;

[0011] Fig. 2 is a schematic diagram depicting a technique for fitting a dynamic network model and calculating a sink index, according to various embodiments;

[0012] Fig. 3 is a schematic diagram detailing a technique for calculating a sink index, according to various embodiments;

[0013] Fig. 4 illustrates distributions of source-sink indices for healthy individuals, according to various embodiments;

[0014] Fig. 5A illustrates sink index distributions both pre-surgery and post- virtual-surgery for a patient with a successful outcome from a corresponding actual surgery, according to an embodiment;

[0015] Fig. 5B illustrates sink index distributions both pre-surgery and post- virtual-surgery for a patient with a sub-optimal outcome from a corresponding actual surgery, according to an embodiment;

[0016] Fig. 6 presents box plots for an example virtual surgery outcome metric, with individuals grouped according to actual surgical outcomes, according to an embodiment; and

[0017] Fig. 7 is a flow diagram for a method of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient, according to various embodiments.Description of the Examples

[0018] Reference will now be made in detail to example implementations, illustrated in the accompanying drawings. Wherever convenient, the same reference numbers will be used throughout the drawings to refer to the same or like parts. In the following description, reference is made to the accompanying drawings that form a part thereof, and in which is shown by way of illustration specific exemplary examples in which the invention may be practiced. These examples are described in sufficient detail to enable those skilled in the art to practice the invention and it is to be understood that other examples may be utilized and that changes may be made without departing from the scope of the invention. The following description is, therefore, merely exemplary.

[0019] Some embodiments provide a patient-specific dynamic network model (DNM) of the patient’s brain from minutes of inter-ictal (between seizure) brain imaging data, e.g., intracranial EEG (iEEG) data. Some embodiments construct a virtual environment where clinicians can lesion the EZ in silico by nullifying their signals, obtain a model representing the patient’s brain post-surgery, and then examine how the dynamic network properties change in response to the virtual resection. Some embodiments calculate a new brain imaging (e.g., iEEG) biomarker of the EZ pre- and post-virtual surgery and quantify diminishment of the biomarker, providing a likelihoodof the surgical plan succeeding. Embodiments may be used repeatedly to evaluate the likelihood of success for multiple surgical plans. Moreover, clinicians may conduct an actual surgical EZ recession on a patient based on an embodiment’s virtual EZ recission outcome prediction.

[0020] These and other features and advantages are shown and described herein in reference to the figures.

[0021] Figs. 1A and 1 B present a flow diagram for a virtual resection method 100 according to various embodiments. The method 100 may be used to quantitatively evaluate a probability of a successful outcome of EX recission surgery for a patient. Features, techniques, and other elements of the method 100 may be used in a method of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient as shown and described herein in reference to Fig. 7.

[0022] The method 100 includes obtaining a brain imaging signal 102 for the patient. As shown in Fig. 1A, and by way of non-limiting example, the brain imaging signal may be electroencephalogram (EEG) data. The EEG data may be intracranial EEG data (e.g., sEEG, ECoG) or scalp EEG data. Other suitable brain imaging data includes functional Magnetic Resonance Imaging (fMRI) Blood-Oxygen-Level- Dependent (BOLD) data and Magnetoencephalography (Mag) data.

[0023] The method 100 also includes fitting a dynamical network model of the patient’s brain to the patient’s brain imaging signal 102. By way of non-limiting example, the dynamical network model may be based on a series of matrices 104, denoted A=(A) / =i,...,n. Details of fitting such as dynamical network model are shown and described herein in reference to Fig. 2.

