Method for optimizing planning and placement of probes in the brain via multi-modal 3D analysis of brain anatomy

Abstract

A method includes obtaining a first imaging scan and a second imaging scan of a single subject's brain. The first imaging scan is converted to a first data set and the second imaging scan is converted to a second data set. A sequence-adaptive multi-modal segmentation algorithm is applied to the first data set and the second data set. The sequence-adaptive multi-modal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled data set and a second labeled data set. The first labeled data set and the second labeled data set are automatically co-registered to one another to generate a transformation matrix based on the first labeled data set and the second labeled data set. The transformation matrix is applied to align the first data set and the second data set.

Description

[0001] Cross-referencing related applications

[0002] This application claims priority to U.S. Provisional Patent Application No. 62 / 978,868, filed February 20, 2020, entitled “Methods of Identification and Avoiding Visual Deficits After Laser Interstitial Thermal Therapy for MesialTemporal Lobe Epilepsy,” which is hereby incorporated herein by reference in its entirety. Background Technology

[0003] Intracranial electrodes are implanted in patients to record high spatiotemporal resolution intracranial electroencephalogram (icEEG) data and to modulate neural circuits and systems. Patients receiving these implants are most commonly assessed for neurological disorders such as epilepsy, movement disorders, and mental illnesses.

[0004] In some patients with medically refractory epilepsy, seizures can be localized to a definable lesion, and in this exemplary case, surgical intervention may offer the possibility of stopping further seizure activity by directly removing, excising, or destroying pathological brain tissue (including minimally invasive methods, including but not limited to catheter-based tissue ablation). Unfortunately, in many cases, patients do not have lesions that can be identified using only non-invasive studies or assessments, which may include studies of brain activity, including but not limited to electroencephalography (EEG) and magnetoencephalography (MEG), and anatomical imaging modalities for identifying structural lesions, such as magnetic resonance imaging (MRI) or computed tomography (CT). Electrodes are also placed for neuromodulation of the epilepsy—currently either in the anterior nucleus of the thalamus or at the site of the seizure.

[0005] Movement disorders (Parkinson's disease, dystonia, essential tremor) are also common. Treatment for these conditions is usually surgical because medications often cause adverse side effects. Deep basal ganglia nuclei (such as the subthalamic nucleus, the globus pallidus, and the VIM nucleus in the thalamus) and their associated white matter pathways are often targeted for these conditions.

[0006] When medication proves ineffective, mental illnesses are rapidly becoming targets of neuromodulation—these include treatment-resistant depression, obsessive-compulsive disorder, post-traumatic stress disorder, and eating disorders.

[0007] In these exemplary patients, implantation of subdural electrodes (SDE) and / or stereotactic electroencephalography (SEEG) electrodes and / or other probes, catheters, or recording devices is a common strategy for precisely defining the relationship between healthy and / or eloquent brain regions and pathological brain regions that may form the basis of a hypothetical pathological network, for diagnostic purposes or for stimulation to elicit neural modulation. Summary of the Invention

[0008] One approach includes: obtaining a first imaging scan and a second imaging scan of a single subject's brain. The first imaging scan is converted into a first dataset, and the second imaging scan is converted into a second dataset. A sequence-adaptive multimodal segmentation algorithm is applied to the first and second datasets. The sequence-adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled dataset and a second labeled dataset. The first and second labeled datasets are automatically co-registered to generate a transformation matrix based on the first and second labeled datasets. The transformation matrix is ​​applied to align the first and second datasets.

[0009] A non-transient computer-readable medium encoded with instructions executable by one or more processors to obtain a first imaging scan and a second imaging scan of a single subject's brain, converting the first imaging scan into a first dataset, and converting the second imaging scan into a second dataset. The instructions can also be executed by one or more processors to apply a sequence-adaptive multimodal segmentation algorithm to the first and second datasets, wherein the sequence-adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled dataset and a second labeled dataset. The instructions can further be executed by one or more processors to automatically co-register the first labeled dataset and the second labeled dataset with each other to generate a transformation matrix based on the first labeled dataset and the second labeled dataset. The instructions can also be further executed by one or more processors to apply the transformation matrix to align the first and second datasets.

[0010] A system includes one or more processors and a memory. The memory is coupled to the one or more processors and stores instructions. The instructions configure the one or more processors to: acquire a first imaging scan and a second imaging scan of a subject's brain, and convert the first imaging scan into a first dataset and the second imaging scan into a second dataset. The instructions may also configure the one or more processors to apply a sequence-adaptive multimodal segmentation algorithm to the first and second datasets, wherein the sequence-adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled dataset and a second labeled dataset. The instructions further configure the one or more processors to automatically co-register the first labeled dataset and the second labeled dataset with each other to generate a transformation matrix based on the first labeled dataset and the second labeled dataset. The instructions further configure the one or more processors to apply the transformation matrix to align the first and second datasets. Attached Figure Description

[0011] For a detailed description of the various examples, reference will now be made to the accompanying drawings, in which:

[0012] Figure 1 A flowchart of a method for co-registration in brain imaging scans according to this disclosure is shown.

[0013] Figure 2 A flowchart of a method for generating surface models of cortical and subcortical brain regions according to this disclosure is shown.

[0014] Figure 3 A flowchart of a method for automatically segmenting the cerebral vascular system according to the present disclosure is shown.

[0015] Figure 4 A flowchart of a method for visualizing the underlying brain structure according to this disclosure is shown.

[0016] Figure 5 A flowchart of a method for automatically planning electrode or probe implantation according to this disclosure is shown.

[0017] Figure 6 A flowchart of a method for automatically locating, naming, and visualizing previously implanted electrodes or brain-penetrating probes according to this disclosure is shown.

[0018] Figure 7 A graphical representation of the co-registration of different neuroimaging modalities performed on a single subject according to this disclosure is shown.

[0019] Figures 8A-8F A graphical representation of a generated 2D / 3D surface model depicting the hippocampus and thalamus according to this disclosure is shown.

[0020] Figures 9A-9DExample steps for segmenting the human cerebral vascular system according to this disclosure are shown.

[0021] Figure 10A-10W The diagram shows arbitrary angles intersecting with surface and volume models, using cutting planes to optimize visualizations of cortical and subcortical structural and / or functional representations, according to this disclosure.

[0022] Figure 11A-11R An example of population-derived anatomical targeting for electrode or penetrating probe implantation according to this disclosure is shown.

[0023] Figures 12A-12E A graphical representation of automatic electrode positioning and marking is shown.

[0024] Figure 13 A block diagram of a computational system suitable for implementing the methods disclosed herein is shown. Detailed Implementation

[0025] Throughout this specification and claims, certain terms are used to refer to specific system components, as understood by those skilled in the art, and different entities may use different names to refer to components. This document is not intended to distinguish between components with different names but identical functions. In this disclosure and claims, the terms "comprising" and "including" are used in an open-ended manner and should therefore be interpreted as "including but not limited to...". Furthermore, the term "coupled" is intended to indicate an indirect or direct wired or wireless connection. Thus, if a first device is coupled to a second device, the connection can be either a direct connection or an indirect connection via other devices and connections. The use of "based on" is intended to mean "at least partially based on". Thus, if X is based on Y, then X can be a function of Y and any number of other factors.

[0026] There are many clinical justifications for implanting electrodes or other medical devices in the brain, and focal epilepsy provides a single exemplary condition, which this article focuses on for illustrative purposes. In all such patients who meet the criteria for invasive monitoring or interventional surgery, the clinical process can be broadly divided into three phases: 1) planning; 2) data acquisition; and 3) intervention. Typically, in the planning phase, high-resolution anatomical imaging of the patient is performed using MRI, which can also be performed after the injection of contrast agents into the bloodstream to enhance vascular imaging. Recently, with advancements in computational modeling techniques, data from these imaging scans can be used to generate 2D and 3D models of the patient's brain anatomy, which can better inform surgeons and clinicians in planning subsequent surgical interventions, while taking into account key or important anatomical structures.

[0027] Following planning, the patient will undergo intracranial electrode implantation, followed by postoperative imaging (e.g., CT brain scan) to precisely and accurately correlate the implanted electrodes with the patient's cortical anatomy. Similarly, 2D and / or 3D computational models of the implanted electrodes or probes can be generated from repeated imaging scans and used to correlate the acquired electrophysiological data with the underlying cortical anatomy after co-registration (e.g., alignment to a common coordinate space) of different (pre- and post-implantation) imaging data.

[0028] During the intervention phase, data collected from Phases 1 and 2 (planning and implantation) are used as part of a comprehensive clinical assessment of the patient to determine if the presumed pathological lesion can be located, and if so, how it might relate to critical and non-critical brain structures. Surgeons use this information to optimize the final surgical plan for removing the epileptic focus, minimizing damage to healthy or important brain regions.

[0029] Precision and accuracy are critical at each of these stages to ensure that patients do not suffer any transient or permanent adverse neurological outcomes that could otherwise be avoided. Despite numerous technological, imaging, and computational advances over the past 20 years, significant technical hurdles remain. These include challenges such as: accurate co-registration across different imaging modalities; automated planning of electrode implantation; methods for minimizing the risk of damage to critical structures (e.g., blood vessels) using only non-invasive imaging data; and automated and / or semi-automated integration of post-implantation imaging data with neuroanatomical and functional data to inform potential surgical interventions. The methods and systems disclosed herein overcome these limitations using novel approaches described below.

