Method, system and apparatus for medical image enhancement to optimize transducer array placement

Abstract

A method, system, and apparatus for medical image enhancement to optimize transducer array placement. A computer-implemented method to generate a three-dimensional model, where the computer includes one or more processors and a memory accessible by the one or more processors, and the memory stores instructions that, when executed by the one or more processors, generate a three-dimensional model. The instructions cause a computer to execute the computer-implemented method, the method comprising: receiving first image data of a first portion of a patient's body in a first image modality (1110), receiving second image data of a second portion of the patient's body in a second image modality (1120), modifying the second image data from the second image modality to the first image modality (1160), and generating a three-dimensional model of the first and second portions of the patient's body based on the first image data in the first image modality and the modified second image data in the second image modality (1170).

Description

[0001] This is a divisional application. The parent application is entitled "Method, System and Apparatus for Medical Image Enhancement with Optimized Transducer Array Placement", filed on January 19, 2022, with application number 202280023809.9.

[0002] Cross-references to related applications This application claims priority to U.S. Provisional Application No. 63 / 140,635, filed January 22, 2021, and U.S. Non-Provisional Application No. 17 / 578,241, filed January 18, 2022, both of which are incorporated herein by reference in their entirety for all purposes. Background Technology

[0003] A tumor therapeutic field (TTField) is a low-intensity alternating electric field in the mid-frequency range that can be used to treat tumors, as described in U.S. Patent No. 7,565,205. The TTField is non-invasively induced in the region of interest by placing transducers on the patient's body and applying an AC voltage between the transducers. To determine the effective localization of the transducers on the patient's body, a three-dimensional model of a portion of the patient's body can be evaluated. However, sufficient image data of the patient may not be available for generating a three-dimensional model because the available image data may be missing a portion of the body, because the resolution of the image data may be insufficient to generate a three-dimensional model, or because the image data of the first part of the body has a different image modality than that of the second part of the body. Accordingly, any of these problems may hinder the generation of a three-dimensional model of a portion of the patient's body, and thus hinder the effective localization of the transducers used to induce the TTField on the patient's body. Summary of the Invention

[0004] One aspect of the present invention relates to a computer-implemented method for generating a three-dimensional model, the computer including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: receiving first image data of a first portion of a patient's body in a first image modality; receiving second image data of a second portion of a patient's body in a second image modality; modifying the second image data from the second image modality to the first image modality; and generating a three-dimensional model of the first portion and the second portion of the patient's body based on the first image data in the first image modality and the modified second image data in the second image modality.

[0005] The above aspects of the invention are exemplary, and other aspects and variations of the invention will become apparent from the following detailed description of the embodiments. Attached Figure Description

[0006] Figure 1 This is a flowchart of an example method for generating three-dimensional images of body parts of a patient based on two image scans of the patient.

[0007] Figure 2 This is a flowchart of an example method for generating three-dimensional images of a patient's body parts based on a single image scan of the patient.

[0008] Figure 3 This is a flowchart of an example method for generating high-resolution 3D images of patient body parts based on low-resolution images of those parts.

[0009] Figure 4 This is a flowchart of an example method for determining the layout of a transducer array for delivery of a TTField to a part of a patient's body.

[0010] Figure 5 It is a block diagram depicting the example operating environment.

[0011] Figure 6 An example device for electrotherapy is shown. Detailed Implementation

[0012] As discovered by the inventors, the disclosed subject matter provides a method and system for generating a three-dimensional model of a part of a patient's body given an incomplete or inconsistent set of images. The three-dimensional model can then be used to determine the location for placing a transducer on the patient's body to generate a TTField.

[0013] An incomplete or inconsistent set of images of the patient's body can be, for example: an image set missing a part of the patient's body; an image set with insufficient resolution to generate a three-dimensional model; or an image set of a first part of the patient's body having an image modality different from the image data of a second part of the patient's body. Using one or more of the techniques of the present invention, a three-dimensional model of a part of the patient's body can then be generated given such an incomplete or inconsistent set of images.

[0014] Figure 1 This is a flowchart of an example method 1100 for generating three-dimensional images of patient body parts based on two image scans of a patient, wherein at least a portion of the two image scans includes different parts of the patient's body parts, and wherein the two image scans have different image modalities. The two images of the patient in the different image modalities can each be an image of the same patient body part. The method described herein can be implemented for any body part of a patient.

[0015] At 1110, the patient support system 1002 can receive first image data of a first portion of a patient's body part in a first image modality. For example, the first portion of the patient's body part may be a first portion of the patient's head. Furthermore, the first image data may not include at least a portion of a second portion of the patient's head. For example, the first image data may include the lower portion of the patient's head (e.g., or other body part), but may not include at least a portion of the upper portion of the patient's head (e.g., or other body part).