[0024] The method 100 further includes performing a virtual resection 108 of the patient’s brain using the dynamical network model. To do so, the method 100includes modifying the patient’s brain imaging signal 106, which may be generated by the dynamical network model using the matrices 104. Specifically, one or more identified brain resection locations, which may be designated according to their corresponding electrodes (e.g., as shown in Fig. 1A and by way of non-limiting example, R’11 , T’2, 09, etc.) are provided, e.g., by a clinician. The method 100 then constructs a synthetic brain imaging signal 110, where the synthetic brain imaging signal 110 corresponds to a resection at the identified brain resection location(s). For example, relative to the patient’s brain imaging signal 106, the synthetic brain imaging signal 110 is zeroed out at the channels that correspond to the identified brain resection region(s). By way of non-limiting example, the synthetic brain imaging signal 110 may be constructed incrementally as xt+1= Aix't, where / '=1 ,... , for n brain imaging channels and t = 1 , ... ,m for m timesteps, where x'tcorresponds to the patient’s brain imaging signal xt, except that the channel(s) that correspond to the resections location(s) are set to zero.

[0025] The method 100 further includes fitting a dynamical network model of the patient’s brain to the synthetic brain imaging signal 110. The dynamical network model here may utilize the same form as the dynamical network model that is fit to the patient’s brain imaging signal 102, except that instead of being fit to the patient’s brain imaging signal, it is fit to the synthetic brain imaging signal 110. Thus, by way of nonlimiting example, the dynamical network model may be based on a series of matrices 112, denoted A = ( )i=1 n. Details of fitting such as dynamical network model are shown and described herein in reference to Fig. 2.

[0026] The method 100 further includes calculating respective sink indices (SIDs) for the dynamical network model (e.g., based on A) that is fit to the patient’s brain imaging signal and, separately, for the dynamical network model (e.g., based on?1) that is fit to the synthetic brain imaging signal at 114. According to some embodiments, the calculation for the dynamical network model (e.g., based on A) that is fit to the patient’s brain imaging signal may be omitted. Details of calculating an SID are shown and described herein in reference to Fig. 3. Note that according to various embodiments, an SID may or may not be represented as a graphical heatmap. According to some embodiments, for example, an SID may be represented by a numerical matrix.

[0027] The method 100 further includes at 116 determining a measure of a statistical distribution of the SID(s) from 114. For example, the measure of statistical distribution may be for respective histograms of the SID(s). As shown in Fig. 1 B, and by way of non-limiting example, the measure of statistical distribution may be a p-value and / or a Jarque-Bera test. Other suitable measures include, by way of non-limiting examples: KS limiting form, KS Stephens modification, KS Marsaglia method, KS Lilliefors modification, Anderson-Darling test, Cramer-Von Mises test, Shapiro-Wilk test, Shapiro-Francia test, and D’Agostino & Pearson test.

[0028] The method 100 further includes at 118 plotting respective heatmaps and histograms of the SID(s) of 116. Note that according to various embodiments, one or both of the heatmap(s) and / or one or both of the histogram(s) may be omitted. The plots may be displayed to a user, e.g., a clinician.

[0029] By way of non-limiting examples, of for purposes of illustration, example SID heatmaps and histograms are shown in Fig. 1 B. In particular, Fig. 1 B shows a pre-virtual-surgery SID histogram 120 for a patient, a pre-virtual-surgery SID heatmap 122 for the patient, a post-virtual-surgery SID histogram 130 for the patient, and a post-virtual-surgery SID heatmap 132 for the patient. As shown in Fig. 1 B, the post- virtual-surgery SID histogram 130 has a distribution that is more bell-shaped than thatof the pre-virtual-surgery SID histogram 130. Further, the post-virtual-surgery SID heatmap 132 depicts fewer data at the extremes. Both of these observations are indicative of a successful surgery. Further explanation of possible interpretations of the SID heatmaps and histograms for the patient and according to various embodiments is provided in reference to Fig. 5.

[0030] Fig. 2 is a schematic diagram depicting a technique 200 for fitting a dynamic network model and calculating a sink index, according to various embodiments. The technique 200 may be used for a virtual resection method as shown and described herein in reference to Fig. 1 and for a method of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient as shown and described herein in reference to Fig. 7, for example.