[0030] Embodiments of this disclosure relate to robust and accurate co-registration of different brain imaging modalities. Specifically, this disclosure describes novel applications of sequence-adaptive multimodal segmentation algorithms for acquiring the same and / or different imaging modalities of brain imaging to achieve accurate intra-subject multimodal co-registration of these imaging modalities in a robust and automated manner. The sequence-adaptive multimodal segmentation algorithm is applied to a 3D dataset generated from raw brain imaging scans performed on a patient. This 3D dataset may include, but is not limited to, data formats as described by the Neuroimaging Informatics Technology Initiative (NIFTI), and is generated from imaging files of a patient's brain scan. An exemplary embodiment of this 3D dataset may be images stored according to standards defined by the Medical Digital Imaging and Communications (DICOM) standard. A sequence-adaptive multimodal segmentation algorithm is applied to the dataset without further or additional preprocessing (including but not limited to intensity normalization and / or volume conformation) to generate a new labeled segmentation dataset with the same volume and geometry as the original dataset, but in which each voxel (i.e., 3D pixel) has had its original intensity value replaced with a number relating to a unique anatomical region and / or the probability that the voxel belongs to a particular brain region (an exemplary embodiment may include a probability distribution of brain regions defined by a template or reference atlas). The segmentation dataset is then used as a “moving” input to a co-registration algorithm to align with an equivalent segmentation “target” dataset generated from the same and / or different imaging modalities from the same patient. The co-registration algorithm also generates a transformation matrix as part of the computational output, which describes the mathematical operations required to accurately replicate the alignment between the input “moving” dataset and the “target” dataset in a symmetric manner, meaning both forward (i.e., aligning the “moving” dataset to the “target” dataset) and backward (i.e., aligning the “target” dataset to the “forward” dataset). Once the transformation matrix is ​​generated, it can be applied to any dataset sharing the volumetric geometry of the original moving dataset to align it with the original target dataset. This disclosure describes a novel application of the segmentation algorithm to co-register imaging datasets acquired from the same subject using the same or different imaging modalities (e.g., MRI and CT) based on the transformation matrix generated using segmentation. It provides a significant advancement over existing techniques for multimodal co-registration of brain images within a subject. Despite the presence of imaging features that may be common causes of failure in other current co-registration methods, the implementation of this disclosure yields satisfactory results. These common causes include, but are not limited to, anatomical defects, lesions and / or masses, foreign bodies, hemorrhage, and / or intensity differences between imaging datasets and / or differences in the imaging pulse sequences used and / or the scanner platform parameters used to acquire the images.

[0031] Embodiments of this disclosure relate to methods for generating anatomically accurate skin and skull models using extracranial boundary layers generated by a sequence-adaptive multimodal segmentation algorithm, thereby facilitating preoperative planning and post-implantation localization of sEEG electrodes. Segmenting extracranial boundary elements directly from CT imaging is a novel approach that improves upon existing methods. Generating skull boundary layers using methods applied to T1-weighted MRI datasets is also a novel approach in this field, demonstrating improvements over current boundary element modeling methods used to estimate these layers. These improvements provide tangible benefits for the planning of surgical electrode implantation by providing skull-brain and skin-skull boundary layers for electrode implantation.

[0032] Embodiments of this disclosure relate to methods that use cerebrospinal fluid (CSF) volume and segmentation of gray and white matter cortical layers as a mask to aid in segmenting blood vessels from 3D brain imaging datasets, including but not limited to contrast-weighted T1 MR imaging. In some exemplary embodiments, the MR imaging dataset may have its intensity values ​​upscaled to better separate high-intensity voxels that may reflect blood vessels from surrounding tissues (e.g., white matter bundles or from partial volume averaging) with similar (but lower) intensity values. Using a CSF boundary layer to mask and constrain the parameter space used for vascular segmentation using upscaled contrast MRI is a novel approach to facilitate the segmentation of the cerebral vascular system.

[0033] Embodiments of this disclosure relate to a method in which, in some exemplary embodiments, a multi-scale Hessian-based filter is applied to segment blood vessels from the aforementioned CSF-, gray-, and white-masked datasets.

[0034] Embodiments of this disclosure relate to methods for generating 2D / 3D anatomical mesh models of a patient's hippocampus, amygdala, and other subcortical structures, and for generating standardized versions of these anatomical meshes derived from the same high-resolution template volume.

[0035] Embodiments of this disclosure relate to a method that uses a loss function and / or risk metric defined with respect to segmented blood vessel volume to determine the optimal trajectory for implanting electrodes or penetrating probes into the brain.

[0036] Embodiments of this disclosure relate to methods for 2D / 3D visualization of the cerebral vascular system. In some exemplary embodiments, the aforementioned visualization will require reconstructing discontinuities in the vascular volume. In some exemplary embodiments, this visualization is achieved using algorithms originally developed for diffusion tensor imaging. In such exemplary embodiments, digital image intensity modulation is applied to the vascular volume, constrained by a specific directional gradient in 3D space, which in some embodiments is performed using Hessian- or tensor-based decomposition to simulate anisotropic diffusion in the 3D imaging volume. These datasets can then be processed using a diffusion tensor imaging toolkit to model and predict connections between similar but discontinuous imaging features. For example, in this way, possible connections between similar voxels that become discontinuous due to low signal-to-noise ratios or processing artifacts (e.g., for vessels in a 3D imaging volume) can be remodeled and visualized. For example, in this way, continuous vessels can be reconstructed from discontinuous datasets.

[0037] Embodiments of this disclosure relate to methods for generating topologically accurate 2D / 3D surface-based representations and / or anatomical mesh models of the hippocampus, amygdala, thalamus, basal ganglia, and other subcortical structures for surgical planning. Parcellations of cortical regions are also generated, and any of these structures can be visualized and manipulated independently relative to adjacent and contralateral structures. Furthermore, standardized surface models of these subcortical and / or other deep structures can be generated using population-based atlases and templates, enabling the translation and application of surface-based co-registration and analysis techniques originally developed for cortical-based intersubjective analyses and demonstrating significant improvements in accuracy and outcomes. These methods represent a novel contribution to the field, providing significant improvements in modeling the pathology of the hippocampus and / or other subcortical regions or in understanding the pathology of the hippocampus / amygdala or other deep brain structures through direct visualization or representation of functional or electrocardiographic data on mesh surface models, or for modeling the implantation of electrodes, penetrating probes, or catheters into these regions.

[0038] Embodiments of this disclosure relate to a method of aligning a trajectory obtained from a robotic, mechanical, or human implanted device or procedure with anatomical T1MR images of a subject using volumetric geometry and transformation matrices obtained from contrast-weighted T1MR imaging of the subject.

[0039] Embodiments of this disclosure relate to a method that uses automatic partitioning techniques and linear and / or nonlinear deformation algorithms to precisely identify anatomical targets to be implanted or targeted, so as to warp the subject’s brain to a standard template space in which the target is predefined based on anatomical structures, or vice versa.

[0040] Embodiments of this disclosure relate to a method for accurately identifying anatomical targets to be implanted or targeted using a prior probability distribution derived from a previously implanted population.

[0041] Embodiments of the present invention relate to the allocation of anatomical targets to inspire unsupervised design of implantation trajectories using depth electrodes placed to identify epilepsy indicated by specific symbolic features or characteristics of the epilepsy. This can be applied to the placement of laser probes, brain electrodes for recording or modulation, stereotactic biopsy probes, and the injection of biological, cellular, genetic, or chemical materials into the brain via trajectories using anatomical constraints and prior implantation cases.

[0042] Embodiments of this disclosure relate to a method for generating a 3D surface model predicting the ablation volume (i.e., the expected volume of the hippocampus to be affected) using a prior probability distribution derived from previous laser ablation volumes across the entire population. This is a novel contribution to the field and will improve preoperative planning and informed trajectory modeling for subsequent laser ablation or other similar catheter-based treatments.

[0043] Embodiments of this disclosure relate to automated techniques for identifying white matter pathways involved in critical functions such as motor, sensory, auditory, or visual processes (identified via deterministic or probabilistic tractography derived from diffusion imaging) and for avoiding damage to these white matter pathways.

[0044] Embodiments of this disclosure relate to a method for automatically segmenting and locating sEEG electrodes using intensity-upscaling of post-implantation CT brain imaging datasets and a volume-based clustering algorithm for resulting metal artifacts. Linear regression modeling is used to fit trajectories from the robotic implantation device that were previously aligned with the same imaging space to facilitate the identification of metal electrode artifacts from noise. The line-fitting model enables the automated method to account for deviations in electrode placement. Artifact clusters identified using a 3D volume clustering search algorithm are iteratively searched, while masking any cortical regions not within the currently interested cluster to ensure that overlapping or merged artifacts can be resolved. Using information about the parallelism between the trajectory path and the trajectory of the identified clusters, and information about the centroid of the clusters, the identified clusters can be aligned with the robotic trajectory so that the trajectory is refitted once a sufficient number of electrodes have been identified. Real-time visualization of the search algorithm allows for concurrent updates of cluster search results for informational and debugging purposes.

[0045] Embodiments of this disclosure relate to a method for validating the implantation planning of multiple stereotactic depth probes along an inclined trajectory, aided by simultaneously visualizing a cortical mesh model of the surface topology and deep anatomical structures revealed by structural magnetic resonance imaging along planar slices collinear with the proposed trajectory. Slicing the brain surface on any given plane enables visualization of the deeply embedded cortex in 3D and also allows clinicians to quickly confirm surgical plans.

[0046] Embodiments of this disclosure relate to methods for manipulating 3D spatial geometry to selectively visualize or interact with different surface models (e.g., skin, skull, blood vessels, brain, hippocampus, amygdala, etc.) along any plane. Surfaces can be selectively traversed along any plane. Similarly, visualizations of partitions or intracranial EEG activity can be presented along any plane, either along the visualized surface or deep into the visualized surface.

[0047] Embodiments of this disclosure relate to a method for converting intracranial electroencephalogram (icEEG) activity from a surface-based representation to a DICOM or 3D dataset format that depicts user-defined activations of interest constrained by an underlying cortical band that reflects the gray matter region likely responsible for the activation.

[0048] Embodiments of this disclosure relate to a method for performing empirical source localization estimation to model a unique neural generator from recorded icEEG data.

