Simulating structures in images

Abstract

Systems and methods for simulating structures and images are disclosed. Techniques described herein may include obtaining a first image of a subject. The techniques may include determining a location within the first image for simulating the structure. The techniques may include simulating a shape for the structure as a function of the location. The techniques may include generating a mask according to the location and a shape for the structure. The techniques may include applying a mask to the first image to generate a second image simulating the structure.

Description

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Application No. 63 / 388,995, filed on July 13, 2022, entitled “Abnormality Simulation Improves Model Performance,” which is incorporated herein by reference in its entirety. Technical Field

[0003] The present invention relates generally to the field of medical imaging including multi-segment magnetic resonance (MR) imaging. Background Art

[0004] Magnetic resonance imaging (MRI) systems can be used to generate images of the interior of the human body. MRI systems can be used to detect magnetic resonance (MR) signals in response to applied electromagnetic fields. MRI techniques can include fast spin echo (FSE) imaging, in which a series of radio frequency (RF) pulses are used to excite protons in tissue. FSE imaging techniques are used to construct images from multiple echoes generated by refocusing the magnetization of the excited protons to increase the speed of image acquisition. Summary of the invention

[0005] The system and method of the technical solution provide techniques for simulating structures and / or images. By utilizing the techniques described herein, one or more structures (such as normal structures (e.g., healthy tissues) and / or abnormal structures (e.g., lesions), etc.) can be simulated for at least one body part of a subject (e.g., a patient or an individual). Body parts may include, but are not limited to, the brain, heart, lungs, liver, kidneys, or other parts of the body. These structures can be applied to images of various body parts associated with one or more subjects. In some aspects, these techniques can simulate one or more images by applying the simulated structure to at least one historical image or aggregating the simulated structure to simulate / generate at least one new image. By utilizing these techniques, the simulated structures and / or images can be used to train machine learning (ML) or artificial intelligence (AI) models (e.g., sometimes referred to as ML / AI models) to improve model performance.

[0006] At least one aspect of the present disclosure relates to a method for simulating a structure in an image. The method may include obtaining a first image of a subject. The method may include determining a location within the first image for simulating a structure. The method may include simulating a shape for the structure based on the location. The method may include generating a mask based on the location and the shape for the structure. The method may include applying the mask to the first image to generate a second image simulating the structure.

[0007] In some implementations, to determine the location within the image, the method may include identifying at least one anatomical region associated with a body part of the subject. The method may include extracting a plurality of regions associated with the first image using the identified at least one anatomical region. The method may include selecting at least one first region associated with the first image as the location for the mask.

[0008] In some implementations, after selecting the at least one first region, the method may include selecting at least one second region associated with the first image as a second location for another mask for generating a third image. In some implementations, to determine the location and model the shape, the method may include receiving an indication of at least one region associated with the first image as the location. The method may include modeling the shape of the structure based on the at least one region associated with the first image.

[0009] In some implementations, to generate the mask, the method may include setting a seed to the first image at the location. The method may include growing at least one region around the seed using at least one region growing algorithm. The method may include generating the mask based on the grown seed. In some implementations, the at least one region growing algorithm may include at least one of: region growing, region merging, split and merge, watershed transform, connected component labeling, and graph cut segmentation.

[0010] In some implementations, the seed may be set at at least one voxel having a first intensity. A region around the seed may be grown to at least one neighboring voxel having a second intensity, wherein the second intensity is substantially similar to the first intensity. In some implementations, to simulate the shape, the method may include generating an elliptical shape based on one or more parameters, the one or more parameters including a major axis or a minor axis of the first image defining a size of the structure. The method may include applying an elastic warp to the elliptical shape to simulate the shape of the structure.

[0011] In some implementations, to generate the mask, the method may include obtaining a plurality of historical masks associated with a plurality of images from one or more subjects. The method may include training a model using at least one machine learning technique based on the plurality of historical masks. The method may include using the trained model to generate a mask for the location based on the plurality of historical masks. The method may include refining the mask for simulating the structure based on a comparison between the generated mask and the plurality of historical masks associated with the location.

[0012] In some implementations, to generate the mask, the method may include determining an appearance of the mask associated with at least an intensity of one or more voxels of the first image. In some implementations, to determine the appearance, the method may include selecting an aggregate pixel intensity of the mask. The method may include applying at least one pattern to the mask. The method may include simulating at least one noise to the mask. In some implementations, the at least one pattern may include at least one of: an edema pattern, a hemorrhagic pattern, a necrotic pattern, a cystic pattern, an inflammatory pattern, a tumor pattern, and an ischemic pattern.

[0013] In some implementations, to determine the appearance and apply the mask, the method may include providing at least a portion of the first image for applying the mask and the generated mask as input to a model trained using machine learning techniques. The method may include using the model to generate a third image, the third image including the at least a portion of the first image and the generated mask. The method may include updating the third image based on a comparison of the third image with at least one historical image, the at least one historical image having a second mask at the location, the second mask having a second shape similar to the shape of the mask.