[0031] The technique 200 includes brain imaging data acquisition 202. As shown in Fig. 2, and by way of non-limiting example, the brain imaging data may be iEEG data. Note however, embodiments are not limited to iEEG brain imaging data. Other suitable brain imaging data includes, for example, sEEG data, ECoG data, scalp EEG data, fMRI BOLD data, and Mag data. Note that, as shown and described herein in reference to Fig. 4, disclosed embodiments may be brain imaging modality independent, such that any of the aforementioned brain imaging data, or other brain imaging data, may be used.

[0032] Thus, the acquired brain imaging data may be in the form of EEG scalp imaging data 204. The brain imaging data may be represented as a multivariate time series (%i(t))i=in;t=im, where n represents the number of channels and m represents the number of timesteps. The data from each channel for a single time step t may be represented

[0001] The technique 200 includes fitting a dynamical network model, at 206.The overall dynamical network model includes a sequence of linear time invariant models of the form:(1 ) x(t + 1) = Atx(t)

[0002] In Equation (1 ), / = 1 , 2, ... , m, where there are m time windows, and Atdescribes how each channel influences each other dynamically within a particular time window of the iEEG. Each time window may be, by way of non-limiting example, 500 msec. Thus, for n channels and m time windows, the dynamical network model may include a sequence A of n x n matrices Atfor / = 1 , 2, ... , m.

[0003] The sink index is then derived from the linear time varying dynamical network model Ai)iim, estimated from the interictal brain imaging data. The term “source” refers to a group of brain regions that are actively influencing the electrical activity of other regions, while “sink” refers to a group of regions that are mostly being influenced by others’ activity.

[0004] In the context of EEG or iEEG data, source-sink connectivity considers how electrical activity propagates through the brain network, from the sources to the sinks. It is hypothesized that the epileptogenic zone in a patient is inhibited by other regions during non-clinical seizure periods and thus are sinks. See Gunnarsdottir KM, Li A, Smith RJ, et al. Source-sink connectivity: A novel interictal EEG marker for seizure localization. Published online November 19, 2021 :2021.10.15.464594. doi: 10.1101 / 2021 .10.15.464594. Gunnarsdottir et al. investigated this by creating an algorithm that identified two groups of nodes within an interictal iEEG network. These groups were nodes that continuously inhibit neighboring nodes (sources) and the nodes that are inhibited (sinks). Gunnarsdottir et al. estimated patient-specific dynamical network models from several minutes of interictal iEEG data, and theresulting connectivity properties as gleaned from the A matrices helped identify the top sources and sinks within the network. Specifically, Gunnarsdottir et al. quantified each node using source-sink metrics derived from the A matrices.

[0005] Toward deriving a sink index (SID) for the brain imaging data, each rr*m metric matrix is summarized into an r?x1 vector of coordinates 208, where n is the number of channels. In particular, the vector of coordinates 208 may include pairs, where each pair represents a row rank rn and a column rank cn, for / = 1 , ... , . The vectors of coordinates 208 are then converted into normalized sink indices, e.g., as shown and described herein in reference to Fig. 3. A SID heatmap 210 may be generated for each timestep, and the heat maps 210 for the timesteps may be joined to generate a SID heatmap 212 for the entire brain imaging data. Note that according to various embodiments, the SID heatmap need not be explicitly displayed; rather, according to various embodiments, the SID heatmap data itself may be used to predict an epileptogenic zone resection surgery outcome.

[0006] Fig. 3 is a schematic diagram detailing a technique 300 for calculating a sink index, according to various embodiments. The technique 300 may be used to generate a sink index (SID) from the vectors of coordinates obtained from brain imaging data as shown and described herein in reference to Fig. 2. The technique 300 may thus be used for a virtual resection method as shown and described herein in reference to Fig. 1 and for a method of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient as shown and described herein in reference to Fig. 7, for example.