[0049] Using the method disclosed herein, template matching search is employed to automatically resolve implanted structures along a known planned trajectory in postoperative imaging. This provides clinicians with the final location of all implanted materials in an automated manner (e.g., without requiring manual identification of each electrode) and enables rigorous measurement of surgical precision. In practical applications, the method disclosed herein helps avoid visual disturbances following laser interstitial thermal therapy for medial temporal lobe epilepsy.

[0050] Figure 1 A flowchart of method 100 is shown, which is used for co-registration of brain imaging scans obtained for a subject through different imaging modalities. Although depicted sequentially for convenience, at least some of the actions shown can be performed in a different order and / or in parallel. Additionally, some implementations may perform only some of the actions shown. The operation of method 100 can be performed by a computational system as disclosed herein.

[0051] In box 102, one or more imaging scans of the subject's brain are obtained. The imaging scans can be performed using scanning modalities such as magnetic resonance imaging (MRI) sequences, computed tomography (CT) sequences, magnetoencephalography (MEG), positron emission tomography (PET), or any combination thereof.

[0052] In box 104, the imaging scan is converted into a file format / dataset that can be used to store, analyze, and manipulate the brain imaging data contained within the imaging scan. For example, the imaging scan can be converted to NIFTI format.

[0053] In box 106, each dataset generated in box 104 is fed as input to a sequence-adaptive multimodal segmentation algorithm to generate labeled partitions and segmented datasets. Typically, the segmentation algorithm performs this by aligning a probability map with the patient dataset. In one exemplary embodiment, given the intensity values ​​of voxels, the map helps assign the probability of a particular brain region label to any voxel in the dataset. In one exemplary embodiment, this can be achieved using Bayesian analysis, where the map provides prior probabilities of voxels belonging to a particular tissue category. As an example, the algorithm can then leverage a likelihood distribution to define a relationship between a given brain region label and the distribution of intensity values ​​in the voxels of that dataset. The term tissue category can refer to white matter, gray matter, cerebrospinal fluid, brain tumors, and / or other brain regions. The term alignment, as used herein, can refer to linear or nonlinear methods. Examples of the use of segmentation algorithms for sequence adaptive segmentation of brain MRI can be found, for example: Puonti O., Iglesias JE, Van Leemput K. (2013) Fast, Sequence Adaptive Parcellation of Brain MR Using Parametric Models, in: Mori K., Sakuma I., Sato Y., Barillot C., Navab N. (eds. Medical Image Computing and Computer-Assisted Intervention – MICCAI 2013, MICCAI 2013, Lecture Notes in Computer Science, Vol. 8149, Springer, Berlin, Heidelberg, https: / / doi.org / 10.1007 / 978-3-642-40811-3_91. The application of sequencing algorithms for co-registration across different imaging modalities is unknown in the art (e.g., between MRI and CT) and is a novel application here. Notably, the application of this algorithm to achieve fast, accurate, and robust co-registration between different imaging modalities (an exemplary embodiment of which could be co-registration of MRI and CT imaging scans) represents a significant improvement over existing technologies. In the labeled dataset, imaging data (example data units of imaging data are 3D pixels with a resolution of 1 mm x 1 mm x 1 mm (referred to herein as voxels)) are replaced with digital labels assigned according to the extracranial and intracranial structures represented. In various embodiments, voxels can have any size that the user deems useful.The term parcellation is used to refer to the cortical region of the marker, while the term segmentation refers to the subcortical region of the marker. In this disclosure, these two terms are used interchangeably to refer to both the cortical and subcortical regions of any marker.

[0054] In box 108, the labeled dataset is input into a co-registration algorithm, where any two datasets are aligned to each other's coordinate space, generating a mathematical transformation matrix that allows the coordinate transformation to be applied to any other dataset sharing the volumetric geometry of either of the two input datasets, thereby aligning that other dataset to the coordinate space of the second input dataset. In an exemplary embodiment, the transformation matrix may be a 4x4 matrix M that defines a linear, rigid, and / or affine transformation from point p to another point p', as defined by the equation p' = Mp. In this example, point p is defined using a column vector containing the x, y, z coordinates of a voxel and the number 1 to define the position of a voxel in the dataset. For example, p = (x, y, z, 1). The matrix-vector product multiplies the column vector from matrix M with the corresponding (x, y, z, 1) values ​​from column vector p. The summation of the scalar vector products generates the output vector p': (x', y', z', 1). In such an example, the upper 3x4 elements of matrix M can contain real numbers to store combinations of translations, rotations, scaling, and / or shearing (and other operations) applied to p. The last row in this example is (0 0 0 1). This form of transformation matrix can be generated using one or more of a variety of neuroimaging analysis software. Using a labeled dataset with partitions / segments as input for co-registration, especially in the case of CT imaging, overcomes many limitations of existing techniques (e.g., related to intensity scaling differences or tissue boundary differences) and generates reliable and accurate co-registration even in the presence of anatomical defects (e.g., stroke), brain injuries (e.g., tumors), or other imaging artifacts.

[0055] In box 110, the datasets generated in box 104 are aligned with each other using transformation matrices.

[0056] Figure 2 A flowchart of a method 200 for generating surface models of cortical and subcortical brain regions according to this disclosure is shown. Although depicted sequentially for convenience, at least some of the actions shown can be performed in a different order and / or in parallel. Additionally, some implementations may perform only some of the actions shown. The operations of method 200 can be performed on the labeled dataset generated by method 100 by a computational system as disclosed herein.

[0057] The surface model generated by method 200 includes a normalized mesh model derived from a population-level map, which allows for point-to-point correspondence between any point on the surface of one subject and the same point on the surface of another subject.

[0058] In box 202, all voxels that match the labeled values ​​of the cortical or subcortical region of interest are extracted into a new 3D dataset containing only those voxels of interest. For example, the new dataset could include the right hippocampus identified during the segmentation process of method 100.

[0059] In box 204, a standard volume-to-surface transformation is used to convert the segmented 3D dataset of the cortical or subcortical region of interest formed in box 202 into a surface mesh model. Generally, the standard volume-to-surface transformation can be implemented using existing open-source neuroimaging software. General and exemplary embodiments of such methods include volume binarization, followed by surface tessellation using the binarized values ​​to generate a mesh (e.g., as provided by Freesurfer: https: / / surfer.nmr.mgh.harvard.edu / fswiki / mri_tessellate) or a marching cube algorithm provided by the open-source vascular modeling toolkit software (VMTK: http: / / www.vmtk.org / vmtkscripts / vmtkmarchingcubes.html). In one exemplary embodiment, the surface / anatomical mesh model can be defined as points in 3D space connected by lines to form triangles, each triangle having 2D faces and combined according to specific volume geometry to form a topologically accurate representation of the modeled object. In one exemplary embodiment, the surface / anatomical mesh model may be a 2D model (e.g., a plane) that can then be folded in 3D space to depict a 3D object (e.g., a brain surface).

[0060] In box 206, the curvature and sulcal features of the resulting surface model are calculated. These features are then used to non-linearly align an expanded version of the surface model to match a high-resolution atlas of the same region generated from labeled population data. In this case, the high-resolution atlas may refer to a reference template pattern of cortical curvature data, which in some exemplary embodiments is previously prepared as an average pattern calculated from a representative set of subjects and made available as part of a standard repository. In other exemplary embodiments, such template data may be generated using a selected population of subjects (e.g., patients at a specific institution operated in a certain way).

[0061] In box 208, a normalized surface mesh, already aligned with the atlas, is overlaid on the subject's original surface model, and the coordinates of this normalized surface mesh are replaced by resampling of the surrounding coordinates in the subject's native coordinate space. The surface mesh consists of thousands of points in space called nodes, which are connected by lines to form triangles, the faces of which form the mesh of the surface model. In the case of a normalized surface, the mesh model contains a fixed number of nodes and maintains a specific node-to-atlas region correspondence (i.e., each node corresponds to the same region in the atlas), and this correspondence can be preserved across subjects. To preserve the correspondence, for each subject, both the normalized surface mesh and the subject's own original surface mesh are deformed non-linearly to align with a spherical template mesh derived from the aforementioned high-resolution population atlas. Both the subject and the normalized mesh are warped to maximize the overlap between brain sulci and curvature patterns. Once both the subject and the normalized surface mesh are aligned to the template atlas and thus aligned with each other, the nodes of the normalized surface mesh are assigned the average of the coordinates of a subset of the surrounding nodes of the subject's original surface mesh (an exemplary embodiment of which may be the four nearest nodes). In this way, the normalized surface is distorted to the subject's anatomical coordinate space, while both are aligned to the template, thus preserving a one-to-one correspondence between the normalized surface and the template atlas. Once deflated from the spherical configuration used during co-registration, the normalized surface mesh reveals the topology of the subject's own anatomy, while continuing to maintain the one-to-one correspondence between its nodes and anatomical atlas identities. In this manner, surface-based comparisons can be performed across subjects with high accuracy by comparing specific nodes with identical nodes across surfaces.

[0062] The operation of method 200 can be repeated for the contralateral hemisphere region, and the operation of method 200 can be repeated for any other additional cortical or subcortical or other marked / segmented or partitioned brain surface to be generated. For information on generating normalized surfaces for cortical surfaces, see, for example, Saad, ZS, Reynolds, RC, 2012. Suma, NeuroImage 62, 768–773, http: / / dx.doi.org / 10.1016 / j.neuroimage.2011.09.016; Kadipasaolu CM, Baboyan VG, Conner CR, Chen G, Saad ZS, Tandon N, Surface-based mixed effects multilevel analysis of grouped human electrocorticography, Neuroimage 1 November 2014; 101:215-24. doi:10.1016 / j.neuroimage.2014.07.006. Epub 12 July 2014, PMID:25019677). However, no such method is known for generating standardized surface-based meshes for subcortical regions. Exemplary examples of such regions may include the hippocampus, amygdala, thalamic nuclei, and basal ganglia. This surface model could be used to create standardized subcortical surfaces for individual anatomy, thereby achieving consistency of these subcortical structures across individuals in an unprecedented manner.

