TREATMENT OF PATIENTS WITH TTFIELDS WITH ELECTRODE POSITIONS OPTIMIZED WITH THE USE OF DEFORMABLE GA

专利摘要:
the modalities receive images of an area of a patient's body; identify abnormal tissue in the image; generate a data set with the masked abnormal tissue; deform a model template in space so that the features in the deformed model template align with the corresponding features in the data set; put data that represents the abnormal tissue back to the deformed model template; generate a model of electrical properties of tissues in the body area based on the deformed and modified model template; and determine an electrode placement layout that maximizes the field strength in the abnormal tissue using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed in the body area. the layout can then be used as a guide for placing respective electrodes for the patient's body area to apply tfields to the body area.
公开号:BR112019012033A2
申请号:R112019012033-5
申请日:2017-12-13
公开日:2020-03-03
发明作者:Bomzon Zeev；Urman Noa
申请人:Novocure Gmbh；
IPC主号:

专利说明:

TREATMENT OF PATIENTS WITH TTFIELDS WITH ELECTRODE POSITIONS OPTIMIZED WITH THE USE OF DEFORMABLE KITS
REMISSIVE REFERENCE TO RELATED ORDERS [OO1] This request claims the benefit of US Provisional Order No. 62 / 433,501, (filed on December 13, 2016), which is incorporated herein in its entirety, for reference.
BACKGROUND [002] The use of electric fields and currents for the treatment of neurological disorders and brain diseases is becoming widely distributed. Examples of such treatments include, but are not limited to: Transcranial Direct Current Stimulation (TDCS), Transcranial Magnetic Stimulation (TMA) and Tumor Treating Fields - TTFields. These treatments depend on providing low frequency electromagnetic fields to target regions within the brain. See, for example, Woods et. al., Clinical Neurophysiology, 127 1031-1048 (2016), which reviews technical aspects of TDCS; and Thielscher et. al., Conference Proceedings, Institute of Electrical and electronics Engineers (IEEE), Engineering in Medicine and Biology Society, 222-225 (2015), which teaches methods to simulate TMS. As yet another example, Miranda et. al., Physics in Medicine and Biology, 59, 4137-4147 (2014), teaches the creation of a computational head model of a healthy individual to simulate the supply of TTFields using a magnetic resonance imaging data set (magnetic resonance imaging - MRI), where the model is created in a semi-automatic way. In addition, Wenger et. al., Physics in Medicine and Biology, 60 7339-7357 (2015), teaches a method to create a computational head model of a healthy individual to simulate the supply of TTFields, where the model is created from data sets from MRI of a
Petition 870190054241, of 6/12/2019, p. 10/59
2/37 healthy individual.
[003] In the case of TDCS and TMS, treatment involves providing electromagnetic fields to target regions in the brain where they stimulate specific neurons. In the case of TTFields, the position of the transducer arrays in the patient's head is optimized to provide maximum field strength to the tumor region. See, for example, Wenger et. al., International Journal of Radiation Oncology Biology Physics, 941137-43 (2016), which teaches how Diffusion Tensor Imaging (DTI) data can be incorporated into models to simulate the supply of TTFields in the head. DTI data is used to derive anisotropic conductivity tensors for each voxel in the head model.
[004] TTFields are low intensity alternating electric fields (for example, 1 to 3 V / cm) within the intermediate frequency range (100 to 300 kHz), which can be used, for example, to treat tumors as described in US patent 7,565,205, which is incorporated herein by reference in its entirety. TTField therapy is an approved monotreatment for recurrent glioblastoma (GBM), and an approved combination therapy with chemotherapy for newly diagnosed patients. These alternating electrical fields are induced non-invasively by transducer arrays (ie capacitively coupled electrode arrays) placed directly on the patient's scalp (for example, using the Novocure Optune ™ system). TTFields also appear to be beneficial for treating tumors in other parts of the body.
[005] In vivo and in vitro studies show that the effectiveness of TTFields therapy increases as the intensity of the electric field increases in the target region, and the intensity in the target region is dependent on the placement of the transducer arrays on the patient's scalp .
[006] One way to optimize the placement of the transducer arrays is to use
Petition 870190054241, of 6/12/2019, p. 11/59
3/37 a computer simulation. The use of a computer is necessary due to the large amount of imaging data that is processed and the simulation / optimization process that is computationally intensive and complex as described here. Normally, when performing the simulations, an anatomically accurate computational model is constructed, and the electrical properties are attributed to the various types of tissues. Once the model has been built, the simulated model electrodes are positioned on the head model and the appropriate boundary conditions, such as voltage on the electrodes, are applied. The electric field inside the head is then calculated. By using various computer-implemented and computationally intensive optimization schemes, it is then possible to find the electrode layout and boundary conditions that generate the optimal distributions of the electromagnetic field within the head (and specifically, the target regions). However, individual patients vary in the details of their anatomy, and these variations influence the field distribution within the individual's head. Therefore, in order to use simulations to optimize treatments involving the supply of electromagnetic fields to target regions, so far it has been necessary to build a customized computational model for each individual.
[007] A conventional approach to forming a head model is as follows. First, a set of medical images is acquired. Usually, the images include magnetic resonance imaging (MRI) and / or computed tomography (Computed Tomography - CT). The images are then segmented to determine which portions of the images correspond to each of the different types of tissue possible (for example, white matter, gray matter, cerebrospinal fluid (cerebrospinal fluid - CSF), skull, etc.). Then, a series of meshes for each type of fabric in the segmented image is constructed and incorporated into the model, and representative values of conductivity are assigned to each type of fabric. Finally, the electrodes are positioned on the model
Petition 870190054241, of 6/12/2019, p. 12/59
4/37 and field distribution is resolved using an appropriate numerical technique, such as a finite element method or a finite difference method (based on the positions in 3D space of the various types of tissue and the conductivities assigned to each one of these types of fabric).
[008] Although many steps in the process described above are implemented by a computer, the process still requires a great deal of human intervention because the automatic algorithms for segmenting medical images of a head, especially images in which tumors are present, are not robust and often require user intervention to obtain reliable results. See, for example, Menze et. al., IEEE Transactions on Medical Imaging, 34 1993-2024 (2014), which investigates the performance of multiple algorithms for automatic segmentation of tumors. In addition, the regularization of the grid is a lengthy process that requires user supervision, as described, for example, in Miranda et. al., Physics in Medicine and Biology, 59, 4137-4147 (2014), Wenger et. al., Physics in Medicine and Biology, 60 7339-7357 (2015), and Wenger et. al., International Journal of Radiation Oncology Biology Physics, 941137-43 (2016). Specifically, when creating a finite element model of a volume, the volume is meshed into volumetric elements. In order to ensure the conversion of the numerical solution, it is desirable that the quality of all elements is high (with the definition of quality varying depending on the type of mesh being created). In addition, it is important to check that the elements do not intersect and that, in general, the quality of the mesh is sufficient. Regularization is a process in which a mesh is processed to improve the conditioning of its elements and its overall quality. For a basic discussion, see S. Makarow et. al., Low Frequency Electromagnetic Modeling For Electrical and Biological systems Using Matlab, John Wiley and Sons, 2010, pages 36 to 81.
[009] Between segmentation and regularization of the grid, man-hours
Petition 870190054241, of 6/12/2019, p. 13/59
5/37 required to create a single model can vary from hours to days, depending on the quality of the images and the complexity of the model being created.
SUMMARY OF THE INVENTION [010] One aspect of the invention is directed to a first method for improving the treatment of a tumor with the use of Tumor Treatment Fields (TTFields). The first method includes receiving, through a computer system processor, a three-dimensional image of an area of a patient's body, identifying portions of the image that correspond to the abnormal tissue and generating a data set corresponding to the image with the masked abnormal tissue . The first method further includes retrieving a model template from a computer system memory device, the model comprising tissue probability maps that specify positions of a plurality of tissue types in a healthy version of the patient's body area and deform the model template in space so that the features in the deformed model template are aligned with the corresponding features in the data set. The first method also includes modifying the deformed model template portions that correspond to the masked portion of the data set, so that the modified portions represent the abnormal tissue, and generating a model of electrical properties of tissues in the body area based on (a) positions of the plurality of fabric types in the deformed and modified model template and (b) abnormal tissue positions in the deformed and modified template template. The first method also includes determining an electrode placement layout that maximizes field strength in at least a portion of the abnormal tissue by using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed on a plurality of different sets of candidate positions for the body area, and selection of one of the sets. The first method also includes placing the electrodes on the patient's body area based on
Petition 870190054241, of 6/12/2019, p. 14/59
6/37 in the placement layout of the given electrode; and use the electrodes placed to apply the TTFields to the body area.
[O11] Another aspect of the invention is directed to a second method for improving an electrotherapeutic treatment. The second method includes receiving, through a computer system processor, a three-dimensional image of an area of a patient's body, identifying portions of the image that correspond to the abnormal tissue and generating a data set corresponding to the image with the masked abnormal tissue . The second method also includes retrieving a model template from a computer system memory device, where the model template specifies positions of a plurality of tissue types in a healthy version of the patient's body area and deforming the template. model in space so that the features in the deformed model template are aligned with the corresponding features in the data set. The second method also includes modifying the deformed model template portions that correspond to the masked portion of the data set, so that the modified portions represent the abnormal tissue, and generating a model of electrical properties of tissues in the body area based on (a) positions of the plurality of fabric types in the deformed and modified model template and (b) abnormal tissue positions in the deformed and modified template template. The second method also includes determining an electrode placement layout that maximizes the field strength in at least a portion of the abnormal tissue by using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed on a plurality of different sets of candidate positions for the body area, and selection of one of the sets. The second method also includes producing the electrode placement layout determined for subsequent use as a guide for placing electrodes corresponding to the patient's body area prior to using the electrodes for electrotherapeutic treatment.