[0007] The technique 300 includes obtaining vectors of coordinates representing brain imaging data, e.g., as shown and described herein in reference to Fig. 2. For each vector of coordinates 302, at 304, the technique 300 represents thevector of coordinates 302 in a two-dimensional space with respective axes for row rank (rr) and column rank (cr). The technique 300 then computes, for each vector of coordinates 302, a Euclidean distance to the coordinates of an ideal sink, which may be located at, for example, (1 , 1 / / V) in the two-dimensional space. This distance may be subtracted from the square root of two, producing a sink index quantity for the particular vector of coordinates. This process may be repeated for each timestep, and the heat maps 306 for the timesteps may be joined to generate a SID heatmap 308 for the entire brain imaging data. As noted above in reference to Fig. 2, for various embodiments, the SID heatmap need not be explicitly displayed, but instead may be used to predict an epileptogenic zone resection surgery outcome as described herein.

[0008] Fig. 4 illustrates distributions of source-sink indices for healthy individuals, according to various embodiments. In particular, Fig. 4 shows brain imaging data in the form of fMRI BOLD data 402 for a first healthy individual, as well as an SID histogram 404 and an SID heatmap 406 for the first individual. Fig. 4 also shows brain imaging data 414 in the form of scalp EEG data 412 for a second healthy individual, as well as an SID histogram 414 and an SID heatmap 416 for the second individual. Note that both histograms 404, 414 have a roughly bell-shaped distribution. According to various embodiments, a measure of SID distribution normality (e.g., a measure of how bell-shaped the SID distribution is) may be used to gauge whether the respective individual has (or will have) epilepsy. For example, the more bellshaped the SID histogram, the less likely that the individual has (or will have) epilepsy. Some embodiments quantify a measure SID statistical distribution normality, e.g., as described herein in reference to Fig. 1 , for use in predicting an epileptogenic zone resection surgery outcome for an epilepsy patient based on a virtual resection.

[0009] Fig. 4 also illustrates SID heatmaps 406, 416 for the first and second healthy individuals. Note that these heatmaps 406, 416 generally lack significant data at the extremes. Some embodiments may quantify this property as an indication of whether the respective individual has epilepsy.

[0010] Note that both SID histograms 404, 414 shown in Fig. 4 have generally bell-shaped distributions, which are indicative of a lack of epilepsy, even though the SID histograms 404, 414 originated from different types of brain imaging data. In particular, the SID histogram 404 for the first individual was generated from scalp EEG data, whereas the SID histogram 414 for the second individual was generated from fMRI BOLD data. Thus, Fig. 4 illustrates that embodiments are generally brain imaging modality independent. By way of non-limiting examples, various embodiments may utilize any of the following brain imaging data: EEG data (e.g., intracranial EEG data such as sEEG and ECoG, or scalp EEG data), fMRI BOLD data, or Mag data.

[0011] Fig. 5A illustrates sink index distributions 500 both pre-surgery and post- virtual-surgery for a patient with a successful outcome from a corresponding actual surgery, according to an embodiment. In particular, the data represented in Fig. 5A shows histograms 502, 512 and SID heatmaps 504, 514 for a patient that underwent an actual successful epileptogenic zone resection surgery. The actual surgery included ablating brain locations corresponding to intracranial EEG contacts L’1 , L’2, L’3, L’4, and L’5. After the actual surgery, the patient had an Engle score of 1 , which is considered indicative of a lack of epilepsy. The SID histogram 502 and the SID heatmap 504 were generated from actual pre-surgery brain imaging data for the patient. The SID histogram 512 and SID heatmap 514 were generated by an embodiment and depict results of a virtual epileptogenic resection surgery. The virtual resection surgery simulated ablation of the same brain locations that were ablated bythe actual surgery. The post-virtual-surgery SID histogram 512 shows more of a bell shape than the pre-surgery SID histogram 512, and the post-virtual-surgery SID heatmap 514 shows fewer data at the extremes. Both of these observations are indicative of a successful virtual surgery. Moreover, the actual surgery, which was subsequently simulated as a virtual surgery using an embodiment, was successful for the individual. Note that embodiments may quantify these indications, e.g., using a normality metric for SID histograms, and such numeric quantification may provide thresholds that may be used to predict an epileptogenic zone resection surgery outcome for an epilepsy patient.