[0016] At 1120, the patient support system 1002 can receive second image data of a second portion of the patient's body in a second image modality different from the first image modality. For example, the second portion of the patient's body part may be the second portion of the patient's head. Furthermore, the second image data may not include at least a portion of the first portion of the patient's head. For example, the second image data may include the upper part of the patient's head (e.g., or other body part), but may not include at least a portion of the lower part of the patient's head (e.g., or other body part).

[0017] The received first / second image data may be in the image modality of X-ray computed tomography (CT) data, and the second image data may be X-ray CT data. In another example, the first / second image modality may be any of single-photon emission computed tomography (SPECT) data, magnetic resonance imaging (MRI) data, positron emission tomography (PET) data, etc., and the second image may include SPECT data, MRI data, PET data, etc. The first / second image data may be received from imaging data 610, local database 1018, or remote image database 1020 via predictive modeling application 1014. The first and second image data may be acquired at the same or different orientations of the patient's body parts. The first and second image data of the patient's body parts may have been acquired at the same or different times.

[0018] At 1130, the patient support system 1002 can determine that the first image modality of the first image data does not match the second image modality of the second image data. The predictive modeling application 1014 can compare the modality fields of the files of each of the first and second image scans to determine whether the modalities of each of the first and second image scans are the same or different. The modality fields can provide indicators indicating the modality of the image scan. If the predictive modeling application 1014 compares the information in the modality fields of each modality of the first and second image scans and determines that they are the same, then the predictive modeling application 1014 can use the first and second image scans substantially as described at 1170 to generate a complete three-dimensional image of the patient's body part. In this example, the predictive modeling application 1014 compares the information in the modality fields of each image modality of the first and second image scans and determines that the modalities are different.

[0019] The predictive modeling application 1014 may be able to access multiple images of other subjects' body parts. These multiple images may include a first portion image and a second portion image, the first portion image including image data of a first portion of the subject's body part in a first modality, and the second portion image including image data of a second portion of the subject's body part in a second image modality. These multiple images may be stored in an image database 1018 of the patient support system 1002, and / or may be accessed from another image database 1020 that may be located remotely from the patient support system 1002.

[0020] Predictive modeling application 1014 can query databases 1018 and 1020 to retrieve multiple images for developing models to convert images from one image modality to another. For example, for subjects other than patients, the query can determine which images in the database belong to the same subject, including image data of a first portion of the subject's body part in a first image modality and separate image data of a second portion of the subject's body part in a second image modality. The set of image data satisfying the query can be selected for analysis performed during the creation of the modality conversion model.

[0021] The number of subject images used to create the modality transformation model can be configurable and can be any number greater than the image data of a single subject excluding the patient. In some example embodiments, a target or threshold number of subjects satisfying the query criteria must be met in order to create the modality transformation model. In some example embodiments, the target threshold can be the image data of at least 5 subjects satisfying the query. For example, the target threshold can be in the range of 15-50 subjects satisfying the query.

[0022] In some example embodiments, the predictive modeling application 1014 may collect image data of the first portion of a body part in a first image modality and image data of the second portion of a body part in a second image modality from a number of subjects equal to a target threshold. In other example embodiments, the predictive modeling application may collect image data of the first portion of a body part in a first image modality and image data of the second portion of a body part in a second image modality from any number of subjects that meet the target threshold and are available in an image database.

[0023] The queries for databases 1018 and 1020 may also include one or more other query optimization factors. For example, at least a portion of these factors may be based on one or more physical attributes of the patient / subject. For instance, query optimization factors may include one or more of the following: patient's age, age range, patient's height, height range, patient's gender, patient's ethnicity, patient's weight, weight range, one or more diseases, conditions, or abnormalities of the patient, one or more dimensions of body parts, ratios of one or more dimensions of body parts, etc. In some example embodiments, multiple super-resolution models may be generated based on one or more of these query optimization factors. The determination of the number and / or type of factors to be included in the query may be user-configurable and / or user-determined.

[0024] At 1140, the predictive modeling application 1014 may receive first plurality of image data of at least a first portion of body parts from multiple other subjects. The images of at least a first portion of these other subjects' body parts may be in a first image modality (e.g., MRI). The images of at least a first portion of the body parts may be received based on queries to databases 1018, 1020. The queries to the databases may or may not include one or more query optimization factors.

[0025] At 1150, the predictive modeling application 1014 (e.g., or another part of the patient support system 1002) can receive a second plurality of image data for at least a second portion of a body part of a plurality of subjects for which it has received the first image data. The image data for at least a second portion of the body parts of these subjects may be in a second image modality (e.g., x-ray CT). The images of the second portion of the body parts may be received based on queries to databases 1018, 1020. The queries to the databases may or may not include one or more query optimization factors.