Petition 870190054241, of 6/12/2019, p. 15/59
7/37 [012] In some modalities of the second method, deformation of the model template includes determining a mapping that maps the data set to a model template coordinate space, and applying an inverse of the mapping to the model template. Optionally, in these modalities, the mapping is determined for points in the data set that are outside the masked portion. Optionally, in these modalities, the model template comprises tissue probability maps, where the mapping maps the data set to the tissue probability maps.
[013] Optionally, in these modalities, tissue probability maps are derived from images of a healthy individual from whom the model template was derived. Optionally, in these modalities, tissue probability maps are derived by simultaneously registering and segmenting images of the healthy individual using existing tissue probability maps, and in which existing tissue probability maps are derived from images of multiple individuals .
[014] Optionally, in these modalities, tissue probability maps are tissue probability maps derived from images of multiple individuals.
[015] Optionally, in these modalities, the inverse of the mapping is applied to each of the tissue probability maps, in which the tissue probability maps inversely mapped are combined into a segmented image comprising the deformed model template. Optionally, in these modalities, the combination of the tissue probability maps inversely mapped includes assigning to each voxel the type of tissue that is most likely to occupy that voxel along the tissue probability maps inversely mapped. Optionally, in these modalities, the combination of tissue probability maps inversely mapped includes the use of a
Petition 870190054241, of 6/12/2019, p. 16/59
8/37 query to assign a type of tissue to each voxel that is associated with more than one type of tissue through the inversely mapped tissue probability maps.
[016] In some modalities of the second method, the identification of the portions of the image that correspond to the abnormal tissue comprises the segmentation of the image. In some embodiments of the second method, the model of electrical properties of tissues comprises a model of electrical conductivity or resistivity. In some embodiments of the second method, the image comprises an MRI image, a CT image, or a combination of MRI and CT images. In some embodiments of the second method, the body area comprises the patient's head. In some embodiments of the second method, the portions of the image that correspond to the abnormal tissue correspond to a tumor. In some modalities of the second method, electrotherapeutic treatment comprises TTFields.
[017] In some modalities of the second method, determining the electrode placement layout involves applying a boundary condition to the simulated electrodes in each of at least two electrode placement layouts, resolving a field distribution in the body area to each of the at least two electrode placement layouts, and choose the electrode placement layouts that produce the strongest field within the abnormal region. Optionally, in these modalities, the boundary condition corresponds to voltages or currents applied to the simulated electrodes.
[018] In some modalities of the second method, the model template is selected from a plurality of model templates based on the similarities between the image and each of the model templates.
[019] Some modalities of the second method also include placing the electrodes in the patient's body area based on the determined electrode placement layout; and use the electrodes to apply the TTFields to the body area.
Petition 870190054241, of 6/12/2019, p. 17/59
9/37 [020] Another aspect of the invention is directed to an electrotherapeutic treatment device that comprises a processor configured to execute instructions stored in one or more memory devices to carry out an electrotherapeutic treatment. In these modalities, the treatment includes receiving, through the processor, a three-dimensional image of an area of a patient's body, identifying portions of the image that correspond to the abnormal tissue and generating a data set corresponding to the image with the masked abnormal tissue. Treatment also includes retrieving a model template from one or more memory devices, in which the model template specifies positions of a plurality of tissue types in a healthy version of the patient's body area, deforming the model template in space so that the features in the deformed model template align with the corresponding features in the data set, and modify the deformed model template portions that correspond to the masked portion of the data set, so that the modified portions represent the abnormal tissue . Treatment also includes generating a model of electrical properties of tissues in the body area based on (a) positions of the plurality of tissue types in the deformed and modified model template and (b) abnormal tissue positions in the deformed template template and modified. The treatment also includes determining an electrode placement layout that maximizes the field strength in at least a portion of the abnormal tissue by using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed in a plurality of different sets of candidate positions for the body area, and selection of one of the sets. The treatment also includes producing the electrode placement layout determined for subsequent use as a guide for placing electrodes corresponding to the patient's body area prior to using the electrodes for electrotherapeutic treatment.
BRIEF DESCRIPTION OF THE DRAWINGS
Petition 870190054241, of 6/12/2019, p. 18/59
10/37 [021] FIG. 1 is a flowchart of a modality that performs electrotherapeutic treatment by creating a realistic model of a patient's head using a deformable template.
[022] FIG. 2 represents an original MRI image obtained from a patient with an abnormality (for example, a tumor).
[023] FIG. 3 illustrates the MRI image of FIG. 2 with the masked abnormality.
[024] FIG. 4 represents the normalization / registration process that generates the mapping and the reverse mapping between FIG. 3 and a deformable model of a healthy individual.
[025] FIG. 5 shows how the deformable template of FIG. 4 is deformed to match the shape of the patient's MRI image.
[026] FIG. 6 shows the implantation of the abnormality back to the deformed model.
[027] FIG. 7 represents a system for electrotherapeutic treatment according to a modality.
[028] FIG. 8 is another flowchart of a modality that performs electrotherapeutic treatment by creating a realistic model of a patient's head using a deformable template.
DETAILED DESCRIPTION OF THE PREFERENTIAL MODALITIES [029] The modalities described here generate a customized realistic head model for each individual patient, applying a non-rigid deformation to a preexisting realistic head model template, thus reducing the time and human work needed to create the head model. After the custom head model is generated for each individual patient, conventional simulation approaches are used to determine the ideal position for the transducer on the patient's body. Optionally, the head model template
Petition 870190054241, of 6/12/2019, p. 19/59
11/37 realistic background for the healthy patient can include tissue probability maps (TPMs). TPMs provide a model in which each point is represented by the respective probabilities of that point, belonging to various types of tissue, such as white matter, gray matter, CSF, etc.
[030] Optionally, patient images can be complemented with other magnetic resonance data, such as Diffusion Tensor Imaging (DTI) or Electric Impedance Water Tomography (Wept) data to obtain more accurate representations of conductivity in the head of the patient. patient, for example. E. Michel, D. Hernandez, and SY Lee, Electrical conductivity and permittivity maps of brain tissues derived from water content based on T 1 -weighted acquisition, Magnetic Resonance in Medicine, 2016. Magnetic resonance imaging techniques, such as DTI or Wept, provide information on tissue conductivity as disclosed, for example, in US Order No. 15 / 336,660, which is incorporated herein by reference in its entirety.
[031] The modalities of FIG. 1 and FIG. 8 describe workflows to create an individualized realistic head model for each patient with reduced user intervention and using these head models to optimize the Tumor Treatment Fields (TTFields) matrix layouts in patients. Once a realistic model has been built for any patient, optimization can be performed either fully automatically or semi-automatically using a sequence of algorithms that is also described here. Although these workflows are described in the context of TTFields, they can also be used in alternative contexts.
[032] The modalities of FIG. 1 and FIG. 8 start with a deformable model that is a realistic head model of a healthy individual (as opposed to a realistic head model of the real patient). This head model can be obtained
Petition 870190054241, of 6/12/2019, p. 20/59
12/37 using any conventional approach. For example, the realistic head model can be created in a standard coordinate system, such as the Montreal Neurological Institute (MNI) or Talairach spaces. For example, Holmes et. al., Journal of Computer Assisted Tomography, 22 324-333 (1998), which is incorporated by reference, teaches mapping and recording MRI images in the standard MNI space. If the model does not exist in a desired standard coordinate space, the transformation from a standard coordinate space to the head model is preferably known and can be used to map the model to the standard coordinate space. An example of a realistic head model built in a standard coordinate space is the model based on the COLIN27 data set (as described in Holmes et. Al., Journal of Computer Assisted Tomography, 22 324-333 (1998)) created by Miranda et. al. (as described in Miranda et. al., Physics in Medicine and Biology, 59, 4137-4147 (2014), which is incorporated herein by reference). But a wide variety of alternative models of realistic heads for the healthy individual can be used in place of Miranda's model. It is desirable that the magnetic resonances from which the model was created are also available for the purposes that will be described below.
[033] In some embodiments, the healthy individual's realistic head model feedback provides TPMs of tissue types. That is, each point in the model is represented by the respective probabilities of that point belonging to various types of tissue, such as white matter, gray matter, CSF, etc. In some embodiments, the healthy individual's realistic head model template provides one TPM per tissue type (for example, 6 TPMs for 6 types of white matter, gray matter, skull, scalp, CSF and air tissue).
[034] FIG. 1 describes a process 100 for using the realistic head model of the healthy individual to create a realistic head model for any patient, using the existing head model as a template
Petition 870190054241, of 6/12/2019, p. 21/59
Deformable 13/37.