[0012] Fig. 5B illustrates sink index distributions 550 both pre-surgery and post- virtual-surgery for a patient with a sub-optimal outcome from a corresponding actual surgery, according to an embodiment. In particular, the data represented in Fig. 5B shows histograms 552, 562 and SID heatmaps 554, 564 for a patient that underwent an actual epileptogenic zone resection surgery, with sub-optimal results. The actual surgery included ablating brain locations corresponding to intracranial EEG contacts L'4, G’1 -3, L'2-3, L’5-6. After the actual surgery, the patient had an Engle score of 2, which is considered sub-optimal, although an improvement relative to the patient’s pre-surgery Engle score. The SID histogram 552 and the SID heatmap 554 were generated from actual pre-surgery brain imaging data for the patient. The SID histogram 512 and SID heatmap 564 were generated by an embodiment and depict results of a virtual epileptogenic resection surgery. The virtual resection surgery simulated ablation of the same brain locations that were ablated by the actual surgery. The post-virtual-surgery SID histogram 562 shows slightly more of a bell-shape than the pre-surgery SID histogram 512, and the post-virtual-surgery SID heatmap 564 shows slightly fewer data at the extremes. Both of these observations are indicativeof an improvement from the virtual surgery, but not a complete absence of epileptic activity. Moreover, the actual surgery, which was subsequently simulated as a virtual surgery using an embodiment, produced sub-optimal actual results for the individual.

[0013] Fig. 6 presents box plots 600 for an example virtual surgery outcome metric, with individuals grouped according to actual surgical outcomes, according to an embodiment. Each dot represents an individual that underwent an actual epileptogenic zone resection surgery. From left to right, the box plots represent those individuals whose actual surgery resulted in Engle scores of 1 , 2, 3, and 4, respectively. For the y-axis, epileptogenic zone resection surgery was simulated according to an embodiment based on the same brain locations that were ablated in the actual surgery. The null hypothesis was that a successful surgery outcome will occur if the SID histogram distribution modulates into a normal distribution. Normality scores for the pre- and post-virtual-surgery SID histograms were calculated using the Jarque-Bera normality test. The surgery outcome likelihood metric was the p-value differences (Jarque-Bara normality before and after virtual surgery) of the SID histograms. The y-axis represents these differences. As shown in Fig. 6, the p-value difference distribution clearly servers as a classifier, with individuals having an Engle score of 1 after the actual surgery, grouped in the leftmost box plot, well above the remaining box plots along the y-axis.

[0014] Fig. 7 is a flow diagram for a method 700 of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient, according to various embodiments. The method 700 may be implemented consistent with the method 100 as shown and described in reference to Figs. 1A and 1 B and with the technique 200 as shown and described herein in reference to Fig. 2.

[0015] At 702, the method 700 includes receiving an identification of a brain location. One or multiple brain location identifications may be received at 702. The brain locations may be identified according to respective intracranial EEG contact designations, by way of non-limiting example. The brain location identifications may be received as input by a clinician using a computer interface on a tablet computer, by way of non-limiting example. The computer interface may be a graphical user interface, for example, which may depict an image of the patient’s brain or a representative generic brain. The brain image may include labels that show locations of various ablation candidate sites. By way of non-limiting example, the clinician may identify the brain locations per 702 by touching on a touch screen, clicking with a mouse, or otherwise providing a selection of brain locations on the brain image in the graphical user interface.