[0014] In some implementations, to apply the mask, the method may include simulating a space-occupying effect around the mask. The method may include applying the mask to a first image with the simulated space-occupying effect to generate the second image simulating the structure. In some implementations, the first image may be a first magnetic resonance (MR) image, and the second image may be a second MR image.

[0015] In some implementations, to generate the mask, the method may include simulating a shape deformation associated with the shape. The method may include generating the mask based on the position and the shape deformation for the structure. In some implementations, simulating the shape deformation may correspond to simulating hydrocephalus of the body part of the subject.

[0016] In some implementations, to apply the mask, the method may include identifying a plurality of anatomical regions associated with a body part of the subject. The method may include simulating a contrast for the first image based on at least one of the plurality of anatomical regions and one or more sequence parameters. The method may include applying the mask and the simulated contrast to the first image to generate the second image simulating the structure.

[0017] In some implementations, the one or more sequence parameters may include at least one of a relaxation time, an echo time, an inversion time, and a flip angle. In some implementations, each of the plurality of anatomical regions is represented as a two-dimensional (2D) image or a three-dimensional (3D) image. In some implementations, the contrast may indicate at least one of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) associated with the body part of the subject.

[0018] In some implementations, to simulate the contrast, the method may include: using at least one signal equation, the at least one signal equation including at least one of the following: spin echo, spoiled gradient echo, rewrap gradient, time reversal gradient echo, and random. In some implementations, to generate the mask, the method may include determining an appearance of the mask based on the simulated contrast.

[0019] In some implementations, the method may include obtaining a plurality of images of a plurality of subjects, the plurality of images including at least a fourth image and a fifth image associated with a body part of at least one subject. The method may include conforming the fifth image to the fourth image. The method may include simulating a sixth image based on the conformed fifth image.

[0020] In some implementations, to conform the fifth image to the fourth image, the method may include at least one of the following operations: conforming a first contrast of the fifth image to a second contrast of the fourth image; and conforming a first size of the fifth image to a second size of the fourth image. In some implementations, the fourth image may be a neonatal image of a body part of the at least one subject, the fifth image may be a developmental image of the body part of the at least one subject, and the sixth image may be another neonatal image simulated based on the developmental image of the body part of the at least one subject.

[0021] In some implementations, to conform the first contrast to the second contrast and the first size to the second size, the method may include at least one of: injecting one or more shapes into the fifth image based on the fourth image; overlaying one or more values ​​associated with pixels or voxels of the fifth image based on the fourth image; and applying one or more distortions to the fifth image based on the fourth image.

[0022] At least one other aspect of the present disclosure relates to a system for simulating a structure in an image. The system may include one or more processors configured to obtain a first image of a subject. The one or more processors may determine a position within the first image for simulating a structure. The one or more processors may simulate a shape for the structure based on the position. The one or more processors may generate a mask based on the position and the shape for the structure. The one or more processors may apply the mask to the first image to generate a second image simulating the structure.

[0023] Yet another aspect of the present disclosure relates to a non-transitory computer-readable storage medium storing processor-executable instructions which, when executed by at least one processor, cause the at least one processor to: obtain a first image of a subject; determine a position within the first image for simulating a structure; simulate a shape for the structure based on the position; generate a mask based on the position and the shape for the structure; and apply the mask to the first image to generate a second image simulating the structure.

[0024] These and other aspects and implementations are discussed in detail below. The foregoing information and the following detailed description include illustrative examples of various aspects and implementations, and provide an overview or framework for understanding the nature and characteristics of the claimed aspects and implementations. The accompanying drawings provide illustrations and further understanding of various aspects and implementations, and are incorporated into and constitute a part of this specification. Various aspects can be combined, and it will be readily understood that the features described in the context of one aspect of the present disclosure can be combined with other aspects. Various aspects can be implemented in any convenient form. In a non-limiting example, by an appropriate computer program, the computer program can be carried on an appropriate carrier medium (computer-readable medium), which can be a tangible carrier medium (such as a disk) or an intangible carrier medium (such as a communication signal). Various aspects can also be implemented using suitable equipment, which can take the form of a programmable computer running a computer program that is arranged not to implement the aspect. As used in the specification and claims, unless the context clearly provides otherwise, the singular forms of "a", "an" and "the" include plural references. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The drawings are not intended to be drawn to scale. Like reference numerals and designations in the various figures indicate like elements. For purposes of clarity, not every component may be labeled in every figure. The patent or application file contains at least one drawing executed in color. Copies of this patent or patent application publication with color drawing(s) will be provided by the Office upon request and payment of the necessary fee. In the drawings:

[0026] Figure 1 Illustrate example components of a magnetic resonance imaging system that may be used to implement techniques for simulating a structure or an image according to one or more implementations;