[035] Process 100 starts at step S1, which is the acquisition of an appropriate set of MRI images. In step S1, an MRI data set for an individual patient is acquired using any conventional approach. This data set preferably includes magnetic resonances that carry structural data (such as that obtained from T1 or T2 MRI sequences). Optionally, additional sequences can also be purchased, such as DTI or perfusion imaging, which can carry additional information that can be useful for creating models, as will be described below. In some cases, MRI sequence parameters are optimized to increase the contrast between specific tissue types. Contrast enhancement is useful for segmenting the image following the steps described below, for example, as in the sequence described in Windhoff et. al., Human Brain Mapping, 34 923-935 (2013), which is incorporated herein by reference.
[036] Preferably, MRIs are acquired at the highest resolution that is practically possible. Generally, a resolution of better than 1 mm x 1 mm x 1 mm is desired. However, images with lower resolution can also be used.
[037] Optionally, diffusion-weighted magnetic resonance imaging (DWI) or DTI imaging data is also acquired. These data can be used to map the conductivity (or conductivity tensor) within each voxel as described in Wenger et. al., International Journal of Radiation Oncology Biology Physics, 941137-43 (2016), and Basser et. al., Biophysical Journal, 66 259267 (1994), which are incorporated herein by reference. In alternative modalities, different imaging modalities can be used instead of MRI images, such as CT images, a combination of MRI and CT images, etc.
[038] Process 100 continues at Step S2, which is image pre-processing. However, in some cases, no pre-processing is necessary and the
Petition 870190054241, of 6/12/2019, p. 22/59
14/37
Step S2 can be skipped. In step S2, image pre-processing is performed on the data obtained in step S1 to obtain a cleaner image. FIG. 2 shows an example of an MRI 200 image resulting after performing image preprocessing in step S2. Pre-processing can be implemented using any conventional approach. In some embodiments, the image pre-processing step includes image alignment and distortion correction. For example, image alignment can be implemented to remove artifacts due to the movement of images using any conventional approach. Realignment can be performed using the affine record, using any suitable conventional approach, such as Statistical Parametric Mapping (SPM), implemented in the SPM 8.0 toolbox, developed for the construction and evaluation of spatially extended statistical processes used to test hypotheses about functional imaging data. In addition, image distortion (for example, caused by induced currents) can be corrected at this stage. Image realignment is necessary when more than one data set is used to create the models, in which case these multiple data sets need to be aligned. For example, when sets of axial and coronal images are used for super-resolution, they need to be aligned. As another example, when DTI data is used in addition to Ti data, DTI data and Ti data may need to be aligned.
[039] In some modalities, an additional pre-processing step of handling the MRI header is performed (for example, in the Neuroimaging Computer Technologies Initiative (Nifti) format), so that the origin of the file coincides with the origin of the TPM model. The origin of the file refers to the origin of the axes in the file. This step helps to facilitate the registration of the MRI image in the deformable space, as described in step S4 below. In some modalities,
Petition 870190054241, of 6/12/2019, p. 23/59
15/37 the origin of the axes in the patient's MRI images and in the files associated with the deformable template is positioned in similar voxels to help facilitate the execution of step S4.
[040] Optionally, super-resolution algorithms that combine multiple sets of MRI data from a single patient into a single image can be used. These algorithms are useful for creating a data set that shows the patient's complete head when all other data sets truncate the head at different points or for creating a high resolution image (or slice spacing) when the original data is low resolution. High resolution data sets and data sets showing the complete 3D head are useful for creating an accurate head model. An example of a super-resolution algorithm is described in Woo, et al. Reconstruction of highresolution tongue volumes from MRI. IEEE Transactions on Biomedical Engineering, 59.12 (2012). This algorithm employed several pre-processing steps, including motion correction and intensity normalization, followed by a region-based maximal random Markov (MRF) approach to combine three volumes of orthogonal images from data sets of MRI in a single reconstruction of isotropic volume of super-resolution of the tongue. The output super-resolution image was superior to the input images in terms of signal to noise ratio (S - R) and resolution.
[041] In many cases, background noise and aliasing can be present and can deteriorate the quality of the head model created using deformable models. In particular, when background noise is present, the skull contour obtained during model creation is often inaccurate and includes part of the background. Consequently, some modalities may implement various thresholding schemes known to those skilled in the relevant techniques to remove background noise and aliasing. Aliasing, as mentioned here, refers to
Petition 870190054241, of 6/12/2019, p. 24/59
16/37 an artifact in MRI images that results in a faint shadow of the object being imaged to appear in the background (that is, the shadow is caused by aliasing). The shadow is typically upside down and directly connected to the main image. In this case, a thresholding scheme can be used to remove the faint shadow in the background. An example of a thresholding scheme that can be used to improve image quality is a semi-automatic method in which the user selects a single value representing background noise and the software applies that value as a threshold to automatically detect the leather contour hairy and reset the intensity of the background noise slice by slice. A wide variety of alternative approaches can also be used, as will be appreciated by people skilled in the relevant techniques.
[042] Alternatively or additionally, specific scanner pre-processing can be applied. For example, images can be converted from the Digital Imaging and Communications in Medicine (DICOM) format to NitTL [043] Process 100 continues in Step S3, which hides abnormal regions in the head. Step S3 is implemented only if there is a tumor or other abnormality (eg, skull defects / flaps) in the patient's MRI images. In step S3, these abnormal regions are masked as shown in image 300 of FIG. 3. Optionally, the regions that are masked can extend beyond the tumor / abnormality, if necessary, to include all regions where the normal brain structure has been significantly disturbed due to the presence of the tumor or other defects.
[044] One way to accomplish this masking step is to use supervised segmentation to properly mark the abnormal head regions. During this stage of supervised segmentation, multiple types of abnormalities are marked to achieve the desired level of detail in the final model,
Petition 870190054241, of 6/12/2019, p. 25/59
17/37 as will be described below. Supervised segmentation can be performed in a semi-automatic manner using, for example, tools such as ITK-SNAP (see, for example, Yushkevich et. Al, Neuroimage, 31 1116-1128 (2006), which is incorporated by reference).
[045] Alternatively, masking can be performed using automatic segmentation algorithms. For example, Porz, et al. Multi-modal glioblastoma segmentation: man versus machine. Public Library of Science (PLOS) One, 9.5 (2014), teaches a method for automatic segmentation of preoperative MRI images. In some situations, manual corrections to the results of the automatic segmentation process may be necessary to ensure accurate masking of the tumor.
[046] In some modalities, the regions that are masked are determined manually. One way to achieve this is to present the MRI data to a user, and ask the user to outline the tumor in the data. The data presented to the user may include structural MRI data (for example, Τι, T2 data). The different MRI modalities can be registered with each other, and the user can be presented with the option of viewing any of the data sets and delineating the tumor. The user may be asked to outline the tumor in a 3D volumetric representation of the MRI data, or the user may have the option to view individual 2D slices of the data and mark the tumor boundary on each slice. Once the limits have been marked on each slice, the tumor within the anatomical volume can be found. In this case, the volume marked by the user would correspond to the tumor. In some situations, margins of a predefined width (for example, 20 mm) are added to the tumor and the resulting volume is used as the region to be masked.
[047] Note that when there is no tumor or other abnormality in the patient's MRI images (for example, when the patient is healthy), step S3 is
Petition 870190054241, of 6/12/2019, p. 26/59
18/37 omitted.
[048] For some patients, segmentation results will reveal that the tumor is not homogeneous, in which case the tumor can also be segmented into several sub-regions, so that this segmentation information can be used to more precisely plant the tumor back to the realistic head model after the deformation step, as will be described in more detail here. Examples of such subregions are active / improving tumor, necrotic regions, resection cavity, etc. Conventional automatic segmentation algorithms can be used for detailed GBM segmentation. An example of a publicly available algorithm is the recent Brain Tumor Image Analysis (BraTumlA) software that distinguishes the necrotic nucleus, edema, the tumor that is not improving, and the improved tumor, requiring four different image modalities (Ti, Ti contrast, contrasted, and FLAIR). Techniques that only need a Ti as an input also exist. But, regardless of any variation within the tumor, all regions of the tumor are masked from the patient's original image. If the skull defects are in the image, these regions will be segmented and masked as well.
[049] Note that although a variety of approaches to identifying the abnormal region in the image are described above, a wide variety of alternative approaches will be evident to those skilled in the relevant techniques.
[050] Process 100 continues in step S4, which is Standardization / Spatial Registration. In step S4, a mapping that distorts the current set of MRI images for a given patient in the standard template model space is identified. FIG. 4 represents the normalization / registration process 400 that generates the mapping and inverse mapping between an MRI image of patient 402 (with a masking abnormality) and the deformable model template 404 of a
Petition 870190054241, of 6/12/2019, p. 27/59
19/37 healthy individual. The reverse of this mapping is also identified (for use in step S5 below to map from the standard space to the space of the patient's MRI set).
[051] For example, one approach to generating this mapping is to register the patient's MRI images in a standard coordinate space, such as the MNI space or the Talairach space. The image register refers to the spatial transformation of an image, so that certain features of the image align with the corresponding features in another image / space. This can be done by any known methods that will be evident to people versed in the relevant techniques, for example, using readily available software packages including, but not limited to, FSL FLIRT and SPM.