[0016] At 704, the method 700 includes obtaining a patient brain imaging signal, where the patient brain imaging signal represents interictal activity of the patient’s brain. Note that embodiments are independent of the particular brain imaging modality, as shown and described herein in reference to Fig. 4. Thus, the patient brain imaging signal may be any of a variety of forms, including, by way of non-limiting example, EEG data (e.g., intracranial EEG data such as sEEG and ECoG, or scalp EEG data), fMRI BOLD data, or Mag data. The brain imaging data may be obtained directly from an imaging device (e.g., an EEG machine, an fMRI machine, or a Mag machine) or from electronic storage of data previously obtained using such an imaging device.

[0017] At 706, the method 700 includes fitting a first dynamical network model to the patient brain imaging signal. The first dynamical network model may be selectedand fit as shown and described herein in reference to Figs. 1A, 1 B, 2, and / or 3, for example.

[0018] At 708, the method 700 includes calculating a first sink index from the second dynamical network model. The first sink index may be calculated as shown and described herein in reference to Figs. 1 A, 1 B, 2, and / or 3, for example.

[0019] At 710, the method 700 includes constructing, using the first dynamical network model, a synthetic brain imaging signal, where the synthetic brain imaging signal corresponds to a resection at the brain location. The construction of 708 may be performed as shown and described herein in reference to Fig. 1 , for example.

[0020] At 712, the method 700 includes fitting a second dynamical network model to the synthetic brain imaging signal. The second dynamical network model may be selected and fit as shown and described herein in reference to Figs. 1A, 1 B, 2, and / or 3, for example.

[0021] At 714, the method 700 includes calculating a second sink index from the second dynamical network model. The second sink index may be calculated as shown and described herein in reference to Figs. 1 A, 1 B, 2, and / or 3, for example.

[0022] At 716, the method 700 includes providing, based on the first and second sink indexes, an outcome prediction for resecting the patient’s brain at the brain location. The outcome prediction may be determined as shown and described herein in reference to Figs. 1A, 1 B, 4, 5A, and / or 5B, for example. The outcome prediction may be provided by display on a computer monitor, transmission over a computer network, or recordation in electronic memory, e.g., of a clinical information system, by way of non-limiting examples. The outcome prediction may be in the form of a quantitative value, e.g., a normality score (e.g., a Jarque-Bera test score, a KS limiting form score, a KS Stephens modification score, a KS Marsaglia method score, a KSLilliefors modification score, an Anderson-Darling test score, a Cramer-Von Mises test score, a Shapiro-Wilk test score, a Shapiro-Francia test score, or a D’Agostino & Pearson test score), a p-score, an Engle score, a difference of any of the preceding, or a different metric. In addition, or in the alternative, the outcome prediction may include a qualitative designation, such as “optimal,” “suboptimal,” or “unsuccessful.”

[0023] Thus, various embodiments provide a system and computer-readable medium for, and method of, predicting an epileptogenic zone resection surgery outcome for an epilepsy patient. As presented herein, e.g., in reference to Fig. 7, embodiments that utilize dynamical network models may be used to test surgical plans prior to surgery and may improve outcomes. Embodiments may utilize a sink index distribution as a brain imaging biomarker to perform in-silico resections to test and modify surgical plans prior to actual surgery.

[0024] Certain examples can be performed using a computer program or set of programs. The computer programs can exist in a variety of forms both active and inactive. For example, the computer programs can exist as software program (s) comprised of program instructions in source code, object code, executable code or other formats; firmware program(s), or hardware description language (HDL) files. Any of the above can be embodied on a transitory or non-transitory computer readable medium, which include storage devices and signals, in compressed or uncompressed form. Exemplary computer readable storage devices include conventional computer system RAM (random access memory), ROM (read-only memory), EPROM (erasable, programmable ROM), EEPROM (electrically erasable, programmable ROM), flash memory, and magnetic or optical disks or tapes.

[0025] Aspects of the present disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), andcomputer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented using computer readable program instructions that are executed by an electronic processor.