[0027] Figure 2 depicts an example pipeline for simulating a structure for at least one image according to one or more implementations;

[0028] Figure 3 Depicting examples of brain region classification according to one or more implementations;

[0029] Figure 4 depicting an example of at least one region selected as a target location for a simulated lesion according to one or more implementations;

[0030] Figure 5 depicts examples of generating lesion masks of different shapes and sizes according to one or more implementations;

[0031] Figure 6 depicting example graphs of volume distribution for real lesions and simulated lesions with lesion metastases according to one or more implementations;

[0032] Figure 7 Depicting examples of generating 2D slices of a lesion including interpolation for use according to one or more implementations;

[0033] Figure 8 depicting an example bar graph of intensity distribution for a lesion on diffusion weighted imaging (DWI) according to one or more implementations;

[0034] Fig. 9 depicting examples of simulated edema effects around a lesion according to one or more implementations;

[0035] Fig.10 depicting example images of acquisition noise in a simulated lesion according to one or more implementations;

[0036] Fig.11 depicts examples of different noise levels for an image according to one or more implementations;

[0037] Fig.12depicts an example of simulated footprint effects around a mask according to one or more implementations;

[0038] Fig.13 depicting an example of a hydrocephalus simulation according to one or more implementations;

[0039] Fig.14 depicts examples of MR images with different contrasts according to one or more implementations;

[0040] Fig.15 depicts examples of tissue region probability maps for gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) according to one or more implementations;

[0041] Fig.16 Depicting examples of the same brain using a spin echo signal equation and generating different contrasts according to different repetition times (TR), echo times (TE), and / or inversion times (TI) according to one or more implementations;

[0042] Fig.17 depicts examples of contrast generated using random signal equations according to one or more implementations;

[0043] Fig.18 depicting examples of pediatric simulations according to one or more implementations;

[0044] Fig.19 A flowchart depicting an example method for simulating a structure or image according to one or more implementations; and

[0045] Fig. 20 is a block diagram of an example computing system suitable for use in the various arrangements described herein, according to one or more example implementations.

[0046] Specific implementation aspects

[0047] The following is a detailed description of various concepts related to techniques, approaches, methods, devices, and systems for suppressing artifacts in multi-segment FSE imaging and implementations of these techniques, approaches, methods, devices, and systems. The various concepts introduced above and discussed in detail below can be implemented in any of a number of ways, as the concepts described are not limited to any particular implementation. Examples of specific implementations and applications are provided primarily for illustrative purposes.

[0048] Imaging techniques (such as magnetic resonance imaging (MRI) techniques, computed tomography (CT) imaging techniques, ultrasound imaging techniques, etc.) can be used in clinical practice to scan patients / subjects. When utilized, these imaging techniques can produce or generate images depicting various aspects of the scanned portion of the subject (e.g., a body part). These aspects may include soft tissue and / or hard tissue. For example, taking MR imaging as an example, a magnetic field and a series of radio waves can be used to create detailed images of the internal structure of the body. The operation of MR imaging may include aligning hydrogen atoms in the body with a magnetic field and applying radio waves to excite these atoms. When the atoms return to their original alignment, radio signals are emitted, which can be detected by the MRI machine / system for processing (e.g., generating cross-sectional images). Some uses of MR imaging may include imaging the brain, spine, or joints (and other body parts), or identifying lesions or abnormalities in soft tissues. These images provide information useful for diagnosing and treating various health conditions.

[0049] These images are also useful for training or improving the performance of ML (or AI) models (e.g., sometimes generally referred to as models) to identify various aspects of body parts (including but not limited to lesion detection or analysis). For example, in some systems, a model can be trained to detect lesions (or other abnormalities) based on or by utilizing historical images of examples including normal and abnormal conditions (e.g., body parts with lesions). These historical images can be provided as a training data set for the model. During training, the ML model can learn patterns or features that distinguish lesions from healthy tissue. The model can use any suitable type of ML technique for training and execution (such as convolutional neural networks (CNNs), support vector machines (SVMs), clustering algorithms, etc.) to extract relevant features, mark lesions within images, or perform other operations for diagnosing the subject.

[0050] In order to improve the performance of the model, a large training data set may be required. However, obtaining a real-world data set (e.g., actual scanned images of the subject) for training the model may consume excessive resources and time, where certain characteristics of the lesion (e.g., location, shape, or appearance) are rarely or rarely encountered for analysis or training purposes. In addition, these new images from the imaging device / machine may be manually labeled (e.g., by a technician or healthcare provider) or automatically labeled by an existing model, which may introduce erroneous training data sets (e.g., including labeling a portion of healthy tissue as a lesion or labeling a portion of a lesion as healthy tissue).