[0063] Method 200 is applied to brain regions (e.g., the hippocampus and / or amygdala and / or subcortical regions) that are not typically included in surface-based modeling or analysis, and is combined with high-resolution anatomical atlases of the regions to enable the generation of standardized surface meshes. This is a significant improvement over existing techniques that previously limited this approach to cortical regions (e.g., strictly speaking gray or white matter surfaces).

[0064] Figure 3 A flowchart of method 200 for automatically segmenting the cerebral vascular system and generating a model based on 2D / 3D surfaces and volumes, according to this disclosure, is shown. Although depicted sequentially for convenience, at least some of the actions shown can be performed in a different order and / or in parallel. Additionally, some implementations may perform only some of the actions shown. The operations of method 200 can be performed by a computing system as disclosed herein.

[0065] In box 302, one or more imaging scans of the subject's brain are obtained. The imaging scans include contrast-weighted MRI scans (e.g., T1-weighted MRI with contrast, which will be referred to as the contrast MRI dataset).

[0066] In box 304, the imaging scan is converted from the original imaging storage format (e.g., DICOM) to a 3D dataset according to method 100. During the conversion, the intensity values ​​of the contrast MRI are variablely amplified (e.g., 100x) to facilitate the differentiation of contrast-enhancing structures (e.g., blood vessels) from their surrounding structures.

[0067] In box 306, according to method 100, a labeled dataset is generated from contrastive MRI using a sequence adaptive multimodal segmentation algorithm.

[0068] In box 308, a mask from a labeled dataset generated as described in Method 100 for comparison with the MRI dataset is used to sub-select all voxels identified as belonging to cerebrospinal fluid (CSF) regions. The CSF mask (sub-selected voxels) provides a new improvement to the vessel segmentation algorithm because high-intensity voxels representing vessels are most frequently located in the CSF, adjacent to the pia mater surface. This is especially true for vessels considered to have the greatest clinically significant bleeding risk (typically vessels with a diameter ≥1.5 mm). Similar masks are also generated for regions labeled with both gray and white matter.

[0069] In box 310, a multi-scale filtering algorithm designed to enhance tubular features in imaging data is used to extract vessels from neighboring voxels that reflect CSF or background noise. Typically, the filtering algorithm utilizes Hessian-based eigenvalue decomposition to derive feature values ​​and vectors at different spatial scales at each pixel in the dataset to select tubular structures corresponding to vessels of different diameters (see, for example: Frangi, Alejandro F. et al., Multiscale vessel enhancement filtering, Medical Image Computing and Computer-Assisted Intervention—MICCAI'98, Springer Berlin Heidelberg, 1998, 130-137). The software algorithm returns an output that assigns a "vessel" weight to each voxel, ranging, for example, from 0 to 1, where a higher weight indicates a voxel with more vessel-like features (e.g., tubular).

[0070] In box 312, information from the vascular weighted dataset is integrated with the contrastive MRI dataset to weight the intensity values ​​of non-zero voxels by the relative “vascular” weights of the non-zero voxels and penalize those voxels that overlap with white or gray matter.

[0071] In box 314, the vascular dataset is aligned with the anatomical MRI dataset using the transformation matrix generated in box 306 (as described in method 100).

[0072] In box 316, the vascular dataset is converted into a surface anatomy mesh model (as described in box 204 for method 200), which can be visualized using different levels of transparency and lighting.

[0073] Figure 4 A flowchart of a method 400 for visualizing underlying brain structures according to this disclosure is shown. Although depicted sequentially for convenience, at least some of the actions shown can be performed in a different order and / or in parallel. Additionally, some implementations may perform only some of the actions shown. The operations of method 400 can be performed by a computational system as disclosed herein.

[0074] In box 402, a cutting plane in 3D space (e.g., a 2D cutting plane) intersects a 3D volume or surface on any user-defined axis. At the intersection of the cutting plane and a given surface mesh model or 3D volume, all components of the mesh on either side of the cutting plane can be selectively rendered as visible, invisible, or translucent. Any surface and / or volumetric (structural or functional) data can be visualized simultaneously on this plane. Typically, the operation of box 402 is performed by defining the cutting plane along one and / or more 2D anatomical imaging planes of the imaging dataset (exemplary embodiments of which include coronal, sagittal, or axial planes of MRI and / or CT scans). The intersection of the cutting plane with the imaging dataset can be defined along any arbitrary geometry (exemplary embodiments of which can include orthogonal and / or oblique angles), and voxels of the associated 3D volumetric imaging dataset are identified along the points where the two planes intersect, which will be used for further display and / or analysis. These voxels can then be selectively visualized alongside a 3D surface model of the subject. Furthermore, the relationship between the voxel coordinates and the coordinates of the 3D surface model (e.g., at their intersection points) can be used to selectively make components of the surface model visible, invisible, or translucent on either side of the cutting plane, or along the cutting plane itself. The coordinates of the voxels along the cutting plane can also be used to determine their distance from the implanted electrodes within the subject's body, which can then be used to calculate and generate a surface and / or volumetric data representation associated with the cutting plane.

[0075] In box 402, when applied to the extracranial and intracranial anatomical meshes defined in methods 200 and / or 300, qualitative and quantitative analysis of surfaces or volumes intersecting the cutting plane can be calculated, including but not limited to calculating morphological features such as the curvature, thickness, and area of ​​cortical gray / white matter and / or subcortical structures, as well as the curvature, thickness, and area of ​​the hippocampus and amygdala. Furthermore, the edges of the intersecting surfaces and / or volumes can be selectively raised or lowered to improve visualization accuracy. In one exemplary embodiment, the intersection of the cutting plane with specific elements of a 3D surface can include various cortical and / or subcortical surface layers, exemplary embodiments of which include pia mater and / or white matter surfaces. For these exemplary embodiments, the intersection of the cutting plane with these surfaces will determine the intermediate gray matter between these two exemplary surfaces. The intermediate gray matter may also be referred to as the cortical band. And the thickness of the cortical band can be calculated by calculating the distance between the pia mater and white matter surfaces along the point where they intersect the cutting plane (e.g., the orthogonal distance between the two surfaces at their intersection with the cutting plane). The area can be calculated by integrating the thickness across the length of the cortical band. In another exemplary embodiment, the curvature of a surface (e.g., a pia mater surface mesh) can be calculated by drawing orthogonal lines outward from the faces of surface mesh triangles to determine whether these lines intersect with the faces of another mesh triangle. Such intersection occurs when the faces of two triangles point towards each other, as is the case with brain sulci. Using the angles and distances of such intersections, local topological features, such as surface curvature and brain sulcus boundaries, can then be determined.

[0076] In box 404, when applied to extracranial and intracranial anatomical meshes as defined in methods 200 and / or 300, visualization of brain structural data (including one or more of MRI, CT, PET, fMRI, DTI) and / or brain activity data (including one or more of EEG or MEG or brain stimulation) can be selectively depicted relative to the cutting plane of various anatomical mesh models.

[0077] In box 406, when applied to the extracranial and intracranial anatomical networks defined in methods 200 and / or 300, users can perform virtual cuts of arbitrary shapes or geometries (e.g., dome-shaped surfaces or surfaces matching craniotomy) by selecting the path along the surface and the depth of the cut applied within the surface. In this way, users can model and visualize surgical approaches or various anatomical boundaries for clinical evaluation and / or surgical planning or training and / or educational visualization.

[0078] Figure 5A flowchart of a method 500 for automatically planning electrode or probe implantation according to this disclosure is shown. Although depicted sequentially for convenience, at least some of the actions shown can be performed in a different order and / or in parallel. Additionally, some implementations may perform only some of the actions shown. The operation of method 500 can be performed by a computing system as disclosed herein.

[0079] Method 500 uses a prior probability distribution derived from a population of previously implanted anatomical targets to be implanted or targeted to provide precise and automated planning of electrode or penetrating probe implantation trajectories. The prior probability distribution is generated using the entry and / or target coordinates of trajectories from a population of previously implanted trajectories, which have been aligned to the subject's brain (or vice versa) using linear or nonlinear deformation algorithms. Furthermore, the usual implantation strategy is derived from clinical considerations of possible anatomical regions using a description of clinical electrical syndromes associated with the subject's seizure symptomology or other electroclinical characteristics of epilepsy. In each case, the trajectory will have an additional objective: avoiding critical structures (e.g., blood vessels) using a loss function and risk metric. Additionally, in one embodiment, these trajectories can be created individually by a physician or surgeon in the field of stereotactic medicine by defining entry and target points of interest.

[0080] In box 502, one or more imaging scans of the subject's brain are obtained. The imaging scans include target anatomical MRI scans (e.g., T1-weighted MRI without contrast) and contrast-weighted scans (e.g., T1-weighted MRI with contrast).

[0081] In box 504, the imaging scans are converted from the original imaging storage format (e.g., DICOM) into datasets according to method 100. T1-weighted MRI without contrast is converted into a dataset called the Anatomical MRI dataset, and T1-weighted MRI with contrast is converted into a dataset called the Contrast MRI dataset.

[0082] In box 506, according to method 100, the vascular dataset and mesh model are generated and jointly registered to the anatomical MRI dataset and its associated mesh model.

[0083] In box 508, the predicted entry and target point coordinates in the current subject's own anatomical space are defined using target and entry point coordinates curated from a previously implanted cohort of subjects. Coordinates from the cohort have previously been co-registered to a high-resolution template atlas. In one exemplary embodiment, the template coordinate space may be a coordinate space defined in a standard coordinate space (e.g., Talairach space; Montreal Institute of Neuroscience space). Co-registration can be implemented using nonlinear or linear / rigid / affine transformations and computed in an inversely consistent and symmetric manner, such that the template coordinates can be transformed to the patient's coordinate space in a topologically accurate manner and an inverse transformation can be applied after electrode implantation in the current subject to further add to the cohort prior dataset.