[052] Notably, abnormal regions masked in step S3 are omitted from the registration process. By ignoring masked regions during registration, it is ensured that the registration is performed using only healthy regions of the head, which can be effectively mapped to the model's TPMs that describe the probability that a specific voxel in the standard space belongs to a type of tissue specific. Advantageously, the omission of abnormal regions improves the robustness of the registration process. In some modalities, TPMs are built in the template template space.
[053] Alternatively, non-rigid registration algorithms (as described, for example, in Zhuang et. Al, IEEE Transactions on Medical Imaging, 30 0278-0062 (2011), which is incorporated by reference and teaches an algorithm for registration using mutual information) can be used to record the patient's MRI images in a standard coordinate space (for example, a realistic model template for a healthy individual) or in a voxelized version of the corresponding segmented model template. Note that a variety of algorithms for mapping patients' MRI images in a standard space is well
Petition 870190054241, of 6/12/2019, p. 28/59
20/37 known to people skilled in the relevant techniques. Moving in the opposite direction (that is, from the standard space for the patient's MRI images, as described below) we will use the reverse of these same mappings.
[054] The mappings described above are found for points on the patient's head that fall outside the masked areas. The transformations in the region (s) that were masked before registration can be estimated, for example, by questioning the deformation map found in the rest of the head in these regions, or by using any of a variety of alternative approaches that will be apparent to people skilled in the relevant techniques. In some modalities, it may not be necessary to find a transformation for the region (s) that were masked before registration. This is due to the fact that the deformable model template areas that correspond to the masked region contain information related to some natural structure (for example, healthy tissue). Therefore, after the mappings described above are applied to the deformable model template for points that are outside the masked regions, the deformed model template already includes some model data in those regions, as the natural structure is maintained in those regions. For example, if a sphere is masked from the left hemisphere in patient images and the mappings are applied to the deformable model template only for points that are outside the sphere, the content of the sphere in the left hemisphere of the deformed template template will resemble with some natural structure.
[055] In some modalities, the model's TPMs are used to find the mapping from the standard space to the patient space. In some embodiments, model TPMs can be derived from the MRI data set from which the deformable model was derived. The use of TPMs derived from this MRI data set can lead to a more accurate representation of the patient in the final model than with the use of other TPMs. The reason for this is as follows.
Petition 870190054241, of 6/12/2019, p. 29/59
21/37
TPMs describe the probability of a voxel in a standard space belonging to each type of tissue. TPMs are generally derived from multiple MRIs from different subjects. Thus, TPMs represent the probability that a voxel belongs to each type of tissue over a population of individuals. This implies that when performing the registration using TPMs derived from multiple individuals, the output mapping represents a mapping in some representative space that, by definition, smoothes the anatomical variation between the individuals from which the TPMs were derived. However, when creating patient models by deforming a healthy individual's head model, it may be desirable that the mapping calculated by recording the patient's MRI in the PMSs captures the anatomical characteristics of the healthy head model as accurately as possible. This precision ensures that when the deformable template is subsequently deformed in the patient's space in step S5 below, the resulting model resembles the patient as accurately as possible. Therefore, it is desirable that the TPMs from which registration in step S4 is performed represent the individual from whom the healthy head model was derived, as opposed to a population of individuals from whom TPMs are typically derived.
[056] One approach to creating TPMs that represent the healthy individual from which the deformable model template was derived is to simultaneously register and segment MRI images of the healthy individual using an existing set of generic TPMs (for example, TPMs built on standard space using data from multiple individuals). An example of an algorithm that does this is the Ashburner and Friston unified segmentation algorithm (Unified segmentation). Neuroimage 26.3 2005) which is implemented in SPM 8.0 toolbox described above. The outputs of this process include probability maps describing the probability that a voxel (from the MRI images recorded in the standard space) belongs to a specific type of tissue. The number of probability maps generated in that
Petition 870190054241, of 6/12/2019, p. 30/59
22/37 process is equal to the number of tissue types in the model (typically 6), and each voxel on a map is assigned a value from 0 to 1, which indicates the probability that the voxel belongs to a particular type of tissue. By definition, these probability maps are TPMs that represent the healthy individual from whom the healthy head model (deformable model) was derived.
[057] In some cases, manual corrections are made to the TPMs to obtain a better representation of the deformable model. For example, the skull and scalp probability maps can be modified to improve the limits of the skull or scalp. This can be done, for example, by manually assigning probability values to specific voxels, so that the probability of the voxel belonging to one type of tissue is close to 1, and the probability of belonging to other types of tissue is close to 0. A final step in creating TPMs from these probability maps is to apply a smoothing filter to the individual maps. Smoothing is important to allow adjustments to an MRI scan for any individual. The smoothing can be carried out, for example, with the use of a Gaussian filter with a smoothing core of 4 mm x 4 mm x 4 mm of FWHM (full width at half height).
[058] Process 100 continues in Step S5, which is deforming / distorting the model in the desired space, in step S5, the reverse mapping found in step S4 is applied to the deformable model template to map the deformable model template in the coordinates of the patient's MRI images. FIG. 5 represents the deformation / torsion process 500 that applies the inverse mapping to a deformable model template 502 to obtain the distorted model 504. In some embodiments, the inverse mapping applies a three-dimensional transformation to the deformable model template 502, thereby making that the 502 deformable model is adapted to the patient's specific anatomical attributes.
Petition 870190054241, of 6/12/2019, p. 31/59
23/37 [059] It should be noted that, prior to twisting, model 502 template is a brain model of a healthy reference individual; and after twisting, the distorted model 504 will represent an approximation of what the patient's brain would be like if it were healthy. In other words, this step results in a model of a healthy individual who has been twisted to fit the head shown in the patient's MRI images, but does not have a tumor. Notably, despite the fact that this distorted model originates from a model template (instead of each patient's head), it is still useful to analyze the electrical fields that can be induced inside each patient's head.
[060] The deformation in step S5 can be applied to a voxelized version of the model or to a mesh version of the model. In the voxelized version, each voxel indicates a type of tissue (or tissue type probabilities) at the location of the coordinates of that voxel. In the mesh version, each mesh defines a boundary between different types of fabrics, and deformation is applied to these meshes in the deformable model template. In some embodiments, a binary image of each type of fabric is created and each resulting binary image is deformed separately.
[061] Optionally, any holes that may appear in the deformed image of a type of fabric can be assigned to one of the types of fabric that appear in that image. An example of a procedure designed to assign fabric types to holes that appear between binary masks can be found in Timmons, et al. End-to-end workflow for finite element analysis of tumor treating fields in glioblastomas, Physics in Medicine & Biology, 62.21 (2017), when using ScanIP software, a Gaussian filter function smoothes the boundaries between masks to avoid convergence problems . The cavities in the mask are filled, and islands above a threshold (which may vary with the type of fabric) are removed. The current mask is duplicated and then expanded (from one to three voxels, depending on the fabric mask) and Boolean is added to the next mask on all slices.
Petition 870190054241, of 6/12/2019, p. 32/59
24/37
Any of a variety of alternative approaches to filling the holes that appear in the warped image can also be used.
[062] After the formation of images for each type of individual tissue, all binary images are combined into a single image representing a segmented image of the deformed head model.
[063] In cases where a voxel in the combined model is assigned to more than one type of tissue, a heuristic logic can be used to determine the type of tissue in the final image. For example, logic can claim that all voxels in which gray and white matter overlap in the combined model are attributed only to white matter, or vice versa.
[064] In modalities where the model template includes TPMs (that is, each fabric in the model template is represented by a 3D matrix that describes the probability that each voxel belongs to a specific type of fabric), the TPMs are deformed and the deformed TPMs are combined into a final model so that each voxel in the combined model is assigned a tissue type based on some heuristic logic. For example, each voxel is assigned to the type of tissue that is most likely to occupy that voxel.
[065] In some modalities, the probability attributed by different TPMs to each voxel is used to determine the combination of the conductivity properties in the created voxelized model. In other words, it is assumed that the voxel does not necessarily contain a certain type of tissue, and the final conductivity is attributed to the voxel as a weighted sum of the conductivities of all types of tissue, with the weights derived from the probability values assigned to each type of tissue in that voxel.
[066] In some modalities, conductivity values are assigned to tissue maps, additionally incorporating information obtained from MRI imaging techniques, such as DTI or Wept, which are known to provide
Petition 870190054241, of 6/12/2019, p. 33/59
25/37 information on tissue conductivity as disclosed, for example, in US Order No. 15 / 336.660 (published as LIS2017 / 0120041), which is incorporated herein by reference in its entirety. This information can be incorporated into the model, for example, by assigning conductivity to each voxel based on the weighted average of the conductivity derived from the model and the conductivity derived from Wept / DTI.