[0026] At 1160, the predictive modeling application 1014 can convert a second image scan of a second portion of a patient's body part from a second image modality to a first image modality. For example, the predictive modeling application 1014 can employ artificial intelligence techniques to generate an image modality transformation model for converting image data in the second image modality into image data in the first image modality, using first plurality of image data of at least a first portion of other subjects' body parts in the first image modality and second plurality of image data of at least a second portion of other subjects' body parts in the second image modality.

[0027] For example, predictive modeling application 1014 can apply generative adversarial network (GAN) analysis to generate image modality transformation models. For instance, predictive modeling application 1014 can apply MedGAN analysis to generate image modality transformation models. In other examples, predictive modeling application 1014 can apply another form of GAN analysis, including but not limited to super-resolution GAN, pix2pix GAN, CycleGAN, DiscoGAN, and Fila-sGAN. In other example embodiments, predictive modeling application 1014 can apply another form of modeling to generate image modality transformation models, such as projective adversarial networks (PAN) or variational autoencoders (VAE).

[0028] Once an image modality conversion model has been generated based on a first plurality of image data of at least a first portion of the body parts of other subjects in the first image modality and a second plurality of image data of at least a second portion of the body parts of other subjects in the second image modality, the predictive modeling application 1014 can apply the model to image scans of the second portion of the patient's body parts in the second image modality to convert the second image scan from the second image modality (e.g., x-ray CT) to the first image modality (e.g., MRI) and / or to the same image modality as the first image scan of the first portion of the patient's body parts.

[0029] At 1170, the predictive modeling application 1014 can generate a complete 3D model of the patient's body part based on first image data of a first portion of the patient's body part in a first image modality and transformed second image data of a second portion of the patient's body part in the first image modality. For example, since the first image and the transformed second image are in the same image modality, the predictive modeling application 1014 can overlay or otherwise combine all or part of the transformed second image data of the second portion of the patient's body part onto the first image data of the first portion of the patient's body part, and can add portions of the patient's body part in the transformed second image data but not in the first image data to the first image data. For example, the body part can be the patient's head. The first image can include a portion of the patient's head, but may also omit another portion of the patient's head (e.g., at least a portion of the upper part of the patient's head). The transformed second image data can include portions of the desired body part that are missing from the first image data. For example, the transformed second image data can include the upper part of the patient's head, but may not include the entire patient's head. Predictive modeling application 1014 can generate a complete 3D model of a patient's head by obtaining image data of a transformed second image of a part of a body part missing in the first image data and adding the image data to the first image data to create a digital representation in 3D space of all or part of the patient's body parts (including internal structures such as tissues, organs, tumors, etc.).

[0030] Figure 2 This is a flowchart of an example method 1200 for generating a three-dimensional image of a patient's body part based on a single image scan that includes a portion of the patient's body part, wherein the image data does not include another portion of the patient's body part.

[0031] At 1210, the patient support system 1002 can receive first image data of a first part of a patient's body part. The first image may not include at least a portion of a second part of the patient's body part.

[0032] At 1220, the patient support system 1002 can determine that a second portion of the patient's body part is needed to generate a complete three-dimensional model of the body part. For example, the patient support system 1002 can evaluate first image data and determine that the image data includes only a portion of the body part required to model the delivery of the TTField to the patient's body part.

[0033] At 1230, the predictive modeling application 1014 can query a database for image data of body parts identical to those of one or more subjects and the same body part as the patient. In response to this query, the predictive modeling application 1014 can receive multiple image data of body parts from multiple subjects other than the patient. The predictive modeling application 1014 can query databases 1018 and 1020 to retrieve multiple images for developing a complete model of the body part, thereby adding image data representing additional parts of the body part to the first image data of the patient's body part. For example, adding image data representing additional parts of the body part to the first image data of the body part can produce a complete image or a more complete image of the patient's body part. This query can, for example, determine which images in the database are images of body parts identical to those of the subjects other than the patient. The query can be narrowed down to include image data of body parts identical to those of the subjects and the patient, where the subject's image data represents a complete image of the body part or an image of a body part more complete than the patient's first image data. Image data satisfying this query can be selected for analysis performed in the creation of the complete model of the body part.

[0034] At 1240, the predictive modeling application 1014 can segment the received image data of each subject into at least two parts. For example, the predictive modeling application 1014 can segment the received image data of each subject's body part into a first part including a first portion of the body part and a second part including a second portion of the body part. For example, the first part can be a portion of a body part that is typically included in a clinical scan. For the example head, the first part can be most of the head except for the top portion and / or one or more lateral portions of each subject's head. For example, the second part can be a portion of a body part that is not typically included in a clinical scan.