[0051] Therefore, the systems and methods of the technical solutions discussed herein can provide techniques for simulating structures or images for improving model performance. By utilizing the techniques discussed herein, the systems and methods can simulate healthy tissue and abnormal structures for images of various body parts. For example, the techniques described herein can simulate healthy tissue and lesions for images of at least a portion of the body. The techniques described herein can simulate one or more lesions within an image. Although various aspects of the present disclosure are described in conjunction with MRI (e.g., MR images), it should be understood that the techniques described herein can be used to simulate structures for other types of imaging (such as, but not limited to, CT images, ultrasound images, or positron emission tomography (PET) scan images, etc.). For the purpose of providing examples herein, the body part of the subject may refer to the brain, although similar techniques discussed herein can be used to simulate images or structures of other body parts (including but not limited to the brain, lungs, liver, kidneys, etc.).

[0052] Figure 1 An example MRI system that can be utilized in conjunction with the structural simulation techniques described herein is illustrated. Figure 1 1 , the MRI system 100 may include a computing device 104, a controller 106, a pulse sequence repository 108, a power management system 110, and a magnetic assembly 120. The MRI system 100 is illustrative and Figure 1 In addition to or in place of the components shown Figure 1 Components illustrated, the MRI system may have one or more other components of any suitable type. In some implementations, one or more components of the MRI system 100 may be operated (including direct or remote operation) by a user 102 or other authorized personnel. In addition, the implementation of the components used by a specific MRI system may be different from those described herein. Examples of low-field MRI systems may include portable MRI systems that may have a field strength that may be less than or equal to 0.5T, less than or equal to 0.2T, in a non-limiting example, may be in the range of 1mT to 100mT, may be in the range of 50mT to 0.1T, may be in the range of 40mT to 80mT, may be about 64mT, and the like.

[0053] The magnetic assembly 120 may include a B0 magnet 122, a shim 124, an RF transmitting and receiving coil 126, and a gradient coil 128. The B0 magnet 122 may be used to generate a main magnetic field B0. The B0 magnet 122 may be a magnetic assembly of any suitable type or combination that may generate a desired main magnetic field B0. In some embodiments, the B0 magnet 122 may be one or more permanent magnets, one or more electromagnets, one or more superconducting magnets, or a hybrid magnet including one or more permanent magnets and one or more electromagnets or one or more superconducting magnets. In some embodiments, the B0 magnet 122 may be configured to generate a B0 magnetic field having a field strength that may be less than or equal to 0.2T or in the range of from 50mT to 0.1T.

[0054] In some implementations, the B0 magnet 122 may include a first B0 magnet and a second B0 magnet, which may each include permanent magnet blocks arranged in concentric rings around a common center. The first B0 magnet and the second B0 magnet may be arranged in a bi-planar configuration such that the imaging zone is located between the first B0 magnet and the second B0 magnet. In some embodiments, the first B0 magnet and the second B0 magnet may each be coupled to and supported by a ferromagnetic yoke configured to capture and direct magnetic flux from the first B0 magnet and the second B0 magnet.

[0055] The gradient coils 128 may be arranged to provide a gradient field, and in a non-limiting example, may be arranged to generate gradients in three substantially orthogonal directions (X, Y, and Z) in the B0 field. The gradient coils 128 may be configured to encode the transmitted MR signals by systematically varying the B0 field (the B0 field generated by the B0 magnet 122 and / or the shim 124) to encode the spatial position of the received MR signals as a function of frequency or phase. In a non-limiting example, the gradient coils 128 may be configured to vary the frequency or phase as a linear function of the spatial position along a particular direction, although more complex spatial encoding profiles may also be provided by using nonlinear gradient coils. In some embodiments, in a non-limiting example, the gradient coils 128 may be implemented using a laminate (e.g., a printed circuit board).

[0056] During an FSE scan, the gradient coils 128 may be controlled to generate phase encoding gradients that may be used to sample MR signals at different locations along different directions (e.g., X, Y, and Z orthogonal directions). A phase encoding gradient is a magnetic field gradient that varies linearly along a phase encoding direction, which induces a phase shift in the MR signal depending on the location of the MR signal in that direction. The gradient coils 128 may be controlled (e.g., by the controller 106) so that the amplitude of the phase encoding gradient varies for each acquisition, so that different locations along the phase encoding direction are sampled. The MR signals may be spatially encoded in the phase encoding direction. The number of phase encoding steps may affect the resolution of the MR image along the phase encoding direction. For example, increasing the number of phase encoding steps may improve image resolution, but also increase scan time, while reducing the number of phase encoding steps may reduce image resolution and reduce scan time.