[0026] These computer readable program instructions may be provided to a processor of a general-purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the electronic processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the flowchart and / or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and / or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function / act specified in the flowchart and / or block diagram block or blocks.

[0027] In embodiments, the computer readable program instructions may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, statesetting data, configuration data for integrated circuitry, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++, or the like, and procedural programming languages, such as the C programming language or similar programming languages. The computer readable program instructions may executeentirely on a user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server.

[0028] As used herein, the terms “A or B” and “A and / or B” are intended to encompass A, B, or {A and B}. Further, the terms “A, B, or C” and “A, B, and / or C” are intended to encompass single items, pairs of items, or all items, that is, all of: A, B, C, {A and B}, {A and C}, {B and C}, and {A and B and C}. The term “or” as used herein means “and / or.”

[0029] As used herein, language such as “at least one of X, Y, and Z,” “at least one of X, Y, or Z,” “at least one or more of X, Y, and Z,” “at least one or more of X, Y, or Z,” “at least one or more of X, Y, and / or Z,” or “at least one of X, Y, and / or Z,” is intended to be inclusive of both a single item (e.g., just X, or just Y, or just Z) and multiple items (e.g., {X and Y}, {X and Z}, {Y and Z}, or {X, Y, and Z}). The phrase “at least one of” and similar phrases are not intended to convey a requirement that each possible item must be present, although each possible item may be present.

[0030] The techniques presented and claimed herein are referenced and applied to material objects and concrete examples of a practical nature that demonstrably improve the present technical field and, as such, are not abstract, intangible or purely theoretical. Further, if any claims appended to the end of this specification contain one or more elements designated as “means for [perform]ing [a function]...” or “step for [performing [a function]...”, it is intended that such elements are to be interpreted under 35 U.S.C. § 112(f). However, for any claims containing elements designated in any other manner, it is intended that such elements are not to be interpreted under 35 U.S.C. § 112(f).

[0031] While the invention has been described with reference to the exemplary examples thereof, those skilled in the art will be able to make various modifications to the described examples without departing from the true spirit and scope. The terms and descriptions used herein are set forth by way of illustration only and are not meant as limitations. In particular, although the method has been described by examples, the steps of the method can be performed in a different order than illustrated or simultaneously. Those skilled in the art will recognize that these and other variations are possible within the spirit and scope as defined in the following claims and their equivalents.

Claims

What is claimed is:1 . A method of predicting an epileptogenic zone resection surgery outcome for an epilepsy patient, the method comprising: receiving an identification of a brain location; obtaining a patient brain imaging signal, wherein the patient brain imaging signal represents interictal activity of the patient’s brain; fitting a first dynamical network model to the patient brain imaging signal; calculating a first sink index from the first dynamical network model; constructing, using the first dynamical network model, a synthetic brain imaging signal, wherein the synthetic brain imaging signal corresponds to a resection at the brain location; fitting a second dynamical network model to the synthetic brain imaging signal; calculating a second sink index from the second dynamical network model; and providing, based on the first sink index and the second sink index, an outcome prediction for resecting the patient’s brain at the brain location.

2. The method of claim 1 , further comprising: repeating, for a plurality of brain locations, the receiving, the constructing, the fitting the second dynamical model, the calculating, and the providing, whereby a plurality of outcome predictions are obtained; and surgically resecting the patient’s brain based on the plurality of outcome predictions.

3. The method of claim 1 , wherein the providing the outcome prediction for resecting the patient’s brain at the brain location comprises: determining a first measure of a first statistical distribution of the first sink index and a second measure of a second statistical distribution of the second sink index, wherein the outcome prediction is based on first measure and the second measure.

4. The method of claim 3, wherein the first statistical distribution of the first sink index comprises a histogram, wherein the first measure comprises a measure of normality, wherein the second statistical distribution of the second sink index comprises a histogram, and wherein second first measure comprises a measure of normality.