[0084] In box 510, for each probe, the group of ingress and target point coordinates for that probe is averaged to generate an average target and ingress point in the subject's anatomical coordinate system, based on which an average trajectory is defined. For each probe, anatomical partitions are further associated with the probe and can be used to further constrain the predicted trajectory in exemplary cases where the anatomical region of interest is small and / or adjacent to other sensitive anatomical structures, and / or the variability of the target coordinate distribution from previous groups is greater than the diameter of the structure. In this way, in one exemplary embodiment, anatomical partitions and previously implanted trajectory distributions can be used to generate prior information to enable the implantation of penetrating probes. In another exemplary embodiment, the anatomical target may be very large (e.g., as in the cingulate gyrus, it extends C-shaped from the anterior to the posterior of the skull), and in this case, target coordinates from a previous population can constrain the desired target location to the anterior, middle, or posterior part of the cingulate gyrus (example trajectories here are AC = anterior cingulate gyrus; MC = middle cingulate gyrus; PC = posterior cingulate gyrus), while anatomical partitions can further constrain the final target coordinates to remain within the boundaries of the cingulate gyrus, which is known to curve upwards and downwards along its course, and to curve backwards and backwards. In a further exemplary embodiment, a clinician in the art can use the patient's seizure symptomatology to derive information about specific anatomical regions that may cause seizures. This understanding can then be translated into informed trajectory planning by constraining the trajectory to the anatomical partition of interest. In another embodiment, surgical trajectories can be manually created individually by a person skilled in stereotactic practice by defining the entry and target points of interest and optimizing these relative to blood vessels and other possible trajectories. Alternatively, they can be derived by some combination of derivation from population-based average trajectories and manual optimization.

[0085] In box 512, constraints on the anatomical region can be achieved by adjusting the trajectory to intersect with the nearest voxel, which has a label assigned from the anatomical region of interest, wherein the distance is determined by calculating the Euclidean distance between the voxel of the probe trajectory and the labeled voxel in the anatomical region of interest.

[0086] In box 514, the intersection of a trajectory with any critical structure (such as a blood vessel) is evaluated using a loss function by identifying such trajectories that intersect with 2D and / or 3D surface and / or volume regions that are labeled or identified as belonging to critical structures, and penalizing any such intersecting trajectories. Furthermore, the proximity of the trajectory to critical structures such as blood vessels is checked against user-defined constraints (e.g., >2 mm from the edge of a blood vessel, or ≥4 mm from the center of an adjacent probe).

[0087] Automatic loss or optimization features can also be incorporated into trajectory planning, such that—in one exemplary embodiment—the total intracranial length is minimized while the sampled gray matter is maximized to achieve the greatest recording potential.

[0088] In box 516, starting from the initial trajectory estimated by prior information (defined by the average entry point and target point across the population), method 500 begins a local search around this average point until a trajectory that is as close as possible to the average trajectory and meets all safety and optimization criteria is identified. The search area is defined as a frustum, the diameter of each end of which is defined according to the standard deviation of the target distribution, and the center of each end is defined by the average entry and target coordinates from the population.

[0089] In block 518, the final trajectory is overlaid on the anatomical MRI dataset to generate a new planning dataset, which can be exported in any way for use with other software or hardware systems to visualize trajectory planning relative to the subject's anatomy.

[0090] When used in conjunction with the visualization techniques of Method 400, validation of the implantation planning of multiple stereotactic depth probes along an inclined trajectory is aided by simultaneously visualizing a cortical mesh model of the surface topology and deep anatomy revealed by structural magnetic resonance imaging along planar slices collinear with the proposed trajectory. Slicing the brain surface on any given plane enables visualization of the deeply embedded cortex in 3D and also allows clinicians to quickly confirm surgical plans.

[0091] The planning and operation of Method 500, combined with the anatomical visualization and analysis techniques of Methods 100-400, enables clinicians to identify key extracranial and intracranial structures (including but not limited to structures such as ventricles, white matter pathways, blood vessels, and functional areas) and avoid unwanted iatrogenic outcomes, including but not limited to bleeding and / or visual, language, cognitive, spatiotemporal, and / or sensorimotor deficits.

[0092] Combining methods 100-500 with white matter pathway analysis using deterministic or probabilistic fiber tractography (derived from diffusion imaging) can further identify risks in pathways involved in key functions such as motor, sensory, auditory, or visual processes. In specific indications—this approach can be used to reduce visual impairment following hippocampal and / or amygdala laser-mediated interstitial thermotherapy for medial temporal lobe epilepsy. 3D planning for optimal trajectories targeting the medial temporal lobe, combined with visualization of pathways identified by diffusion imaging, can be used to expand the therapeutic window of this technique.

[0093] Figure 6 A flowchart of method 600 is shown for automatically locating, naming, and visualizing previously implanted electrodes or penetrating brain probes, and resolving implanted structures in postoperative imaging using a template matching search algorithm and planned trajectories. Although depicted sequentially for convenience, at least some of the actions shown can be performed in a different order and / or in parallel. Additionally, some implementations may perform only some of the actions shown. The operation of method 600 can be performed by a computational system as disclosed herein.

[0094] In box 602, one or more imaging scans of the subject's brain are obtained. The imaging scans include target anatomical imaging scans (e.g., uncontrast T1-weighted MRI) and post-implantation CT imaging scans obtained after electrode implantation to locate the actual position of each electrode.

[0095] In box 604, imaging scans are converted from the original imaging storage format (e.g., DICOM) to a dataset according to method 100. Uncontrast T1-weighted MRI is converted into a dataset called the Anatomical MRI dataset, and CT imaging scans are converted into a dataset called the CT Electrode dataset.

[0096] In box 606, the two datasets are co-registered, and the CT electrode dataset is aligned to the anatomical MRI according to method 100.

[0097] In box 608, a third imaging scan is obtained. This third imaging scan is actually used by the surgeon during surgery as an anatomical imaging dataset to guide electrode implantation (referred to herein as an implantation MRI dataset or imaging scan). In one exemplary embodiment, this scan may be a contrast-enhanced T1-weighted MRI to provide high-resolution anatomical detail and reveal the location of blood vessels during implantation planning. According to method 100, the implantation MRI imaging scan is imported, co-registered, and aligned to the anatomical MRI.

[0098] In box 610, the trajectory implantation data file (referred to herein as the implantation log) is obtained. The implantation log is created during implantation, and an exemplary embodiment thereof is obtained using a robotic sEEG implantation system (e.g., Zimmer ROSA). TM Implantation files for subjects generated by robots, and another example is implantation files generated by navigation systems (e.g., BrainLab). TM Hemei Dunli Stealth TM The stereotactic file is created. The implantation log includes information about the probe / trajectory name and / or the planning target and / or the coordinates of the entry point and / or the probe trajectory vector defined relative to the patient coordinate space of the implanted MRI (as described in method 500).

[0099] Method 500 may also include a safety / manual verification feature, thereby requiring the user to manually input the probe name, the number of electrodes on each probe, and the number of electrodes to be ignored from each probe (e.g., for electrodes in the anchor, electrodes outside the brain, or electrodes not included in the record, etc.). The aforementioned information can be read directly from the implantation log or its equivalent (if available) to automatically obtain an initial list of the names of each probe and the estimated electrodes, and this initial list can be provided to the user as a template.

[0100] In box 612, the planned target and entry point coordinates provided for each probe from the implantation log, along with the number of electrodes in each probe as verified by the user, are used to calculate an initial list of expected coordinates for each electrode. This calculation is performed using the axes of the trajectory calculated from the entry and target point coordinates, the distance between the entry and target point coordinates, and the spacing between the electrodes. This information is used to generate a “virtual” object at the estimated location of each electrode for each probe trajectory in the probe trajectory. An exemplary embodiment of such a “virtual” object may be a sphere (e.g., a cylinder) centered at coordinates with a given geometry that matches the electrode geometry. This new dataset (referred to herein as the planned trajectory dataset) has the same volumetric geometry and coordinate space as the implanted MRI dataset.

[0101] In box 614, the planned trajectory dataset is aligned to the anatomical MRI dataset using a transformation matrix generated by aligning the implanted MRI dataset to the anatomical MRI dataset.

[0102] In box 616, the CT electrodes and the planned trajectory dataset are co-registered to the subject's anatomical MRI dataset according to method 100, and a binarization operation is performed on the CT electrode dataset, where imaging voxels below an automatically determined threshold level (e.g., 1mm x 1mm x 1mm cubes that may contain intensity information from the image, used as equivalent 3D pixels) are zeroed out. A 3D clustering algorithm is applied to the remaining voxels to identify voxels with high-intensity signals (sometimes referred to as metal artifacts) on the CT scan that have electrode contacts. An exemplary embodiment of the 3D clustering algorithm can be a standard clustering command provided by the underlying neuroimaging analysis software used. An iterative search is performed by adjusting the threshold until the number of clusters obtained is similar to the expected number of electrodes. The coordinates of these clusters are iteratively compared with the coordinates of spherical "virtual" electrodes generated for the planned trajectory dataset. Using line distances in 3D space and centroids for distance metrics, the clusters from the CT electrode dataset, as well as the trajectory paths and spherical object coordinates from the planned trajectory dataset, are iteratively searched and optimized until all expected electrodes are identified and located.

[0103] In box 618, a clustering algorithm, combined with trajectory path information and the expected number of electrodes (as provided by the input logs and represented in the planned trajectory dataset), is used to adjust the final positions of the electrode coordinates. This derives the final coordinate positions (e.g., cluster positions) that best match the imaging data, as well as the physical constraints of the expected trajectory (positions along specific lines separated by a specific distance from adjacent electrodes on the same path).

[0104] In box 620, after all electrode coordinates have been identified, 2D and / or 3D models of these electrodes are presented for visualization, with appropriate electrode names and numbering schemes assigned. The electrodes are visualized using displayable objects (e.g., cylinders or disks) that reflect the size, spacing, and dimensions of each actual electrode. With these electrodes co-registered to an anatomical MRI dataset, they can be visualized relative to 2D and / or 3D surface and volume representations of the relevant extracranial and intracranial structures generated by methods 200 and / or 300.