[0057] MRI scans are performed by exciting and detecting the transmitted MR signals using transmit and receive coils (referred to herein as radio frequency (RF) coils), respectively. The transmit / receive coils may include separate coils for transmitting and receiving, multiple coils for transmitting or receiving, or the same coil for transmitting and receiving. Thus, the transmit / receive assembly may include one or more coils for transmitting, one or more coils for receiving, or one or more coils for transmitting and receiving. The transmit / receive coils may be referred to as Tx / Rx or Tx / Rx coils to generally refer to various configurations of the transmit and receive magnetic components of an MRI system. These terms are used interchangeably herein. Figure 1 , the RF transmit and receive coils 126 may include one or more transmit coils that may be used to generate RF pulses to induce an oscillating magnetic field B1. The transmit coil(s) may be configured to generate any type of suitable RF pulses.

[0058] The power management system 110 includes electronics for providing operating power to one or more components of the MRI system 100. In non-limiting examples, the power management system 110 may include one or more power supplies, energy storage devices, gradient power components, transmit coil components, or any other suitable power electronics required to provide appropriate operating power to power on and operate components of the MRI system 100. Figure 1 As illustrated, the power management system 110 may include a power system 112, (one or more) power components 114, a transmit / receive circuit system 116, and may optionally include a thermal management component 118 (e.g., cryogenic cooling equipment for superconducting magnets, water cooling equipment for electromagnets).

[0059] The power system 112 may include electronics that provide operating power to the magnetic assembly 120 of the MRI system 100. In a non-limiting example, the electronics of the power system 112 may provide operating power to one or more gradient coils (e.g., gradient coils 128) to generate one or more gradient magnetic fields to provide spatial encoding of MR signals. In addition, the electronics of the power system 112 may provide operating power to one or more RF coils (e.g., RF transmit and receive coils 126) to generate one or more RF signals or receive one or more RF signals from a subject. In a non-limiting example, the power system 112 may include a power supply configured to provide power from a mains supply to the MRI system or an energy storage device. In some embodiments, the power supply may be an AC to DC power supply that converts AC power from the mains supply into DC power for use by the MRI system. In some embodiments, the energy storage device may be any one of a battery, a capacitor, a supercapacitor, a flywheel, or any other suitable energy storage device that can bidirectionally receive (e.g., store) power from the mains supply and supply power to the MRI system. Additionally, the power system 112 may include additional power electronics including, but not limited to, power converters, switches, buses, drivers, and any other suitable electronics for supplying power to the MRI system.

[0060] The amplifier(s) 114 may include: one or more RF receive (Rx) preamplifiers that amplify MR signals detected by one or more RF receive coils (e.g., coil 126); one or more RF transmit (Tx) power components that are configured to provide power to one or more RF transmit coils (e.g., coil 126); one or more gradient power components that are configured to provide power to one or more gradient coils (e.g., gradient coil 128), and the amplifier(s) 114 may provide power to one or more shim power components that are configured to provide power to one or more shims (e.g., shims 124). In some implementations, shims 124 may be implemented using permanent magnets, electromagnetic members (e.g., coils), or a combination thereof. The transmit / receive circuit system 116 may be used to select whether the RF transmit coil is being operated or the RF receive coil is being operated.

[0061] like Figure 1As illustrated, the MRI system 100 may include a controller 106 (also referred to as a console), which may include control electronics that send instructions to and receive information from a power management system 110. The controller 106 may be configured to implement one or more pulse sequences that are used to determine instructions sent to the power management system 110 to operate the magnetic assembly 120 in a desired sequence (e.g., parameters for operating the RF transmit and receive coils 126, parameters for operating the gradient coils 128, etc.). In addition, the controller 106 may perform processing for estimating a navigation map used for DWI reconstruction according to various techniques described herein. The pulse sequence may generally describe the order and timing of the operation of the RF transmit and receive coils 126 and the gradient coils 128 to acquire the resulting MR data. In a non-limiting example, the pulse sequence may indicate the order and duration of the transmit pulses, the gradient pulses, and the acquisition time of the receive coils to acquire the MR data.

[0062] The pulse sequence can be organized into a series of cycles. In a non-limiting example, the pulse sequence can include a preprogrammed number of pulse repetition cycles, and applying the pulse sequence can include causing the MRI system to operate according to the parameters of the pulse sequence for up to the preprogrammed number of pulse repetition cycles. In each cycle, the pulse sequence can include parameters for generating RF pulses (e.g., parameters identifying transmit duration, waveform, amplitude, phase, etc.), parameters for generating gradient fields (e.g., parameters identifying transmit duration, waveform, amplitude, phase, etc.), timing parameters for controlling when to generate RF and / or gradient pulses and / or when (one or more than one) receiving coils are configured to detect MR signals generated by the subject, and other functionality. In some embodiments, as described herein, the pulse sequence can include parameters specifying one or more navigation RF pulses.