[067] Process 100 continues at Step S6, which is planting the abnormality back into the deformed template. In step S6, the deformed template is edited so that each voxel in the template that corresponds to the masked region found in step S3 is assigned to an abnormal tissue type (for example, the tumor or the surrounding region). FIG. 6 represents this process 600 where an abnormality identified in the patient's image 602 is implanted in a deformed model template 604. In some modalities, planting is performed by assigning tissue types in each of the abnormal regions according to the segmentation performed in the step S3. More specifically, the type of tissue assigned to each point in the abnormal region after deformation is based on the type of tissue identified for a corresponding point in segmentation in step S3 before deformation. Therefore, if the segmentation in step S3 identifies more than one type of tissue in the abnormal region, then there may be more than one type of tissue assigned to the abnormal region after deformation. In alternative modalities, planting can be done by assigning a standard abnormal tissue type to the abnormal region after deformation. In other alternative modalities, planting can be done by having a user manually assign a type of tissue to the spots in the abnormal region.
[068] Process 100 continues at Step S7, which is the creation of the model. In the modeling step (S7), electrical properties, such as conductivity and permittivity, are attributed to the various types of fabrics. Note that the types of fabric are usually obtained from the deformed template. However, a type of
Petition 870190054241, of 6/12/2019, p. 34/59
26/37 tissue corresponding to the tumor tissue will be assigned to each voxel that corresponds to the implanted abnormality. Electrode models (or transducer arrays) are placed on the model's skin and suitable boundary conditions are applied. In some modalities, the modeling step S7 assumes that each type of tissue is homogeneous and, therefore, a single value for the electrical property is assigned to each type of tissue (as described, for example, in Miranda et. Al., Physics in Medicine and Biology, 59, 4137-4147 (2014), Wenger et. al., Physics in Medicine and Biology, 60 7339-7357 (2015), and Wenger et. al., International Journal of Radiation Oncology Biology Physics, 941137 -43 (2016)). In other models, the conductivity in each voxel is attributed based on DTI or DWI images acquired during the image acquisition stage. The DTI assigns anisotropic electrical properties (a 3x3 tensor) to each voxel, while the DWI assigns an isotropic conductivity (a scalar) to each voxel. Finally, the model is divided into volume elements, for example, by voxelization or, alternatively, by volume meshing.
[069] Process 100 continues at Step S8. After the head model is created and the model electrodes have been added to the head model, a simulation is performed in step S8. This simulation finds an optimal electrode matrix layout by solving the corresponding induced electric field using an appropriate numerical technique including but not limited to finite element methods or finite difference methods.
[070] Optimizing the electrode matrix layouts means finding the matrix layout that optimizes the electric field within the sick regions of the patient's brain (tumor). This optimization can be implemented on the volume targeted to treat (the target volume) within the realistic head model, automatically place transducer matrices and define boundary conditions in the realistic head model; calculate the electric field that develops within the realistic head model, once the matrices were placed in the head model
Petition 870190054241, of 6/12/2019, p. 35/59
27/37 realistic and in the boundary conditions applied; and execute an optimization algorithm to find the layout that produces ideal electric field distributions within the target volume. Although a variety of alternative approaches can be used, an example for implementing these four steps is provided below.
[071] The position and orientation of the matrices in the realistic head model can be calculated automatically for a given iteration. Each transducer array used to deliver TTFields to the Optune ™ device comprises a set of ceramic disk electrodes that are attached to the patient's head through a layer of medical gel. When placing matrices on real patients, the disks naturally line up parallel to the skin, and good electrical contact between the matrices and the skin occurs because the medical gel deforms to match the contours of the body. However, virtual models are made of rigidly defined geometries. Therefore, placing the matrices on the model requires a precise method to find the orientation and contour of the model's surface in the positions where the matrices are to be placed, as well as to find the thickness / geometry of the gel that is necessary to ensure good contact of the model matrices with the realistic model of the patient. To allow for fully automated optimization of field distributions, these calculations need to be performed automatically.
[072] A variety of algorithms to perform this task can be used. The steps of such an algorithm designed for this purpose are presented below.
The. Define the position where the center point of the transducer array will be placed on the model head. The position can be defined by a user or as one of the steps in the field optimization algorithm.
B. Use the input from step (a) together with knowledge about the geometry of the disks and how the disks are organized in the matrix, calculate the
Petition 870190054241, of 6/12/2019, p. 36/59
28/37 approximate positions of the centers of all discs in the transducer matrix within the model.
ç. Calculate the surface orientations of the realistic model in the positions where the discs will be placed. The calculation is performed by locating all the points on the fictional computational skin that are within a disk radius of the designated center of the disk. The coordinates of these points are organized in the columns of a matrix and the decomposition of the singular value is performed in the matrix. The normal for the model's skin is then the eigenvector that corresponds to the smallest eigenvalue found.
d. For each disc in the transducer matrix: calculate the thickness of the medical gel needed to ensure good contact between the discs and the patient's body. This is done by finding the parameters for a cylinder with its height oriented parallel to the normal skin surface. The cylinder is defined with a radius equal to the radius of the discs, and its height is adjusted to extend a predetermined amount (this is a predetermined constant) beyond the points on the skin used to find the normal one. This results in a cylinder that extends at least the predetermined amount out of the dummy surface.
and. In the model, create the cylinders described in (d).
f. Using binary logical operations (for example, subtracting the cylinder head), remove from the model the regions of the cylinder that project onto the realistic model of the patient. The resulting truncated cylinders represent the medical gel associated with the transducer arrays.
g. On the outer side of the truncated cylinders, place the disks that represent the ceramic disks of the transducer arrays.
[073] Then, the distribution of the electric field is calculated within the head model for the given iteration. Once the fictitious head is constructed and the transducer arrays (ie, electrode arrays) that will be used to apply the
Petition 870190054241, of 6/12/2019, p. 37/59
29/37 fields are placed in the realistic head model, a volume mesh, suitable for finite element method analysis, can be created. Then, boundary conditions can be applied to the model. Examples of boundary conditions that can be used include Dirichlet boundary conditions (constant voltage) in the transducer arrays, Neumann boundary conditions in the transducer matrices (constant current) or potential floating boundary condition that defines the potential at this limit. so that the integral of the normal component of the current density is equal to a specified amplitude. The model can then be solved with a suitable finite element solver (for example, a low-frequency quasi-static electromagnetic solver) or, alternatively, with finite difference algorithms. Meshing, the imposition of boundary conditions and model resolution can be performed with existing software packages, such as Sim4Life, Comsol Multiphysics, Ansys or Matlab. Alternatively, the custom computer code that performs the finite element algorithms (or finite differences) could be written. This code can use existing software resources, such as C-Gal (for creating meshes), or FREEFEM ++ (software written in C ++ for quick tests and finite element simulations). The final solution of the model will be a data set that describes the distribution of the electric field or related quantities, such as electrical potential within the computational spectrum for the given iteration. In some embodiments, the model is based on voxel (that is, it comprises volume elements in the form of a box). In these modalities, the Finite Difference Time Domain (FDTD) algorithms can be used to solve the model, for example, using the quasi-electrostatic solver associated with the Sim4Life software package from ZMT Zurich MedTech AG.
[074] Then, an optimization algorithm is used to find the matrix layout that optimizes the supply of the electric field to the sick regions of the
Petition 870190054241, of 6/12/2019, p. 38/59
30/37 patient's brain (tumor) for both directions of application (LR and AP). The optimization algorithm will use the method for automatic matrix positioning and the method to solve the electric field within the head model in a well-defined sequence, in order to find the ideal matrix layout. The optimal layout will be the layout that maximizes or minimizes some target function of the electric field in the diseased regions of the brain, considering both directions in which the electric field is applied. This target function can be, for example, the maximum intensity within the diseased region or the average intensity within the diseased region. It is also possible to define other target functions.
[075] There are several approaches that can be used to find the ideal matrix layouts for patients, three of which are described below. An optimization approach is exhaustive research. In this approach, the optimizer will include a bank with a finite number of matrix layouts that must be tested. The optimizer runs simulations of all matrix layouts in the bank and selects the matrix layouts that produce the ideal field strengths in the tumor (the ideal layout is the bank layout that produces the highest (or lowest) value for the target function of optimization, for example, the strength of the electric field applied to the tumor).
[076] Another optimization approach is an iterative search. This approach encompasses the use of algorithms, such as minimal offspring optimization methods and simplex search optimization. Using this approach, the algorithm iteratively tests different array layouts in the head and calculates the target function of the electric field in the tumor for each layout. In each iteration, the algorithm automatically chooses the configuration to test based on the results of the previous iteration. The algorithm is designed to converge in a way that maximizes (or minimizes) the target function defined for the field in the tumor.
[077] Yet another optimization approach is based on placing a dipole in the center of the tumor in the model. This approach differs from the other two
Petition 870190054241, of 6/12/2019, p. 39/59