[0035] At 1250, predictive modeling application 1014 can determine a complete body part model for generating all or part of the remaining portion of a body part from image data. Predictive modeling application 1014 can employ artificial intelligence techniques and first and second parts of image data of body parts from multiple subjects to determine a complete body part model for generating all or part of the remaining portion of a body part from image data of a patient's body part. In one example, predictive modeling application 1014 can employ statistical shape analysis of the first and second parts of image data of body parts from multiple subjects to determine a complete body part model. In another example, predictive modeling application 1014 can employ active appearance modeling of the first and second parts of image data of body parts from multiple subjects to determine a complete body part model. In yet another example, predictive modeling application 1014 can employ global image statistics of the first and second parts of image data of body parts from multiple subjects to determine a complete body part model. Any of the proposed techniques for determining a complete body part model can model the geometric relationships between head image statistics and head and / or brain structures in the segmented first and second parts of image data of body parts from multiple subjects. After training on a large dataset, a machine learning regressor (e.g., random forest) can be incorporated to predict missing body parts from the first image data. In another example, the predictive modeling application 1014 can employ GAN analysis (e.g., MedGAN, Super-Resolution GAN, pix2pix GAN, CycleGAN, DiscoGAN, and Fila-sGAN) to augment the evaluated dataset to include a large number (e.g., more than 100, more than 1000, more than 5000) of simulated image scans of body parts. Then, the first and second parts of the image data of body parts from multiple subjects can be fed as input into an artificial neural network containing convolutional blocks and trained to output images of entire body parts (e.g., the entire head, torso, arms, legs, etc.), which include the missing portions of the body part within the determined complete body part model.

[0036] At 1260, predictive modeling application 1014 can apply a complete model of the body part to the first image scan of the first part of the patient's body part. For example, the complete model of the body part can be applied to the image data of the first part of the patient's body part using artificial intelligence techniques to determine all or at least a portion of the remaining parts of the patient's body part that are not included in the first image scan of the body part.

[0037] At 1270, the predictive modeling application 1014 can generate second image data of a second part of the patient's body part, which supplements and is based on the image data of the first part of the patient's body part. The image data of the second part of the body part can be a three-dimensional discrete image representing the second part of the patient's body part. In one example, the second image data representing the second part of the patient's body part can be any remaining portion of the body part that is not included in the first image data.

[0038] At 1280, the predictive modeling application 1014 can generate a complete 3D model of the patient's body part (or a part of the body part) based on the first image data of the first part of the patient's body part and the second image data of the generated second part of the patient's body part.

[0039] Figure 3 This is a flowchart of an example method 1300 for generating high-resolution three-dimensional images (e.g., MRI) of a patient's body parts based on low-resolution images (e.g., SPECT scans or PET scans). The image data from the low-resolution image scan can be image data of the entire body part of the patient or a portion of the body part.

[0040] At 1310, the predictive modeling application 1014 may receive multiple first image data of multiple other subjects' body parts at a first resolution. The first resolution may be high resolution (e.g., MRI or X-ray CT images). Each of the first image data may have the same image modality. The multiple first image data may be received based on queries to databases 1018 and 1020. The queries may or may not include optimization factors.

[0041] Predictive modeling application 1014 can query databases 1018 and 1020 to retrieve multiple images for developing models that generate high-resolution image data (e.g., MRI) based on low-resolution image data (e.g., SPECT or PET scans) of a patient's body parts. This query can, for example, determine which images in the database belong to the same subject (excluding the patient) and include both high-resolution and low-resolution image data of the subject's body parts. Sets of image data satisfying the query can be selected for analysis in the creation of the super-resolution model. The number of subject images used for creating the super-resolution model can be configurable and can be any number greater than one subject (excluding the patient). In some example embodiments, a target or threshold number of subjects satisfying the query criteria must be met to create the super-resolution model. In some example embodiments, the target threshold can be image data of at least 100 subjects satisfying the query (e.g., low-resolution and high-resolution image data of body parts). For example, the target threshold can be in the range of 50-5000 subjects satisfying the query. In some example embodiments, the predictive modeling application 1014 may collect only low-resolution and high-resolution image data of a number of body parts of the subject equal to a target threshold. In other example embodiments, the predictive modeling application 1014 may collect any number of low-resolution and high-resolution image data of body parts of the subject that meet the target threshold and are available in image databases 1018, 1020.

[0042] At 1320, the predictive modeling application 1014 can receive multiple second image data of body parts from multiple subjects. Therefore, for each subject, the predictive modeling application 1014 can receive both first and second image data of the body part. Each of the multiple second image data can be at a second resolution. The second resolution can be low resolution. Each of the multiple second image data can have the same image modality and can be different from the image modality of the first image data. The multiple second image data can be received based on queries to databases 1018 and 1020. The queries may or may not include optimization factors.