5. The method of claim 1 , wherein the brain imaging signal comprises an Electroencephalogram (EEG) signal of the patient’s brain.

6. The method of claim 1 , wherein the brain imaging signal comprises a functional Magnetic Resonance Imaging (fMRI) Blood-Oxygen-Level-Dependent (BOLD) signal of the patient’s brain.

7. The method of claim 1 , wherein the brain imaging signal comprises aMagnetoencephalography (MEG) signal of the patient’s brain.

8. The method of claim 1 , wherein the fitting the first dynamical network model to the patient brain imaging signal comprises determining a sequence of linear time-invariant dynamical network models.

9. The method of claim 8, wherein the constructing the synthetic brain imaging signal comprises: removing channel information from segments of the patient brain imaging signal, whereby a plurality of modified segments of the patient brain imaging signal are produced; and applying each of the sequence of linear time-invariant dynamical network models to a respective segment of the plurality of modified segments of the patient brain imaging signal.

10. The method of claim 1 , wherein the receiving the identification of the brain location comprises: displaying a representation of the patient’s brain on a screen; and obtaining an input representing the identification of the brain location.

11. A system for predicting an epileptogenic zone resection surgery outcome for an epilepsy patient, the system comprising a non-transitory computer readable medium comprising instructions, and at least one electronic processor that executes the instructions, to perform operations comprising: receiving an identification of a brain location; obtaining a patient brain imaging signal, wherein the patient brain imaging signal represents interictal activity of the patient’s brain;fitting a first dynamical network model to the patient brain imaging signal; calculating a first sink index from the first dynamical network model; and constructing, using the first dynamical network model, a synthetic brain imaging signal, wherein the synthetic brain imaging signal corresponds to a resection at the brain location; fitting a second dynamical network model to the synthetic brain imaging signal; calculating a second sink index from the second dynamical network model; providing, based on the first sink index and the second sink index, an outcome prediction for resecting the patient’s brain at the brain location.

12. The system of claim 11 , wherein the operations further comprise: repeating, for a plurality of brain locations, the receiving, the constructing, the fitting the second dynamical model, the calculating, and the providing, whereby a plurality of outcome predictions are obtained; and surgically resecting the patient’s brain based on the plurality of outcome predictions.

13. The system of claim 11 , wherein the providing the outcome prediction for resecting the patient’s brain at the brain location comprises: determining a first measure of a first statistical distribution of the first sink index and a second measure of a second statistical distribution of the second sink index, wherein the outcome prediction is based on first measure and the second measure.

14. The system of claim 13, wherein the first statistical distribution of the first sink index comprises a histogram, wherein the first measure comprises a measure of normality, wherein the second statistical distribution of the second sink index comprises a histogram, and wherein second first measure comprises a measure of normality.

15. The system of claim 11 , wherein the brain imaging signal comprises an Electroencephalogram (EEG) signal of the patient’s brain.

16. The system of claim 11 , wherein the brain imaging signal comprises a functional Magnetic Resonance Imaging (fMRI) Blood-Oxygen-Level-Dependent (BOLD) signal of the patient’s brain.

17. The system of claim 11 , wherein the brain imaging signal comprises a Magnetoencephalography (MEG) signal of the patient’s brain.

18. The system of claim 11 , wherein the fitting the first dynamical network model to the patient brain imaging signal comprises determining a sequence of linear time-invariant dynamical network models.

19. The system of claim 18, wherein the constructing the synthetic brain imaging signal comprises: removing channel information from segments of the patient brain imaging signal, whereby a plurality of modified segments of the patient brain imaging signal are produced; andapplying each of the sequence of linear time-invariant dynamical network models to a respective segment of the plurality of modified segments of the patient brain imaging signal.

20. The system of claim 1 , wherein the receiving the identification of the brain location comprises: displaying a representation of the patient’s brain on a screen; and obtaining an input representing the identification of the brain location.

Images