31/37 approaches, as it does not depend on the resolution of the field strength for different matrix layouts. Instead, the ideal position for the matrices is found by placing a dipole aligned with the direction of the expected field at the center of the tumor in the model and solving the electromagnetic potential. The regions on the scalp where the electric potential (or possibly the electric field) is maximum will be the positions where the matrices are placed. The logic of this method is that the dipole will generate an electric field that is maximum at the center of the tumor. By reciprocity, if we could generate the field / tension in the scalp that the calculation produced, then we would expect to obtain a field distribution that was maximum in the center of the tumor (where the dipole was placed). The closest we can get to this with our current system is to place the matrices in the regions where the potential induced by the dipole in the scalp is maximum.
[078] Note that alternative optimization schemes can be used to find a matrix layout that optimizes the electric field within the diseased regions of the brain. For example, the algorithms that combine the various approaches mentioned above. As an example of how these approaches can be combined, consider an algorithm combining the third approach discussed above (that is, placing the dipole at the center of the tumor in the model) with the second approach (ie, the iterative search). With this combination, a matrix layout is found initially using the dipole at the center of the tumor approach. This matrix layout is used as input for an iterative search that finds the ideal layout.
[079] Once the layout that optimizes the electric field within the patient's diseased brain regions has been determined (for example, using any of the approaches explained here, or with an appropriate alternative approach), the electrodes are positioned in certain positions. AC voltages are then applied to the electrodes (for example, as described in US Patent
Petition 870190054241, of 6/12/2019, p. 40/59
32/37
7,565,205, which is incorporated by reference) to treat the disease.
[080] FIG. 7 represents an example of system 700 for electrotherapeutic treatment that can be used after the electrode positions have been optimized as described herein. System 700 includes a controller 702 that applies TTFields to a patient, applying voltages to arrays of capacitively coupled transducers 42, 44 that are affixed to the patient's scalp 40 in the specified positions. Note that the front view of scalp 40 is shown in FIG. 7 and only three of the four electrode sections are visible in the figure and neither the eyes nor the ears are represented.
[081] Optionally, the system can be designed to work with several model templates. In this case, an additional step S3.5 is implemented after step S3 and before step S4. In step S3.5, the similarity of the patient's MRI images to each of a plurality of templates is first measured (using, for example, a correlation or mutual information measure). The deformable template that most closely resembles the patient's MRI images is selected and used in all subsequent steps. Alternatively, in some modalities, the selection of the deformable template that most closely resembles the patient's MRI images can be performed after registering the patient's images in a standard space in step S4 and before step S5. In these modalities, the deformable template that most closely resembles the patient's MRI images is used in all steps subsequent to S4.
[082] Optionally, the system can be configured as a learning system in which each realistic head model created using the process described above serves as a deformable model for future models. Both the deformed healthy model created in step S5 and the resulting model that includes defects (created in step S6) could be added to the database. If a patient's MRI images in the original image stack resemble a
Petition 870190054241, of 6/12/2019, p. 41/59
33/37 stored template of a brain with a tumor in a degree close enough, then it is possible to create a model that represents the patient's MRI images by measuring the deformations in the previously stored template.
[083] Finally, although the concepts presented here are discussed in the context of a magnetic resonance image (MRI) of a patient's head, the same principles can be applied to other parts of a patient's body and / or the imaging modalities other than MRI.
[084] FIG. 8 is a flow chart 800 of a method for optimizing the position of electrodes that will subsequently be used to perform electrotherapeutic treatment by creating a realistic model of a patient's head using a deformable template. Electrotherapeutic treatment can be TDCS, TMS or TTFields.
[085] In S10 one or more 3D images of a patient's body area are received. The 3D images can be MRI images, CT images or images in any other modalities known in the art. The body area can be the patient's head or any other area of the body. Optionally, images can be pre-processed using any of the approaches described here (for example, as described here with reference to step S2 of FIG. 1).
[086] In S20, portions of the image that correspond to the abnormal tissue are identified. For example, when the body area is a patient's head, these portions may correspond to a tumor or abnormality in the skull. The abnormality can be identified manually, automatically or semi-automatically, according to any of the methods described here, or according to any other appropriate methods that will be evident to those skilled in the relevant techniques. In some embodiments, portions of the image that correspond to the abnormal tissue are identified by segmentation of the image.
Petition 870190054241, of 6/12/2019, p. 42/59
34/37 [087] In S30, a data set is generated to match the image with the masked abnormal tissue. This can be achieved, for example, by masking the abnormal tissue which includes ignoring the abnormal regions in the registration process described in S50 below. In some modalities, the abnormal region masking is implemented by marking the data points in that region and excluding all data points flagged during the registration process described in S50 below.
[088] In S40, a model template that specifies the positions of a plurality of tissue types in a healthy version of the patient's body area is retrieved. For example, when the body area is a patient's head and the abnormal tissue corresponds to a tumor in the patient's head, the model template corresponds to the head of a healthy individual and has no tumor. In some embodiments, the template template can be selected from several existing template templates based on the similarities between the image and each of the various template templates. For example, a measure of similarity, such as mutual information or distance, can be determined between the patient's data set (derived from masking abnormalities in the patient's image) and each of the various model templates and the template template which is most similar to the patient's data set (for example, has the shortest distance or the most mutual information) can be selected accordingly. In some embodiments, the model template may include TPMs, and the TPMs may correspond to the same healthy individual from whom the model was derived (and derived from images of the healthy individual) or to multiple individuals.
[089] In S50, the model template is deformed in space, so that the resources in the deformed model template align with the corresponding resources in the data set. In some modalities, the model template is deformed by determining a mapping that maps the set
Petition 870190054241, of 6/12/2019, p. 43/59
35/37 of data for a model template coordinate space; and apply an inverse of the mapping to the model template. In some modalities, the mapping can be determined by registering the data set in a model template coordinate space. That is, the mapping distorts the data set in the model template. Thus, the inverse of the mapping distorts the model template for the data set and thus provides a realistic model for the patient if the patient has no abnormalities. In some embodiments, the mapping of the data set to the model template is determined for points in the data set that are outside the masked portion. In modalities where the model template includes TPMS, the mapping maps the data set to the TPMs and the inverse of the mapping is applied to each of the TPMs and the TPMs inversely mapped are combined into a segmented image that comprises the model template deformed.
[090] In S60, the portions of the deformed template template that correspond to the masked portion of the data set are modified so that the modified portions represent the abnormal tissue. The modification can be carried out according to the information obtained during the identification of the abnormal portions in S20. For example, one or more abnormal tissue types identified in S20 can be assigned to corresponding portions in the deformed model template. Alternatively, a predetermined generic tissue type can be assigned to the masked portion.
[091] In S70, a model of electrical properties of tissues in the body area is generated based on: (a) positions of the plurality of tissue types in the deformed and modified model template and (b) abnormal tissue positions in the template deformed and modified model. The electrical properties can be electrical conductivity, electrical resistivity or any other electrical property pertinent to the electrotherapeutic treatment of the body area. In some
Petition 870190054241, of 6/12/2019, p. 44/59
36/37 modalities, for example, a different electrical property value can be assigned to each type of fabric according to a pre-filled consultation table.
[092] In S80, an electrode placement layout that maximizes field strength in at least a portion of the abnormal tissue is determined using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed in a plurality of different sets of candidate positions corresponding to the body area and selecting one of the sets. In some embodiments, the electrode placement layout is determined by applying a boundary condition to the simulated electrodes in each of at least two electrode placement layouts; resolution of a field distribution in the body area for each of the at least two electrode placement layouts; and choosing the electrode placement layouts that produce the strongest field within the abnormal region. The boundary condition can correspond, for example, to the voltages applied to the simulated electrodes. In some modalities, the field distribution is solved using a numerical technique, such as a finite element method or a finite difference method.
[093] In S90, the determined electrode placement layout is produced for subsequent use as a guide for placing electrodes relative to the patient's body area before using electrodes for electrotherapeutic treatment (eg TTFields).
[094] Models constructed in this way could also be used for other applications in which the calculation of distributions of electric field and / or electric current inside the head can be useful. These applications include, but are not limited to: direct and alternating current trans-cranial stimulation; simulations of field maps of implanted stimulator electrodes; placement planning
Petition 870190054241, of 6/12/2019, p. 45/59
37/37 implanted electrode stimulators; and location of the source on the electroencephalogram (EEG).
[095] Finally, although this patent application describes methods for optimizing matrix layouts on the head, the same steps can be used to optimize matrix layouts in other regions of the body (including, but not limited to, chest or abdomen ).
[096] Although the present invention has been disclosed with reference to certain modalities, numerous modifications, alterations and changes in the described modalities are possible without departing from the scope and scope of the present invention, as defined in the appended claims. Consequently, it is intended that the present invention is not limited to the described modalities, but that it has the full scope defined by the language of the following claims, and their equivalents.