[0043] At 1330, the predictive modeling application can determine a super-resolution model for generating high-resolution image data (e.g., MRI) of a patient's body parts based on low-resolution image data (e.g., SPECT or PET data). For example, predictive modeling application 1014 can employ artificial intelligence techniques to generate a super-resolution model for generating high-resolution image data of a patient's body parts based on low-resolution image data of the patient's body parts, using multiple first image data and multiple second image data of body parts from multiple subjects. For example, predictive modeling application 1014 can apply a form of generative adversarial network (GAN) analysis to the multiple first and multiple second image data of body parts from multiple subjects to generate the super-resolution model. For example, predictive modeling application 1014 can apply MedGAN analysis to generate the super-resolution model. In other examples, predictive modeling application 1014 can apply another form of GAN analysis, including but not limited to super-resolution GAN, pix2pix GAN, CycleGAN, DiscoGAN, and Fila-sGAN. Predictive modeling application 1014 can apply another form of modeling to generate super-resolution models (such as regression models or convolutional networks).

[0044] At 1340, the predictive modeling application 1014 can receive image data of a patient's body parts. The image data of the body parts can be a low-resolution second resolution (e.g., lower than the resolution of MRI image data).

[0045] At point 1350, once a super-resolution model has been generated based on multiple first image data of body parts at a first resolution and multiple second image data of body parts at a second resolution for multiple subjects (e.g., humans), the predictive modeling application 1014 can apply the model to the received image data of the patient's body parts at the second resolution. In some examples, the super-resolution model can be generated before the image data of the patient's body parts is received. In other example embodiments, the super-resolution model can be generated after the image data of the patient's body parts is received.

[0046] At position 1360, predictive modeling application 1014 can generate image data of the patient's body parts at a first resolution. The generation of this first-resolution image data can be based on applying a super-resolution model to the received second-resolution image data of the patient's body parts. The first resolution can be higher than the second resolution. Based on the second-resolution image data of the patient's body parts and the super-resolution model, the generated first-resolution image data of the patient's body parts can be a complete three-dimensional model of the patient's body parts.

[0047] Figure 4 This is a flowchart of an example method 1400 for determining the transducer array layout for delivery of a TTField to a part of a patient's body. Method 1400 may be performed by one or more of device 100, patient support system 1002, patient modeling application 608, and / or any other device / component described herein.

[0048] At point 1410, a three-dimensional model of a part of the patient's body can be received. For example, this three-dimensional model can be received by a patient modeling application 608. This three-dimensional (3D) model can be... Figure 1-3 One or more 3D models are generated in the process, and may include a body part or a portion of the patient's body. At 1420, a region of interest (ROI) can be determined within the 3D model of a portion of the patient's body. At 1430, a simulated electric field distribution can be determined. At 1440, a dose metric can be determined. For example, the dose metric can be determined based on the simulated electric field distribution. For example, the dose metric can be determined for each of a plurality of positioning pairs of transducer arrays. At 1450, one or more sets of positioning pairs among a plurality of positioning pairs that satisfy angular constraints between transducer array pairs are determined. For example, the angular constraints may be and / or indicate orthogonal angles between the plurality of transducer array pairs. For example, the angular constraints may be and / or indicate a range of angles between the plurality of transducer array pairs. At 1460, one or more candidate transducer array layouts can be determined. For example, the one or more candidate transducer array layouts may be determined based on the dose metric and one or more sets of positioning pairs that satisfy the angular constraints. In some instances, method 1400 may include adjusting the simulated orientation or simulated positioning of at least one transducer array at at least one location of one or more candidate transducer array layouts, and determining a final transducer array layout based on the adjusted simulated orientation or simulated positioning of the at least one transducer array.

[0049] Figure 5 This is a block diagram depicting an environment 1000 including a non-limiting example of a patient support system 1002. In one aspect, some or all of the steps of any of the described methods can be performed on a computing device as described herein. The patient support system 1002 may include one or more computers configured to store an electric field generator (EFG) configuration application 606, a patient modeling application 608, imaging data 610, an operating system (O / S) 1012, a predictive modeling application 1014, an image database 1018, and one or more of such.

[0050] The patient support system 1002 may be a digital computer, which, in terms of hardware architecture, generally includes one or more processors 1004, a memory system 1006, an input / output (I / O) interface 1008, and a network interface 1010. These components (1004, 1006, 1008, and 1010) are communicatively coupled via a local interface 1016. The processor 1004 may be a hardware device for executing software, particularly software stored in the memory system 1006. When the patient support system 1002 is in operation, the processor 1004 may be configured to execute the software stored in the memory system 1006, transfer data to and from the memory system 1006, and generally control the operation of the patient support system 1002 according to the software. The patient support system 1002 may be a computer including one or more processors and memory accessible by the one or more processors, wherein the memory stores instructions that, when executed by the one or more processors, cause the computer to perform one or more methods disclosed herein.