[0105] Displayable electrode objects can be manipulated individually (e.g., colored, annotated, numbered, visualized using different shapes or representations, turned on or off). They can be presented as transparent, translucent, or invisible along with any functional data (EEG) collected by the electrodes.

[0106] The methods and techniques described herein can be used in combination with other methods for surface-based representations of recorded intracranial EEG or other general functional activations or general neural relevances measured using implanted electrodes and / or penetrating probes and / or imaging modalities. The methods disclosed herein for surface-based representations can be applied not only to representations on cortical structures but also to anatomical meshes generated for the hippocampus, amygdala, and / or other general subcortical or brain structures. Such methods are disclosed in U.S. Patent No. 10,149,618.

[0107] Using the methods disclosed herein, the data representation of interest can be restricted to specific electrodes and derived into new datasets by superimposing the intensity values ​​of the voxels of interest onto the anatomical MRI dataset to generate new surface or volumetric activation datasets. In surface-based datasets, activations are assigned to surface nodes using a geodesic diffusion function, as described in a previous publication (Kadipasaoglu CM, Baboyan VG, Conner CR, Chen G, Saad ZS, Tandon N, Surface-based mixed effects multilevel analysis of grouped human electrocorticography, Neuroimage, 2014 Nov 1; 101:215-24. doi:10.1016 / j.neuroimage.2014.07.006. Epub 2014 July 12, PMID:25019677). In volume-based datasets, activations are restricted to voxels within the boundary between the pia mater and the surface layer of white matter (cortical band) beneath the electrodes of interest (Christopher R. Conner, Gang Chen, Thomas A. Pieters, Nitin Tandon, Category Specific Spatial Dissociations of Parallel Processes Underlying Visual Naming, Cerebral Cortex, Vol. 24, No. 10, October 2014, pp. 2741-2750, https: / / doi.org / 10.1093 / cercor / bht130). These datasets can be exported to disk in any manner for use with other software or hardware systems to visualize these activations relative to the subject's anatomy.

[0108] Figure 7 A graphical representation of the co-registration of different neuroimaging modalities performed on a single subject according to this disclosure is shown. Figure 7 In this dataset, sequence adaptive segmentation is applied to dataset 702 to produce labeled dataset 706, and sequence adaptive segmentation is applied to dataset 704 to produce labeled dataset 708. Labeled datasets 706 and 708 are co-registered, and the transformation matrix generated by the co-registration is applied to align labeled datasets 706 and 708, as shown in dataset 710.

[0109] Figure 8 shows a graphical representation of the generated 2D / 3D surface model depicting the hippocampus and thalamus according to the present disclosure. Figure 8A From Figure 8B An exemplary illustration of the anatomical mesh model of the right hippocampus generated by segmentation of a 3D volumetric dataset of an anatomical T1 MRI of the object shown.

[0110] Figure 8B and Figure 8C Exemplary illustrations depicting 2D / 3D surface mesh models of the subject's left hippocampus and amygdala, after anatomical atlas-based partitioning, are simultaneously visualized alongside the right hemisphere of the same subject. Figure 8C In the model, the surface model is visualized as distinct structures. The left cortical hemisphere is presented transparently, independent of the right, to allow visualization of the left hippocampus and amygdala. Figure 8D In the middle, the right cortical hemisphere of the partition appears translucent, allowing the solid representation of the underlying right hippocampus and amygdala to be visualized.

[0111] Figure 8E and Figure 8F An exemplary surface-based mesh model of the thalamus and its cell nuclei in a subject, generated using partitions derived from a microscopic stereotactic atlas, is shown. Figure 8E In this context, the surface model is an isolated view. Figure 8F In this study, the same model was viewed in relation to the three principal planes of the subject's original anatomical T1 MRI.

[0112] Figures 9A-9D Example steps for segmenting the human cerebral vascular system according to this disclosure are shown. Figures 9A-9C The original imaging dataset is depicted. Figure 9A In this paper, an exemplary embodiment of the original imaging dataset is a contrast-enhanced T1 MRI, which is subsequently segmented by cerebrospinal fluid volume (CSF). Figure 9B The vascular voxels were masked and then processed using a multi-scale Hessian-based filtering algorithm to accurately segment the blood vessels. Figure 9C ). Figure 9DThe resulting 3D surface model of blood vessels is depicted by superimposing segmented blood vessel volumes (right) and surface cerebral vascular system models (outlines) on three principal planes of the original contrast T1 MRI dataset, demonstrating comprehensive segmentation of the subject's vascular system.

[0113] Figures 10A-10W Graphical representations showing the intersection of 2D and / or 3D surface and volume models at arbitrary angles with 2D and / or 3D surface and volume models using 2D cutting planes (“slicers”) to optimize visualizations of cortical and subcortical structures and / or functions. Figures 10A-10C Depicting in relation to Figure 10A The cutting plane is observed on a 2D sagittal view of an anatomical T1-weighted MRI of a subject with overlapping skull bones as shown in the CT scan. Figure 10B The image shows a 3D surface model of the subject's complete skull, and... Figure 10C The image shows the skull after the cutting planes were applied. The skull is partially transparent to visualize the underlying cortical surface model with the same cutting planes applied.

[0114] Figures 10D-10F A rotated view of the same subject's skull and underlying cortical surface model is displayed. Note that the cutting plane can be rendered as opaque and confined within the boundaries of the 3D surface model to display the associated 2D MRI planar image. Figure 10D Alternatively, the cutting plane can be rendered as semi-transparent and / or the 2D MRI planar view can be extended beyond the boundaries of the underlying surface model. Figure 10E Finally, the cutting planes can be made invisible, and the bottom planes of the surface model can be made transparent, allowing deep anatomical structures to be visualized. Figure 10F ).

[0115] Figure 10G , Figure 10H and Figure 10I This displays sagittal views of a 2D cutting plane and an associated 3D partitioned cortical surface model at various rotation angles. Figure 10I In the model, the edges of the cortical model extend slightly beyond the boundaries of the cutting plane, with the boundaries of the gyri and sulci selectively enhanced to more accurately visualize the underlying anatomical features.

[0116] Figures 10J-10L Showing Figures 10G-10I The three views shown depict the same 2D sagittal cutting plane and 3D cortical surface model, with the edges of the model receding from the cutting plane. Figure 10J ), flush with the plane ( Figure 10K ), and slightly beyond the plane ( Figure 10L ).

[0117] Figure 10M , Figure 10N and Figure 10P Three views are shown: a 2D coronal cutting plane and a 3D skin and segmented cortical surface model. The complete skin model intersects the cutting plane, and the remaining skin and segmented cortical models are visualized. Figure 10M ).exist Figure 10N The partitioned cortical model is shown separately and with reference to the cutting plane, where the edges extend slightly beyond the plane, and then again in the third view, but in this view, the edges are limited to only the gray and white matter boundaries, such that the extension of the gray and white matter boundary edges beyond the cutting plane isolates the intermediate cortical band. Figure 10P In the middle, it is depicted from Figure 10N A magnified and slightly rotated view of the third image, where the white arrow indicates the exemplary region contained within the aforementioned cortical band between the edges of gray and white matter. Figure 10P )

[0118] Figure 10Q-Figure 10W The cortex, simultaneously representing surface and deep anatomical structures, and cortical activity represented as a color scale, are displayed via slices along planes collinear with the depth trajectory. Visualization of brain structural data (including one or more of MRI, CT, PET, fMRI, DTI) and / or brain activity data (including one or more of EEG, MEG, or brain stimulation) relative to the cutting planes of various anatomical mesh models can be selectively depicted to optimize the neocortex ( Figure 10Q , Figure 10R , Figure 10S , Figure 10T and Figure 10U ) and / or hippocampus and amygdala ( Figure 10V and Figure 10W Visualization of functional activity in the subcortical or other brain regions. Relevant viewpoints for visualizing cutting planes and related surfaces ( Figure 10T-Figure 10V () are drawn by line 1002 and arrow 1004 respectively.

[0119] Figures 11A-11R A graphical representation of population-derived anatomical targets for electrode or penetrating probe implantation is shown, which combines priors derived using probability distributions from previously implanted populations and / or partitioning and segmentation based on anatomical atlases. Figures 11A-11D Grouped representations depicting the trajectories of 130 patients with 2600 electrodes implanted for epilepsy detection entering the brain were constructed. These grouped representations were co-registered and aligned to a common brain space and color-coded by entry and target points. Figure 11AElectrodes can be further color-coded based on standard regional nomenclature applied to them, thereby indicating similar entry and target points for specific cortical or subcortical lesions between individuals. An exemplary embodiment is depicted for the right amygdala and hippocampus of a single subject. Figure 11B Using information from previous trajectories in this population, new trajectories can be derived for any specific brain region in a new individual (not one of the previous 130 patients). An exemplary illustration of the analysis is provided for the right anterior hippocampus (RAH) of a single subject, where the new trajectory is depicted as an elongated cylinder, while using each individual probe ( Figure 11C (shorter cylinders) or by visualizing the mean and variance of the population ( Figure 11D The population prior trajectory is depicted using a frustum, which in this exemplary illustration depicts the population's mean and variance using the mean and 1.5 times the standard deviation of the ingress and target point coordinates.

[0120] Figure 11E An integrated approach is depicted, including a tilted cutting plane, a detailed cerebral vascular system and a partitioned anatomical mesh model, and a trajectory planning algorithm for generating automated implantation trajectories for 12 different brain probes (e.g., sEEG probes). The automated algorithm ensures compliance with several safety constraints, an exemplary embodiment of which may be minimum distances to adjacent vessels along the trajectory and to adjacent probes. Panel 11E-2 depicts an exemplary illustration of manual trajectory optimization, where two cylinders are visualized to represent the original (i.e., automatically derived) and manually adjusted trajectories of the right anterior hippocampus (RAH) probe.