[0063] Examples of pulse sequences include zero echo time (ZTE) pulse sequence, balanced steady-state free precession (bSSFP) pulse sequence, gradient echo pulse sequence, inversion recovery pulse sequence, DWI pulse sequence, spin echo pulse sequence (which includes conventional spin echo (CSE) pulse sequence, multi-segment FSE pulse sequence, fast spin echo (FSE) pulse sequence, turbo spin echo (TSE) pulse sequence or any multi-spin echo pulse sequence (such as diffusion-weighted spin echo pulse sequence, etc.)), inversion recovery spin echo pulse sequence, arterial spin labeling pulse sequence, Overhauser imaging pulse sequence, and others.

[0064] When capturing a multi-segment FSE pulse sequence, the controller 106 can control the transmit and receive coils 126 to generate a series of RF excitation pulses separated by time intervals (during these time intervals, the controller 106 controls the gradient coils 128 to apply phase encoding gradients). The total number of phase encoding steps can be divided into smaller segments, where each segment corresponds to a portion of k-space. Each segment can be acquired using a separate RF excitation pulse, and k-space data for each segment can be acquired and stored by the controller 106. The controller 106 can then combine the k-space data from all segments to reconstruct the final image.

[0065] The controller 106 can initiate an example multi-segment FSE imaging process by first generating a strong magnetic field using the B0 magnet 122. The strong magnetic field can align hydrogen protons in the tissue. In some implementations, the B0 magnet 122 can be a permanent magnet that is always in effect. In some implementations, such as when the B0 magnet 122 is an electromagnet, the controller 106 can activate the B0 magnet 122. The controller 106 can then initiate RF pulses via the transmit and receive coils 126 to excite the protons. The controller 106 can activate the transmit and receive coils 126 to generate a series of RF refocusing pulses applied to the tissue, thereby producing a plurality of corresponding echoes. The phase and phase of the RF pulses can be modified according to the techniques described herein.

[0066] After each excitation caused by the RF refocusing pulse, the controller 106 can use the gradient coil 128 to apply a phase encoding gradient to the tissue being imaged. The phase of the phase encoding gradient can be changed simultaneously with the phase of the RF excitation pulse. The controller 106 can change the amplitude of the phase encoding gradient for each excitation, thereby allowing spatial information to be encoded along the phase encoding direction. The controller 106 can then acquire k-space data for each segment separately. As described herein, each segment can correspond to a portion of k-space. The total number of segments depends on the number of phase encoding steps utilized for the desired spatial resolution. The k-space data for the image can be stored in a memory of the controller and, in some implementations, provided to the computing device 104. The controller 106 can then use the k-space data to perform an image reconstruction process to generate an image domain image of the scanned tissue. Although the examples herein provide features of FSE imaging processing, other types of imaging techniques used by MRI systems can be used to generate images of tissue of a subject associated with at least one body part.

[0067] The MRI system 100 may include one or more external sensors 178. The one or more external sensors may assist in detecting one or more error sources (e.g., motion, noise) that degrade image quality. The controller 106 may be configured to receive information from the one or more external sensors 178. In some embodiments, the controller 106 of the MRI system 100 may be configured to control the operation of the one or more external sensors 178, as well as to collect information from the one or more external sensors 178. The data collected from the one or more external sensors 178 may be stored in a suitable computer memory and may be used to assist various processing operations of the MRI system 100.

[0068] The MRI system 100 may be a portable MRI system, and thus may include a portable subsystem 150. The portable subsystem 150 may include at least one power subsystem 152 and at least one motorized transport system 154. The power subsystem 152 may include any device or system that enables or supports the portability of the MRI system 100. In a non-limiting example, the power subsystem 152 may include any of the functionalities of the power supply 112, and may also include other circuit systems that enable power to be provided, including but not limited to batteries and associated circuit systems, AC-DC converters, DC-DC converters, switching power converters, voltage regulators, or battery charging circuit systems, and others. The power subsystem 152 may include connectors that support the portability of the MRI system 100, such as connectors and cables of suitable sizes used for portable systems, etc. In some implementations, the power subsystem 152 may include circuit systems that provide power to the MRI system 100. In some implementations, the power subsystem 152 may include circuitry or connectors that enable the MRI system 100 to receive power from one or more electrical outlets (which may include standard electrical outlets).

[0069] The motorized transport system 154 may include any device or system that enables the MRI system 100 to be transported to different locations. The motorized transport system 154 may include one or more components configured to facilitate moving the MRI system 100 to a location where MRI is required. In some implementations, the motorized transport system 154 may include a motor coupled to a drive wheel. In such an implementation, the motorized transport system 154 may provide motorized assistance when transporting the MRI system 100 to one or more locations. The motorized transport system 154 may include a plurality of casters to assist support and stability and facilitate transport.