[0051] The patient modeling application 608 can be configured to generate a 3D model of a part of a patient's body based on imaging data 610. Imaging data 610 can include any type of visual data, such as single-photon emission computed tomography (SPECT) image data, x-ray computed tomography (CT) data, magnetic resonance imaging (MRI) data, positron emission tomography (PET) data, and data that can be captured by optical instruments. In some implementations, the image data can include 3D data obtained from or generated by a 3D scanner. The patient modeling application 608 can also be configured to generate a 3D array layout map based on the patient model and one or more electric field simulations. To appropriately optimize array placement on a part of the patient's body, imaging data 610, such as MRI imaging data, can be analyzed by the patient modeling application 608 to identify regions of interest, including tumors. In one aspect, the patient modeling application 608 can be configured to determine the desired transducer array layout for the patient based on the location and extent of the tumor. In another aspect, the patient modeling application 608 can be configured to determine a 3D array layout map of the patient.

[0052] Network interface 1010 can be used to transmit and receive data from patient support system 1002. Figure 5 In the example, the software in the memory system 1006 of the patient support system 1002 may include an EFG configuration application 606, a patient modeling application 608, imaging data 610, a predictive modeling application 1014, an image database 1018, and an operating system 1012.

[0053] Predictive modeling application 1014 can be one or more modeling applications used to generate image data models based on image data from multiple subjects. The predictive modeling application can be configured to perform one or more of the following: Generative Adversarial Network (GAN) analysis, MedGAN analysis, super-resolution GAN, pix2pix GAN, CycleGAN, DiscoGAN, Fila-sGAN, Projective Adversarial Network (PAN) analysis, Variational Autoencoder (VAE) analysis, regression analysis, or convolutional network analysis. For example, predictive modeling application 1014 can employ one or more artificial intelligence techniques to analyze subject image data.

[0054] Figure 6 An example apparatus 100 for electrotherapy is shown. The apparatus 100 may include an electric field generator 102 and one or more transducer arrays 104. The apparatus 100 may be configured to generate a TTField via the electric field generator 102 and deliver the TTField to a region of the body via the one or more transducer arrays 104. The electric field generator 102 may include one or more processors 106 in communication with a signal generator 108. The electric field generator 102 may include control software 110 configured to control the performance of the processors 106 and the signal generator 108. The control software 110 may be stored in a memory accessible by the one or more processors 106. The signal generator 108 may generate one or more electrical signals in the form of a waveform or a series of pulses. The signal generator 108 may be configured to generate an AC voltage waveform at a frequency, for example, from approximately 50 kHz to approximately 500 kHz. The voltage allows the electric field strength in the tissue to be treated to be in the range, for example, from approximately 0.1 V / cm to approximately 10 V / cm.

[0055] One or more outputs 114 of the electric field generator 102 may be coupled to one or more conductive leads 112, which are attached at one end to a signal generator 108. The opposite ends of the conductive leads 112 are connected to one or more transducer arrays 104 activated by an electrical signal. Output parameters of the signal generator 108 may include the field strength, wave frequency, and maximum permissible temperature of the one or more transducer arrays 104. These output parameters may be set and / or determined by control software 110 in conjunction with processor 106.

[0056] The one or more transducer arrays 104 may include one or more electrodes 116. The electrodes 116 may be biocompatible and coupled to a flexible circuit board 118. The electrodes 116, hydrogel, and flexible circuit board 118 may be attached to a hypoallergenic medical adhesive bandage 120 to hold the one or more transducer arrays 104 in place on the body and in continuous direct contact with the skin. Each transducer array 104 may include one or more sensors, such as thermistors, to measure the skin temperature beneath the transducer array 104. The one or more transducer arrays 104 may vary in size and may include a varying number of electrodes 116. The transducer arrays 104 may be configured for placement on specific sites on a patient's body, such as the head, torso, arm, or leg.

[0057] In one example, electrode 116 may be a ceramic disk, and each of the ceramic disks may have a diameter of approximately 2 cm and a thickness of approximately 1 mm. In another example, electrode 116 may be a non-disc-shaped ceramic element. In yet another example, electrode 116 may be a non-ceramic dielectric material positioned on a plurality of flat conductors. Examples of non-ceramic dielectric materials positioned on flat conductors may include polymer films disposed on pads on a printed circuit board or on a flat metal sheet. In certain embodiments, transducers using an array of electrodes that are not capacitively coupled may also be used. In this case, each electrode element 116 may be implemented using a region of conductive material configured for placement against the subject's body, wherein no insulating dielectric layer is disposed between the conductive element and the body. In other embodiments, the transducer may comprise only a single electrode element. As an example, a single electrode element may be a flexible organic material or a flexible organic composite material positioned on a substrate. As another example, the transducer may comprise a flexible organic material or a flexible organic composite material without a substrate.

[0058] Other alternative configurations for implementing the transducers used in the embodiments of the present invention may also be used, provided that they are able to (a) deliver the TTField to the body of the subject and (b) be positioned at the location specified herein.