[0121] Figure 11F Similar population-level derivation plans for laser interstitial thermotherapy of the amygdala and / or hippocampus for medial temporal lobe epilepsy are depicted. The optimal novel trajectory for new subjects, along with the predicted ablation amount expected from the population derivation for a given trajectory, are visualized.

[0122] Figure 11G and Figure 11H Exemplary illustrations depict novel trajectories derived from population data of previous implantation trajectories targeting multiple regions within the left cingulate gyrus, including the left rostral cingulate (LRC), left anterior cingulate (LAC), left medial cingulate (LMC), and left posterior cingulate (LPC) regions. These illustrations include side views highlighting the entry point. Figure 11GThe image shows a medial view depicting the proposed trajectory marked by the aforementioned target brain regions, wherein the left hemisphere has been rendered completely transparent, making the cingulate gyrus of the right hemisphere visible (and usable as a visual reference to the contralateral target brain regions), and wherein the proposed trajectory is depicted using its associated frustum (derived from the population) rendered with a semi-transparent overlay. Figure 11H ).

[0123] Figures 11J-11P Top view using the 3D cortical surface model ( Figure 11J – Figure 11L Both the lateral and side views (11M-11P) depict exemplary illustrations of another subset of exemplary suggested trajectories that a subject undergoing stereotactic brain imaging evaluation for refractory epilepsy might expect, with the left hemisphere presented as opaque (11J and 11M) or completely transparent (11K-11L and 11N-11P). The middle illustration depicts population data of previously implanted trajectories used to generate their corresponding new trajectories, visualizing each of the previously implanted probes overlapping their corresponding new trajectories using cylinders color-coded by the target brain region. Figure 11K and Figure 11N The aforementioned frustum was generated using the mean and 1.5-fold standard deviation of the coordinate distribution of the previously implanted trajectories. This frustum is depicted in the rightmost illustration as a semi-transparent overlay with their respective trajectories (11L and 11P). The bottom row depicts an exemplary summary illustration of all trajectories from the previously implanted population data, all visualized with their respective frustums on a 3D cortical surface model of a single exemplary subject, presented as opaque and fully transparent (11Q and 11R, respectively).

[0124] Figures 12A-12E A graphical representation of automated electrode localization and labeling is depicted, incorporating implantation trajectory logs from a robotic sEEG implantation system to constrain and inform the electrode search algorithm, providing probe names and the number of associated electrodes. An initial clustering algorithm applied to a post-implantation CT electrode dataset is described in 12A, demonstrating how an increased intensity threshold, used to zero out voxels with intensities below a threshold, can identify clusters of high-intensity voxels representing artifacts from electrode contacts in the CT scanner. The trajectory implantation logs from the robotic implantation system can also be used to further inform electrode search by limiting the algorithm's search space to more effectively separate signals associated with electrode artifacts from noise. Figure 12B Furthermore, it ensures that the final electrode coordinates are spaced and aligned in a manner consistent with the actual implantation as defined by the spherical virtual electrodes. Figure 12C ).

[0125] Figure 12DA cutting plane is depicted applied at an angle to a model of the subject's skull to visualize the implanted electrodes relative to a model of the subject's right hippocampus and amygdala surface. In this exemplary embodiment, each electrode is presented as a cylinder with inter-electrode spacing and dimensions determined by the implantation trajectory log and the physical dimensions of the actual electrode. Probes and their corresponding electrodes are color-coded by probe name. Figure 12E A more reduced view of the right hippocampus and amygdala of the same subject is depicted, where a subset of implanted probes is visualized as displayable objects and color-coded by probe names, which are also annotated in white. Trajectories from the trajectory implantation log are also depicted here as translucent cylinders with smaller dimensions and spacing to distinguish them from the actual electrode locations. As can be seen from the highlighted electrodes with covered crosses, the final electrode coordinates do not always correspond perfectly to the planned trajectory because the probes may deflect during implantation. The highlighted electrode coordinates correspond to the coordinates of the crosses in adjacent 2D coronal and sagittal plane images of the same subject from a pre-implantation MRI covered by a post-implantation CT scan.

[0126] Figure 13 A block diagram of a computing system 1300 is shown, applicable to implementations of the methods disclosed herein (e.g., methods 100, 200, 300, 400, 500, and / or 600). The computing system 1300 includes one or more computing nodes 1302 and auxiliary storage 1316, which are communicatively coupled (e.g., via network 1318). One or more of the computing nodes 1302 and the associated auxiliary storage 1316 can be used to perform the operations described herein.

[0127] Each compute node 1302 includes one or more processors 1304 coupled to memory 1306, network interface 1312, and I / O devices 1314. In various embodiments, the compute node 1302 may be a single-processor system including one processor 1304, or a multiprocessor system including several processors 1304 (e.g., two, four, eight, or another suitable number). The processor 1304 may be any suitable processor capable of executing instructions. For example, in various embodiments, the processor 1304 may be a general-purpose or embedded microprocessor, graphics processing unit (GPU), or digital signal processor (DSP) implementing any of a variety of instruction set architectures (ISAs). In a multiprocessor system, each of the processors 1304 may typically (but not necessarily) implement the same ISA.

[0128] Memory 1306 may include a non-transient computer-readable storage medium configured to store program instructions 1308 and / or data 1310 accessible by processor(s) 1304. Memory 1306 may be implemented using any suitable memory technology, such as static random access memory (SRAM), synchronous dynamic RAM (SDRAM), non-volatile / flash memory, or any other type of memory. Program instructions 1308 and data 1310 implementing the functions disclosed herein are stored in memory 1306. For example, instructions 1308 may include instructions that, when executed by processor(s) 1304, implement one or more of the methods disclosed herein.

[0129] Auxiliary storage 1316 may include volatile or non-volatile storage and storage devices for storing information, such as program instructions and / or data for implementing the methods described herein. Auxiliary storage 1316 may include various types of computer-readable media accessible from computing node 1302 via network interface 1312. Computer-readable media may include storage media or memory media, such as semiconductor storage, magnetic or optical media, such as magnetic disks or CD / DVD-ROMs or other storage technologies.

[0130] Network interface 1312 includes circuitry configured to allow data exchange between compute node 1302 and / or other devices coupled to network 1318. For example, network interface 1312 may be configured to allow data exchange between a first instance of compute system 1300 and a second instance of compute system 1300. Network interface 1312 may support communication via wired or wireless data networks.

[0131] I / O device 1314 allows computing node 1302 to communicate with various input / output devices, such as one or more display terminals, keyboards, keypads, touchpads, scanning devices, voice or optical recognition devices, or any other device suitable for inputting or retrieving data through one or more computing nodes 1302. Multiple input / output devices may exist in computing system 1300.

[0132] The computing system 1300 is illustrative only and is not intended to limit the scope of the embodiments. In particular, the computing system 1300 may include any combination of hardware or software capable of performing the functions disclosed herein. In some embodiments, the computing node 1302 may also be connected to other devices not shown. Additionally, in some embodiments, the functionality provided by the illustrated components may be combined in fewer components or distributed across additional components. Similarly, in some embodiments, the functionality of some of the illustrated components and / or other additional functions may not be provided, although they may be available.

[0133] The foregoing discussion is intended to illustrate the principles and embodiments of the invention. Once the foregoing disclosure is fully understood, numerous variations and modifications will become apparent to those skilled in the art. The appended claims are intended to be construed as encompassing all such variations and modifications.

Claims

1. A method for brain imaging scanning, the method comprising: Obtain first and second imaging scans of a single subject's brain; The first imaging scan is converted into a first dataset, and the second imaging scan is converted into a second dataset; The sequence adaptive multimodal segmentation algorithm is applied to the first dataset, wherein the sequence adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled dataset; The sequence adaptive multimodal segmentation algorithm is applied to the second dataset, wherein the sequence adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a second labeled dataset; The first and second labeled datasets are automatically registered together to generate a transformation matrix based on the first and second labeled datasets; as well as The transformation matrix is ​​applied to align the first dataset and the second dataset.

2. The method as described in claim 1, characterized in that, The first imaging scan and the second imaging scan are performed using one or more of the following: magnetic resonance imaging (MRI), computed tomography (CT), magnetoencephalography (MEG), or positron emission tomography (PET).

3. The method as described in claim 1, characterized in that, The sequence adaptive multimodal segmentation algorithm assigns a digital label value to each voxel in the first dataset or the second dataset.

4. The method of claim 1, further comprising: Extract voxels from the first dataset, which has labels corresponding to the subcortical regions of interest; A third dataset is formed, containing voxels extracted from the first dataset; The third dataset is converted into a first subcortical surface mesh model; Calculate the curvature and sulcus features of the first subcortical surface mesh model; The curvature and sulcus features are used to align the first subcortical surface mesh model to the subcortical atlas of the region of interest. as well as A first subcortical surface mesh model of the map aligned to the subcortical region of interest is overlaid on a second subcortical surface mesh model, the second subcortical surface mesh model having a standardized number of nodes that realize a one-to-one correspondence between node identity and map location; as well as The coordinates of the nodes of the first subcortical surface mesh model are assigned to the second subcortical structural surface mesh model, so that the second subcortical surface mesh model presents the topology of the first subcortical structural surface mesh model.

5. The method as described in claim 1, characterized in that: The first imaging scan is a contrast-weighted MRI scan and the first dataset is a contrast-weighted dataset; and The method further includes: Based on the labeled dataset, select the voxels in the first dataset that are identified as belonging to the cerebrospinal fluid region; A multi-scale tubular filtering algorithm is applied to identify voxels representing blood vessels in the first dataset and assign blood vessel weight values ​​to each voxel. Integrate the blood vessel weight values ​​into the first dataset; and After aligning the first dataset and the second dataset, the first dataset is converted into a surface anatomy mesh model.