[0070] In some implementations, the motorized transport system 154 includes motorized assistance controlled using a controller (e.g., a joystick or other controller that can be manipulated by a person) to guide the portable MRI system during transport to the desired location. The motorized transport system 154 may include a power-assisted circuit system (e.g., including an accelerometer and a vibration sensor, etc.) that detects when a force is applied to the MRI system, and in response, engages the motorized transport system 154 to provide motorized assistance in the direction of the detected force. In some implementations, the motorized transport system 154 may detect when a force is applied to one or more parts of the MRI system 100 (e.g., by an operator pushing an armrest, a housing, etc. of the MRI system 100 or applying a force to an armrest, a housing, etc. of the MRI system 100), and in response, provides motorized assistance to drive the wheels in the direction of the applied force. Thus, the MRI system 100 can be guided to a location where MRI is required. The power subsystem 152 can be used to provide power to the MRI system 100 (including the motorized transport system 154).

[0071] In some implementations, the motorized transport system 154 may include a safety mechanism for detecting a collision. In a non-limiting example, the motorized transport system 154 may include one or more sensors for detecting contact forces with another object (e.g., a wall, bed, or other structure). Upon detecting a collision, the motorized transport system 154 may generate a signal to one or more motors or actuators of the motorized transport system 154 to cause a motorized motion response away from the source of the collision. In some implementations, the MRI system 100 may be transported by having a person use manual force to move the system to a desired position. In such an implementation, the motorized transport system 154 may include wheels, bearings, or other mechanical devices that enable the MRI system 100 to be repositioned using manual force.

[0072] like Figure 1 As illustrated, the computing device 104 may communicate with the controller 106, for example, to receive MR data. The computing device 104 may be configured to process the MR data from the controller 106. In a non-limiting example, the computing device 104 may process the received MR data using any suitable image reconstruction process to generate one or more MR images. Additionally or alternatively, the controller 106 may process the received MR data for image reconstruction, and the reconstructed image may be provided to the computing device 104, such as for display, etc. The controller 106 may provide information related to one or more pulse sequences to the computing device 104 for the computing device to process the data.

[0073] The computing device 104 may be any electronic device configured to process the acquired MR data and generate one or more images of the subject being imaged. The computing device 104 may include at least one processor and a memory (e.g., a processing circuit). The memory may store processor-executable instructions that, when executed by the processor, cause the processor to perform one or more of the operations described herein. The processor may include a microprocessor, an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), a graphics processing unit (GPU), a tensor processing unit (TPU), etc. or a combination thereof. The memory may include, but is not limited to, an electronic, optical, magnetic, or any other storage or transmission device capable of providing program instructions to the processor. The memory may also include a floppy disk, a CD-ROM, a DVD, a disk, a memory chip, an ASIC, an FPGA, a read-only memory (ROM), a random access memory (RAM), an electrically erasable programmable ROM (EEPROM), an erasable programmable ROM (EPROM), a flash memory, an optical medium, or any other suitable memory from which the processor can read instructions. The instruction may include code generated from any suitable computer programming language. The computing device 104 may include any or all of the components and may be combined Fig. 20 Any or all of the functions of the computing system 2000 described above. In some implementations, the computing device 104 can be located in the same room as the MRI system 100, or coupled to the MRI system 100 via a wired or wireless connection. In some implementations, the computing device 104 can be remote from the MRI system 100 and / or the controller 106 (e.g., in a different location) and configured to receive data or information via a network.

[0074] In some implementations, the computing device 104 may be a fixed electronic device, such as a desktop computer, a server, a rack computer, or any other suitable fixed electronic device that can be configured to process MR data and generate one or more images from MR signals captured using the MRI system 100. Alternatively, the computing device 104 may be a portable device, such as a smart phone, a personal digital assistant, a laptop computer, a tablet computer, or any other portable device that can be configured to process MR signal data. In some implementations, the computing device 104 may include multiple computing devices of any suitable type, as the various aspects of the disclosure provided herein are not limited in this respect. In some implementations, operations described as being performed by the computing device 104 may be performed by the controller 106 instead, or vice versa. In some implementations, certain operations may be performed by both the controller 106 and the computing device 104 via communication between the devices. For the purposes of the examples herein, the computing device 104 may execute instructions to simulate structures or images to improve model performance.

[0075] Figure 2 An example pipeline 200 for simulating a structure for at least one image is depicted according to one or more implementations. As described herein, the computing device 104 (or in some cases, the controller 106) can be configured to simulate various structures or images, for example, by performing features or operations of the example pipeline 200. The example pipeline can include various operations to be performed by the computing device 104 for structure or image simulation, such as, but not limited to, at least operations 208 to 224. One or more of these operations can be part of at least one of the lesion generation stage 202, the lesion transfer stage 204, and / or the result stage 206. The example pipeline 200 can include other operations or stages for performing the techniques described herein.