[0059] The present invention includes other illustrative embodiments, such as the following embodiments.

[0060] Illustrative Example 1: A non-transitory computer-readable medium including instructions for generating a three-dimensional model, the instructions causing the computer, when executed by a computer, to perform a method comprising: receiving first image data of a first portion of a patient's body in a first image modality; receiving second image data of a second portion of a patient's body in a second image modality; modifying the second image data from the second image modality to the first image modality; and generating a three-dimensional model of the first portion and the second portion of the patient's body based on the first image data in the first image modality and the modified second image data in the second image modality.

[0061] Illustrative Example 2: A non-transitory computer-readable medium including instructions for generating a three-dimensional model, the instructions causing the computer to perform a method when executed by a computer, the method comprising: receiving first image data of a first portion of a body part of a patient, wherein the first portion of the body part is less than a complete body part; receiving a plurality of second image data of body parts of a plurality of subjects; determining a complete model of the body part based on the plurality of second image data; generating third image data of a second portion of the body part based on the complete model of the body part and the first image data; and generating a three-dimensional model of the patient's body part based on the first image data and the third image data.

[0062] Illustrative Example 3: A non-transitory computer-readable medium including instructions for generating a three-dimensional model, the instructions, when executed by a computer, causing the computer to perform a method comprising: receiving first image data of a portion of a patient's body at a first image resolution; receiving a plurality of second image data of a plurality of subjects; determining a super-resolution model based on the plurality of second image data for improving the resolution of the first image data; and generating third image data of the portion of the patient's body at a second image resolution based on the super-resolution model and the first image data, wherein the second image resolution is greater than the first image resolution.

[0063] Illustrative Example 4: A system for generating a three-dimensional model, the system including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the system to perform a method comprising: receiving first image data of a first portion of a patient's body in a first image modality; receiving second image data of a second portion of a patient's body in a second image modality; modifying the second image data from the second image modality to the first image modality; and generating a three-dimensional model of the first portion and the second portion of the patient's body based on the first image data in the first image modality and the modified second image data in the second image modality.

[0064] Illustrative Example 5: A system for generating a three-dimensional model, the system including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the system to perform a method, the method comprising: receiving first image data of a first portion of a body part of a patient, wherein the first portion of the body part is less than a complete body part; receiving a plurality of second image data of body parts of a plurality of subjects; determining a complete model of the body part based on the plurality of second image data; generating third image data of the second portion of the body part based on the complete model of the body part and the first image data; and generating a three-dimensional model of the patient's body part based on the first image data and the third image data.

[0065] Illustrative Example 6: A system for generating a three-dimensional model, the system including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the system to perform a method, the method comprising: receiving first image data of a portion of a patient's body at a first image resolution; receiving a plurality of second image data of a plurality of subjects; determining a super-resolution model based on the plurality of second image data for increasing the resolution of the first image data; and generating third image data of the portion of the patient's body at a second image resolution based on the super-resolution model and the first image data, wherein the second image resolution is greater than the first image resolution.

[0066] Illustrative Example 7: A computer-implemented method for generating a three-dimensional model, the computer including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: receiving first image data of a first portion of a patient's body in a first image modality; receiving second image data of a second portion of a patient's body in a second image modality; modifying the second image data from the second image modality to the first image modality; and generating a three-dimensional model of the first portion and the second portion of the patient's body based on the first image data in the first image modality and the modified second image data in the second image modality.

[0067] Illustrative Example 8: A computer-implemented method according to Illustrative Example 7, wherein the first image modality includes magnetic resonance imaging (MRI).

[0068] Illustrative Example 9: The computer-implemented method according to Illustrative Example 7, wherein the first part of the patient's body is the first part of the patient's body parts, and the second part of the patient's body is the second part of the patient's body parts.

[0069] Illustrative Example 10: A computer-implemented method according to Illustrative Example 7, wherein the body part is one of the head, torso, arm, or leg.

[0070] Illustrative Example 11: A computer-implemented method for generating a three-dimensional model, the computer including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: receiving first image data of a first portion of a body part of a patient, wherein the first portion of the body part is less than a complete body part; receiving a plurality of second image data of body parts of a plurality of subjects; determining a complete model of the body part based on the plurality of second image data; generating third image data of a second portion of the body part based on the complete model of the body part and the first image data; and generating a three-dimensional model of the patient's body part based on the first image data and the third image data.

[0071] Illustrative Example 12: The computer-implemented method according to Illustrative Example 11, wherein the three-dimensional model is a complete model of a patient's body parts.