6. The method as described in claim 1, characterized in that: The first imaging scan is a contrast-weighted MRI scan, and the second imaging scan is an anatomical MRI scan; and The method further includes: The predicted target point coordinates and entry point coordinates of the probe are defined based on the target point coordinates and entry point coordinates of the previously implanted probe, or by using user-defined target points and entry points. The trajectory of the probe is defined based on the average target coordinates and the average entry point coordinates; The trajectory is adjusted to intersect with the nearest voxel of the label of the assigned anatomical region of interest; Based on user-defined constraints of the trajectory and / or user-defined modifications to satisfy the user-defined constraints, the proximity of the trajectory to the critical structure is checked; and The trajectory is overlaid on the second dataset to form a planning dataset.

7. The method as described in claim 1, characterized in that: The first imaging scan is an anatomical MRI scan, the second imaging scan is a post-implantation CT imaging scan, the first dataset is an anatomical MRI dataset, and the second dataset is a post-implantation CT electrode dataset; and The method further includes: Obtain a third imaging scan for guiding electrode implantation during surgery; Convert the third imaging scan into a third dataset; Align the third dataset with the first dataset; Obtain the trajectory implantation data file created during the electrode implantation; A planned trajectory dataset is generated based on the trajectory implantation data file, the planned trajectory dataset including virtual objects set at the positions of the electrode geometry; Align the planned trajectory dataset to the anatomical MRI dataset; and Based on the virtual object in the trajectory implantation data file, the electrodes in the CT electrode dataset are automatically identified and labeled.

8. A non-transitory computer-readable medium, said non-transitory computer-readable medium being encoded with instructions executable by one or more processors to: Obtain first and second imaging scans of a single subject's brain; The first imaging scan is converted into a first dataset, and the second imaging scan is converted into a second dataset; The sequence adaptive multimodal segmentation algorithm is applied to the first dataset, wherein the sequence adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled dataset; The sequence adaptive multimodal segmentation algorithm is applied to the second dataset, wherein the sequence adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a second labeled dataset; The first and second labeled datasets are automatically registered together to generate a transformation matrix based on the first and second labeled datasets; as well as The transformation matrix is ​​applied to align the first dataset and the second dataset.

9. The non-transient computer-readable medium as claimed in claim 8, characterized in that, The first imaging scan and the second imaging scan are performed using one or more of the following: magnetic resonance imaging (MRI), computed tomography (CT), magnetoencephalography (MEG), or positron emission tomography (PET).

10. The non-transient computer-readable medium as claimed in claim 8, characterized in that, The sequence adaptive multimodal segmentation algorithm assigns a digital label value to each voxel in the first dataset or the second dataset.

11. The non-transient computer-readable medium as claimed in claim 8, characterized in that, The instructions can be executed by the one or more processors to: Extract voxels from the first dataset, which has labels corresponding to the subcortical regions of interest; A third dataset is formed, containing voxels extracted from the first dataset; The third dataset is converted into a first subcortical surface mesh model; Calculate the curvature and sulcus features of the first subcortical surface mesh model; The curvature and sulcus features are used to align the first subcortical surface mesh model to the subcortical atlas of the region of interest. as well as A first subcortical surface mesh model of the map aligned to the subcortical region of interest is overlaid on a second subcortical surface mesh model, the second subcortical surface mesh model having a standardized number of nodes that realize a one-to-one correspondence between node identity and map location; as well as The coordinates of the nodes of the first subcortical surface mesh model are assigned to the second subcortical structural surface mesh model, so that the second subcortical surface mesh model presents the topology of the first subcortical structural surface mesh model.

12. The non-transient computer-readable medium as claimed in claim 8, characterized in that: The first imaging scan is a contrast-weighted MRI scan and the first dataset is a contrast-weighted dataset; and The instructions can be executed by the one or more processors to: Based on the labeled dataset, select the voxels in the first dataset that are identified as belonging to the cerebrospinal fluid region; A multi-scale tubular filtering algorithm is applied to identify voxels representing blood vessels in the first dataset and assign blood vessel weight values ​​to each voxel. The blood vessel weight values ​​are integrated into the first dataset; as well as After aligning the first dataset and the second dataset, the first dataset is converted into a surface anatomy mesh model.

13. The non-transient computer-readable medium as claimed in claim 8, characterized in that: The first imaging scan is a contrast-weighted MRI scan, and the second imaging scan is an anatomical MRI scan; and The instructions can be executed by the one or more processors to: The predicted target point coordinates and entry point coordinates of the probe are defined based on the target point coordinates and entry point coordinates of the previously implanted probe, or by using user-defined target points and entry points. The trajectory of the probe is defined based on the average target coordinates and the average entry point coordinates; The trajectory is adjusted to intersect with the nearest voxel of the label of the assigned anatomical region of interest; The proximity of the trajectory to the critical structure is checked based on user-defined constraints and / or user-defined modifications to satisfy the user-defined constraints. as well as The trajectory is overlaid on the second dataset to form a planning dataset.

14. The non-transient computer-readable medium as claimed in claim 8, characterized in that: The first imaging scan is an anatomical MRI scan, the second imaging scan is a post-implantation CT imaging scan, the first dataset is an anatomical MRI dataset, and the second dataset is a post-implantation CT electrode dataset; and The instructions can be executed by the one or more processors to: Obtain a third imaging scan for guiding electrode implantation during surgery; Convert the third imaging scan into a third dataset; Align the third dataset with the first dataset; Obtain the trajectory implantation data file created during the electrode implantation; A planned trajectory dataset is generated based on the trajectory implantation data file, the planned trajectory dataset including virtual objects set at the positions of the electrode geometry; Align the planned trajectory dataset to the anatomical MRI dataset; as well as Based on the virtual object in the trajectory implantation data file, the electrodes in the CT electrode dataset are automatically identified and labeled.

15. A system for brain imaging scanning, the system comprising: One or more processors; Memory, coupled to the one or more processors, wherein the memory stores instructions that configure the one or more processors to: Obtain first and second imaging scans of the subject's brain; The first imaging scan is converted into a first dataset, and the second imaging scan is converted into a second dataset; The sequence adaptive multimodal segmentation algorithm is applied to the first dataset, wherein the sequence adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a first labeled dataset; The sequence adaptive multimodal segmentation algorithm is applied to the second dataset, wherein the sequence adaptive multimodal segmentation algorithm performs automatic intensity-based tissue classification to generate a second labeled dataset; The first and second labeled datasets are automatically registered together to generate a transformation matrix based on the first and second labeled datasets; as well as The transformation matrix is ​​applied to align the first dataset and the second dataset.

16. The system as described in claim 15, characterized in that, The first imaging scan and the second imaging scan are performed using one or more of the following: magnetic resonance imaging (MRI), computed tomography (CT), magnetoencephalography (MEG), or positron emission tomography (PET).

17. The system as claimed in claim 15, characterized in that, The sequence adaptive multimodal segmentation algorithm assigns a digital label value to each voxel in the first dataset or the second dataset.

18. The system as described in claim 15, characterized in that, The instructions configure the one or more processors to: Extract voxels from the first dataset, which has labels corresponding to the subcortical regions of interest; A third dataset is formed, containing voxels extracted from the first dataset; The third dataset is converted into a first subcortical surface mesh model; Calculate the curvature and sulcus features of the first subcortical surface mesh model; The curvature and sulcus features are used to align the first subcortical surface mesh model to the subcortical atlas of the region of interest. as well as A first subcortical surface mesh model of the map aligned to the subcortical region of interest is overlaid on a second subcortical surface mesh model, the second subcortical surface mesh model having a standardized number of nodes that realize a one-to-one correspondence between node identity and map location; as well as The coordinates of the nodes of the first subcortical surface mesh model are assigned to the second subcortical structural surface mesh model, so that the second subcortical surface mesh model presents the topology of the first subcortical structural surface mesh model.

19. The system as described in claim 15, characterized in that: The first imaging scan is a contrast-weighted MRI scan and the first dataset is a contrast-weighted dataset; and The instructions configure the one or more processes to be used for: Based on the labeled dataset, select the voxels in the first dataset that are identified as belonging to the cerebrospinal fluid region; A multi-scale tubular filtering algorithm is applied to identify voxels representing blood vessels in the first dataset and assign blood vessel weight values ​​to each voxel. The blood vessel weight values ​​are integrated into the first dataset; as well as After aligning the first dataset and the second dataset, the first dataset is converted into a surface anatomy mesh model.

20. The system as described in claim 15, characterized in that: The first imaging scan is a contrast-weighted MRI scan, and the second imaging scan is an anatomical MRI scan; and The instructions configure the one or more processes to be used for: The predicted target point coordinates and entry point coordinates of the probe are defined based on the target point coordinates and entry point coordinates of the previously implanted probe, or by using user-defined target points and entry points. The trajectory of the probe is defined based on the average target coordinates and the average entry point coordinates; The trajectory is adjusted to intersect with the nearest voxel of the label of the assigned anatomical region of interest; The proximity of the trajectory to the critical structure is checked based on user-defined constraints and / or user-defined modifications to satisfy the user-defined constraints. as well as The trajectory is overlaid on the second dataset to form a planning dataset.

21. The system as described in claim 15, characterized in that: The first imaging scan is an anatomical MRI scan, the second imaging scan is a post-implantation CT imaging scan, the first dataset is an anatomical MRI dataset, and the second dataset is a post-implantation CT electrode dataset; and The instructions configure the one or more processes to be used for: Obtain a third imaging scan for guiding electrode implantation during surgery; Convert the third imaging scan into a third dataset; Align the third dataset with the first dataset; Obtain the trajectory implantation data file created during the electrode implantation; A planned trajectory dataset is generated based on the trajectory implantation data file, the planned trajectory dataset including virtual objects set at the positions of the electrode geometry; Align the planned trajectory dataset to the anatomical MRI dataset; as well as The virtual object in the CT electrode dataset is identified based on the trajectory implantation data file.

Images