权利要求:
Claims (22)
[1]
1. Method to improve the treatment of a tumor using Tumor Treatment Fields (TTFields), the method FEATURED by the fact that it comprises:
receiving, through a computer system processor, a three-dimensional image of a patient's body area;
identify portions of the image that correspond to the abnormal tissue;
generate a data set corresponding to the image with the abnormal tissue masked; retrieving a model template from a computer system memory device, the model template comprising tissue probability maps that specify positions of a plurality of tissue types in a healthy version of the patient's body area;
deform the model template in space so that the features in the deformed model template align with the corresponding features in the data set;
modify the portions of the deformed model template that correspond to the masked portion of the data set so that the modified portions represent the abnormal tissue;
generate a model of electrical properties of tissues in the body area based on (a) positions of the plurality of tissue types in the deformed and modified model template and (b) abnormal tissue positions in the deformed and modified template template;
determine an electrode placement layout that maximizes field strength in at least a portion of the abnormal tissue by using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed in a plurality of different sets of candidate positions corresponding to the body area, and selection
Petition 870190054241, of 6/12/2019, p. 47/59
[2]
2/6 of one of the sets;
place the respective electrodes in the patient's body area based on the determined electrode placement layout; and use the electrodes placed to apply TTFields to the body area.
2. Method to improve an electrotherapeutic treatment, CHARACTERIZED by the fact that it comprises:
receiving, through a computer system processor, a three-dimensional image of a patient's body area;
identify portions of the image that correspond to the abnormal tissue;
generate a data set corresponding to the image with the abnormal tissue masked; retrieving a model template from a computer system memory device, wherein the model template specifies positions of a plurality of tissue types in a healthy version of the patient's body area;
deform the model template in space so that the features in the deformed model template align with the corresponding features in the data set;
modify the portions of the deformed model template that correspond to the masked portion of the data set so that the modified portions represent the abnormal tissue;
generate a model of electrical properties of tissues in the body area based on (a) positions of the plurality of tissue types in the deformed and modified model template and (b) abnormal tissue positions in the deformed and modified template template;
determine an electrode placement layout that maximizes field strength in at least a portion of the abnormal tissue by using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed on a plurality of
Petition 870190054241, of 6/12/2019, p. 48/59
[3]
3/6 different sets of candidate positions for the body area, and selection of one of the sets; and produce the electrode placement layout determined for subsequent use as a guide for placing the respective electrode to the patient's body area prior to using the electrodes for electrotherapeutic treatment.
3. Method, according to claim 2, CHARACTERIZED by the fact that the deformation of the model template comprises: determining a mapping that maps the data set to a coordinate space of the model template; and apply an inverse of the mapping to the model template.
[4]
4. Method, according to claim 3, CHARACTERIZED by the fact that the mapping is determined for points in the data set that are located outside the masked portion.
[5]
5. Method, according to claim 3, CHARACTERIZED by the fact that the model template comprises tissue probability maps, in which the mapping maps the data set to the tissue probability maps.
[6]
6. Method, according to claim 5, CHARACTERIZED by the fact that tissue probability maps are derived from images of a healthy individual from whom the model template was derived.
[7]
7. Method, according to claim 6, CHARACTERIZED by the fact that tissue probability maps are derived by simultaneously registering and segmenting images of the healthy individual using existing tissue probability maps, and in which Existing tissue probabilities are derived from images of multiple individuals.
[8]
8. Method according to claim 5, CHARACTERIZED by the fact that tissue probability maps are tissue probability maps derived from images of multiple individuals.
Petition 870190054241, of 6/12/2019, p. 49/59
4/6
[9]
9. Method, according to claim 5, CHARACTERIZED by the fact that the inverse of the mapping is applied to each of the tissue probability maps, in which the tissue probability maps inversely mapped are combined into a segmented image comprising the deformed template template.
[10]
10. Method according to claim 9, CHARACTERIZED by the fact that the combination of the tissue probability maps inversely mapped includes assigning to each voxel the type of tissue that is most likely to occupy that voxel across the probability maps inversely mapped tissue.
[11]
11. Method, according to claim 9, CHARACTERIZED by the fact that the combination of the tissue probability maps inversely mapped includes the use of a look-up table to assign a type of tissue to each voxel that is associated with more than one tissue type through inversely mapped tissue probability maps.
[12]
12. Method, according to claim 2, CHARACTERIZED by the fact that the identification of the portions of the image that correspond to the abnormal tissue comprises the segmentation of the image.
[13]
13. Method, according to claim 2, CHARACTERIZED by the fact that the model of electrical properties of tissues comprises a model of electrical conductivity or resistivity.
[14]
14. Method according to claim 2, CHARACTERIZED by the fact that the image comprises an MRI image or a CT image.
[15]
15. Method, according to claim 2, CHARACTERIZED by the fact that the body area comprises the patient's head.
[16]
16. Method, according to claim 2, CHARACTERIZED by the fact that the portions of the image that correspond to the abnormal tissue correspond to a tumor.
Petition 870190054241, of 6/12/2019, p. 50/59
5/6
[17]
17. Method, according to claim 2, CHARACTERIZED by the fact that the electrotherapeutic treatment comprises Tumor Treatment Fields (TTFields).
[18]
18. Method, according to claim 2, CHARACTERIZED by the fact that the determination of the electrode placement layout comprises:
apply a boundary condition to the simulated electrodes in each of the at least two electrode placement layouts;
resolve a field distribution in the body area for each of the at least two electrode placement layouts; and choose the electrode placement layout that produces the strongest field within the abnormal region.
[19]
19. Method, according to claim 18, CHARACTERIZED by the fact that the boundary condition corresponds to voltages or currents applied to the simulated electrodes.
[20]
20. Method, according to claim 2, CHARACTERIZED by the fact that the template template is selected from a plurality of template templates based on the similarities between the image and each of the template templates.
[21]
21. Method, according to claim 2, CHARACTERIZED by the fact that it further comprises:
place the respective electrodes in the patient's body area based on the determined electrode placement layout; and use the electrodes to apply the TTFields to the body area.
[22]
22. Electrotherapeutic treatment device, CHARACTERIZED by the fact that it comprises a processor configured to execute instructions stored in one or more memory devices to perform an electrotherapeutic treatment, which comprises:
Petition 870190054241, of 6/12/2019, p. 51/59
6/6 receive, by the processor, a three-dimensional image of an area of a patient's body; identifying the portions of the image that correspond to the abnormal tissue;
generate a data set corresponding to the image with the abnormal tissue masked; retrieving a model template from one or more memory devices, where the model template specifies positions of a plurality of tissue types in a healthy version of the patient's body area;
deform the model template in space so that the features in the deformed model template align with the corresponding features in the data set; modifying the portions of the deformed template template that correspond to the masked portion of the data set so that the modified portions represent the abnormal tissue;
generate a model of electrical properties of tissues in the body area based on (a) positions of the plurality of tissue types in the deformed and modified model template and (b) abnormal tissue positions in the deformed and modified template template;
determine an electrode placement layout that maximizes field strength in at least a portion of the abnormal tissue by using the electrical properties model to simulate electromagnetic field distributions in the body area caused by simulated electrodes placed in a plurality of different sets of candidate positions for the body area, and selection of one of the sets; and to produce the electrode placement layout determined for subsequent use as a guide for placing the respective electrode to the patient's body area before using the electrodes for electrotherapeutic treatment.