[0072] Illustrative Example 13: A computer-implemented method for generating a three-dimensional model, the computer including one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: receiving first image data of a portion of a patient's body at a first image resolution; receiving a plurality of second image data of a plurality of subjects; determining a super-resolution model based on the plurality of second image data for increasing the resolution of the first image data; and generating third image data of the portion of the patient's body at a second image resolution based on the super-resolution model and the first image data, wherein the second image resolution is greater than the first image resolution.

[0073] Illustrative Example 14: A computer-implemented method according to Illustrative Example 13, wherein the first image resolution includes magnetic resonance imaging.

[0074] The embodiments described under any heading or in any part of this disclosure may be combined with the embodiments described under the same or any other heading or other part of this disclosure, unless otherwise indicated herein or otherwise obviously contradicted by the context.

[0075] Numerous modifications, alterations, and variations of the described embodiments are possible without departing from the scope of the invention as defined in the claims. It is intended that the invention be limited to the described embodiments, but rather has the full scope defined by the language of the following claims and their equivalents.

Claims

1. A computer-implemented method for generating a three-dimensional model, the computer comprising one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: Receive first image data of a first part of the patient's body in a first image modality; Receive second image data of a second part of the patient's body in a second image modality; Modify the second image data from the second image modality to the first image modality; and Three-dimensional models of the first and second parts of the patient's body are generated based on the first image data in the first image modality and the modified second image data in the second image modality.

2. The method of claim 1, further comprising generating an image modality transformation model, wherein modifying the second image data from a second image modality to a first image modality comprises applying the image modality transformation model to the second image data in the second image modality.

3. The method of claim 2, wherein generating the image modality transformation model comprises: Receive multiple image data of the first part of the patient's body in the first image modality of multiple subjects; and Receive second multiple image data of the second part of the patient's body in a second image modality from multiple subjects. An image modality transformation model is generated based on the analysis of a first set of multiple image data and a second set of multiple image data.

4. The method of claim 3, wherein the analysis comprises at least one of the following: generative adversarial network (GAN) analysis, MedGAN analysis, super-resolution GAN analysis, pix2pix GAN analysis, cycleGAN analysis, discoGAN analysis, fila-sGAN analysis, projective adversarial network (PAN) analysis, variational autoencoder (VAE) analysis, or regression analysis.

5. The method of claim 1, further comprising determining a transducer array layout along at least one of the first and second parts of the patient's body based on a three-dimensional model of the first and second parts of the patient's body.

6. A computer-implemented method for generating a three-dimensional model, the computer comprising one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: Receive first image data of a first part of a patient’s body part, wherein the first part of the body part is less than the complete body part; Receive multiple second image data of body parts from multiple subjects; Based on the multiple second image data, a complete model of the body parts is determined; Based on the complete model of the body part and the first image data, third image data of the second part of the body part is generated; and Based on the first and third image data, a three-dimensional model of the patient's body parts is generated.

7. The method of claim 6, further comprising dividing each of the plurality of second image data into a first portion of image data and a second portion of image data, wherein the first portion of image data includes a first part corresponding to a body part of the subject, and the second portion of image data includes another part corresponding to a body part of the subject.

8. The method of claim 7, further comprising analyzing the first portion of image data and the second portion of image data for each of a plurality of subjects.

9. The method of claim 8, wherein the analysis comprises at least one of statistical shape analysis, active appearance analysis, or global image statistical analysis.

10. The method of claim 6, further comprising generating a three-dimensional model of the patient's body part by determining image data of a second part of the body part to be desired based on the first image data.

11. The method of claim 6, further comprising determining a transducer array layout along the patient's body parts based on a three-dimensional model of the patient's body parts.

12. The method of claim 6, wherein the body part is the head, and the first image data of the first portion of the body part does not include the top portion of the patient's head.

13. A computer-implemented method for generating a three-dimensional model, the computer comprising one or more processors and a memory accessible by the one or more processors, the memory storing instructions that, when executed by the one or more processors, cause the computer to perform the method, the method comprising: Receive first image data of a part of the patient's body at a first image resolution; Receive multiple second image data from multiple subjects; Based on the plurality of second image data, a super-resolution model for increasing the resolution of the first image data is determined; and A third image of the patient's body at a second image resolution is generated based on a super-resolution model and first image data, wherein the second image resolution is greater than the first image resolution.

14. The method of claim 13, wherein receiving the plurality of second image data from the plurality of subjects comprises: Receive first and a plurality of second image data of the same body parts of the subjects as the patient's body at a first image resolution; and Receive second plurality of second image data of the same parts of the bodies of the plurality of subjects at a second image resolution. Determining the super-resolution model involves analyzing a first set of multiple second image data and a second set of multiple second image data.

15. The method of claim 14, wherein the analysis comprises at least one of regression analysis, convolutional network analysis, generative adversarial network (GAN) analysis, MedGAN analysis, super-resolution GAN analysis, pix2pix GAN analysis, cycleGAN analysis, discoGAN analysis, or fila-sGAN analysis.

Images