类似技术:

---

公开号 | 公开日 | 专利标题

---

US11109773B2|2021-09-07|Treating patients with TTFields with the electrode positions optimized using deformable templates

---

JP2021074585A|2021-05-20|Ttfield treatment with optimization of electrode positions on the head based on mri-based conductivity measurements

---

Acar et al.2010|Neuroelectromagnetic forward head modeling toolbox

---

Horn et al.2016|Toward a standardized structural–functional group connectome in MNI space

---

US20190308016A1|2019-10-10|TTField Treatment with Optimization of Electrode Positions Based on Low Frequency | AC Conductivity Estimates Derived From Two MRI Images Having Different Repetition Times

---

US9508144B2|2016-11-29|Determining an anatomical atlas

---

US20200023179A1|2020-01-23|Using Power Loss Density and Related Measures to Quantify the Dose of Tumor Treating Fields |

---

Puonti et al.2020|Accurate and robust whole-head segmentation from magnetic resonance images for individualized head modeling

---

Farcito et al.2019|Accurate anatomical head segmentations: a data set for biomedical simulations

---

US20200146586A1|2020-05-14|Creating Accurate Computational Head Models of Patients Using Datasets Combining MRI and CT Images

---

Chen et al.2014|Development of an open source software module for enhanced visualization during MR-guided interstitial gynecologic brachytherapy

---

Lau et al.2011|Subject specific modelling of electrical conduction in the body: A case study

---

Doyle et al.2015|Patient-Specific Biomechanical Model as Whole-Body CT Image Registration Tool

同族专利:

---

公开号 | 公开日

---

US20180160933A1|2018-06-14|

---

CN110234394A|2019-09-13|

---

US11109773B2|2021-09-07|

---

WO2018109691A2|2018-06-21|

---

AU2017377003A1|2019-07-18|

---

JP2020501689A|2020-01-23|

---

JP2022031908A|2022-02-22|

---

WO2018109691A3|2018-10-25|

---

KR20190104036A|2019-09-05|

---

EP3554631A2|2019-10-23|

---

CA3047067A1|2018-06-21|

引用文献:

---

公开号 | 申请日 | 公开日 | 申请人 | 专利标题

---

---

US20050202384A1|2001-04-20|2005-09-15|Medtronic, Inc.|Interactive computer model of the heart|

---

JP4750784B2|2004-04-23|2011-08-17|ノヴォキュアー・リミテッド|Treatment of tumors by electric fields of different frequencies|

---

US20160055304A1|2005-06-16|2016-02-25|Aaken Laboratories|Targeted electrical stimulation|

---

US9307925B2|2005-06-16|2016-04-12|Aaken Laboratories|Methods and systems for generating electrical property maps of biological structures|

---

WO2009089341A1|2008-01-09|2009-07-16|Washington University In St. Louis|Method for quantitatively mapping myocardial contractile function with magnetic resonance based multiparametric strain analysis|

---

US10117707B2|2008-04-29|2018-11-06|Virginia Tech Intellectual Properties, Inc.|System and method for estimating tissue heating of a target ablation zone for electrical-energy based therapies|

---

EP2419171B1|2009-04-13|2017-11-01|Research Foundation Of The City University Of New York|Neurocranial electrostimulation models|

---

US9414881B2|2012-02-08|2016-08-16|Angiodynamics, Inc.|System and method for increasing a target zone for electrical ablation|

---

US10296809B2|2012-02-28|2019-05-21|Siemens Healthcare Gmbh|Method and system for fast patient-specific cardiac electrophysiology simulations for therapy planning and guidance|

---

EP2863986A4|2012-06-22|2016-01-20|Thync Inc|Device and methods for noninvasive neuromodulation using targeted transcrannial electrical stimulation|

---

GB201307590D0|2013-04-26|2013-06-12|St Georges Hosp Medical School|Processing imaging data to obtain tissue type information|

---

US10188851B2|2015-10-28|2019-01-29|Novocure Limited|TTField treatment with optimization of electrode positions on the head based on MRI-based conductivity measurements|US10821283B2|2016-04-04|2020-11-03|Novocure Gmbh|Reducing motility of cancer cells using tumor treating fields |

---

US10902621B2|2018-03-25|2021-01-26|Varian Medical Systems International Ag|Deformable image registration based on masked computed tomographyimage|

---

US10953209B2|2018-03-28|2021-03-23|Board Of Regents Of The University Of Texas System|Treating tumors using TTFields combined with a PARP inhibitor|

---

EP3900775A4|2018-07-03|2021-10-27|Edwin Chang|Using alternating electric fields to increase cell membrane permeability|

---

US11179322B2|2018-07-10|2021-11-23|Novocure Gmbh|Methods and compositions for treating tumors with TTFields and sorafenib|

---

EP3967362A1|2018-07-18|2022-03-16|Novocure GmbH|Using power loss density and related measures to quantify the dose of tumor treating fields|

---

DK3773291T3|2018-08-23|2021-11-15|Novocure Gmbh|USE OF ALTERNATIVE ELECTRIC FIELDS TO INCREASE THE PERMEABILITY OF THE BLOOD-BRAIN BARRIER|

---

US11160977B2|2018-09-04|2021-11-02|Novocure Gmbh|Delivering tumor treating fieldsto the infratentorial brain|

---

CN112770806A|2018-09-07|2021-05-07|诺沃库勒有限责任公司|Treatment of autoimmune diseases using alternating electric fields to reduce proliferation of T cells|

---

US11154707B2|2018-10-15|2021-10-26|Novocure Gmbh|Generating tumor treating fieldswith high uniformity throughout the brain|

---

CN109102894B|2018-10-18|2021-09-07|中国医学科学院生物医学工程研究所|Method and device for modeling anisotropic conductivity head model by combining cortex excitability|

---

WO2020084596A1|2018-10-25|2020-04-30|Zeev Bomzon|Delivering alternating electric fieldsto a subject's spinal anatomy|

---

KR20210099609A|2018-11-29|2021-08-12|노보큐어 게엠베하|Enhanced Flexibility Transducer Arrays Delivering Tumor Treatment Plants|

---

WO2020236956A1|2019-05-20|2020-11-26|Neuroelectrics Corporation|Systems and methods for treating tumors using targeted neurostimulation|

---

WO2021092402A1|2019-11-08|2021-05-14|Novocure Gmbh|Perforated hydrogel configurations and methods of production and use thereof|

---

US20210162228A1|2019-12-02|2021-06-03|Novocure Gmbh|Methods, systems, and apparatuses for optimizing transducer array placement|

---

WO2021136971A1|2019-12-31|2021-07-08|Novocure Gmbh|Methods, systems, and apparatuses for fast approximation of electric field distribution|

---

WO2021150877A1|2020-01-22|2021-07-29|Novocure Gmbh|Assemblies containing two conductive gel compositions and methods of production and use thereof|

---

US20210299439A1|2020-03-31|2021-09-30|Novocure Gmb|Methods, systems, and apparatuses for guiding transducer placements for tumor treating fields|

---

US20210346693A1|2020-05-06|2021-11-11|Novocure Gmbh|Conductive pad generating tumor treating field and methods of production and use thereof|

---

KR20210141103A|2020-05-15|2021-11-23|뉴로핏 주식회사|Apparatus and method for providing posion of brain stimulation|

---

KR20210142299A|2020-05-18|2021-11-25|주식회사 필드큐어|Apparatus and Method for Quality Assurance in Electric Field Therapy|

---

KR102321009B1|2021-07-06|2021-11-03|뉴로핏 주식회사|Method, server and computer program for designing customized headgear for transcranial direct current stimulation|

---

CN113827866B|2021-11-25|2022-02-18|北京航空航天大学杭州创新研究院|Non-invasive tumor treatment device based on alternating current electric field|

法律状态:
2021-10-13| B350| Update of information on the portal [chapter 15.35 patent gazette]|

优先权:

---

申请号 | 申请日 | 专利标题

---

US201662433501P| true| 2016-12-13|2016-12-13|

---

US62/433.501|2016-12-13|

---

PCT/IB2017/057901|WO2018109691A2|2016-12-13|2017-12-13|Treating patients with ttfields with the electrode positions optimized using deformable templates|

---