[0076] As an initial step, at operation 208, the computing device 104 may receive / obtain / acquire at least one image of a subject / patient. For purposes of the example used to simulate a lesion, the image may be from a healthy subject (e.g., an image of healthy tissue or structure), although in some other cases an image with an abnormality may be utilized. The computing device 104 may obtain the image from the controller 106. In some cases, the computing device 104 may retrieve the image from a data repository (such as a database on a server or a storage device of the controller 106, etc.). The computing device 104 may use an image that can show at least one body part of the subject (such as a brain, etc.). The image may be a 2D image or a 3D image. The image may include layers or slices of a body part, such as a 2D slice of a 3D image, etc.

[0077] In some aspects, the computing device 104 may use the image to generate a lesion (e.g., in the lesion generation stage 202). In some aspects, the computing device 104 may use the image for lesion transfer (e.g., in the lesion transfer stage 204). In various aspects, for example, the computing device 104 may use the image for a simulation mask (e.g., a lesion mask) to apply or embed the simulated lesion mask on the image. The image may be processed sequentially and / or concurrently at various stages or operations of the example pipeline 200. In some cases, when the mask is applied to the image (e.g., the input image or the first image), the computing device 104 may generate another image (e.g., an image with a simulated lesion, sometimes referred to as an output image or the second image) as part of the result stage 206. For example, the output image may include an appearance similar to the input image, wherein the applied simulated lesion is at a location on the input image. For purposes of example, the operations of the example pipeline 200 may be used to simulate at least one lesion (e.g., an abnormality) in the brain. In some other aspects, the operation of the example pipeline 200 can be used to simulate other structures (such as but not limited to healthy tissue, etc.) in other parts of the body (not limited to the brain).

[0078] In the lesion generation phase 202, the computing device 104 may perform operations 210 to 214 for generating / creating an example lesion (e.g., a simulated lesion mask). The lesion mask may be manually created by the computing device 104. At operation 210, the computing device 104 may perform a regional classification for at least one anatomical region associated with a body part (e.g., at least a portion of a body part). Figure 3 to Figure 4 At least one of (but not limited to) Figure 3 to Figure 4 ) to illustrate operation 210. In various configurations, the computing device 104 may perform at least one or a combination of the following example methods / approaches / techniques for lesion generation:

[0079] Anatomical regions of the brain or cerebrovascular regions (or anatomical regions of other body parts);

[0080] District growth pathways; and / or

[0081] Data-driven pathology mask generation.

[0082] To generate example lesions using brain region anatomy, computing device 104 may identify regions associated with body parts. Figure 3 Depicts an example of brain region classification according to one or more implementations. As shown, Figure 3An example image 300 or slice (e.g., 2D slice) of a brain, for example, from an input image, is included. In some cases, the image 300 may be a plurality of slices of the brain in 3D form. The computing device 104 may process each image 300 to separate / divide or extract a plurality of vascular regions (e.g., 26 vascular regions) associated with the image 300, for example, extracting regions from the brain. The computing device 104 may utilize at least one suitable regional classification or mapping technique to separate regions of a body part, such as anatomical landmarks, voxel-based (or pixel-based) morphometry (VBM), segmentation algorithms, diffusion tensor imaging (DTI) and / or functional connectivity analysis, and at least one of others. For example, regions may include, but are not limited to, major left and right vascular branches, such as the middle cerebral artery (MCA), the anterior cerebral artery (ACA), the posterior cerebral artery (PCA), the anterior inferior cerebellar artery (AICA), the posterior inferior cerebellar artery (PICA), the superior cerebellar artery (SCA), vascular regions, and the like. One or more than one region may be part of at least one anatomical region.

[0083] Based on these separated regions, computing device 104 may generate or obtain a region map for each image 300. The region map may be included as a portion of each image 302 associated with the respective image 300, wherein each contrasting grayscale portion of image 302 may represent a respective region (e.g., a unique vascular region) associated with a body part. The generated image 302 may be in 2D or 3D form.

[0084] At operation 212 , the computing device 104 may identify or select a location (eg, a lesion area) for sampling. Figure 4 Depicting an example of at least one region selected as a target location for simulating a lesion according to one or more implementations. For example, computing device 104 may obtain image 400 for processing. Computing device 104 may separate various regions from image 400 to generate image 402 including a map of these regions. In response to obtaining or detecting the regions, computing device 104 may select at least one of these regions as a target lesion region (e.g., a region to which a simulated lesion is applied). For example, computing device 104 may sample at least one region uniformly or using a customized or predetermined multinomial distribution. Uniformly sampling a region may refer to selecting elements or data points from a region so that each element has a similar probability of being included as part of a selected sample (in this case, a location). Sampling a region using a customized multinomial distribution may involve selecting elements or features from a region using a distribution configuration that is customized for certain characteristics or proportions of elements within the region, thereby allowing certain elements of the region to be more or less apparent as part of the selected region. In some cases, for example, if certain regions are preferred or desired relative to other regions, computing device 104 may use a customized multinomial distribution for region selection. As Figure 4 As shown, in response to sampling at least one region, the computing device 104 may select an associated region (eg, location) for the target lesion. Figure 4 An example region (or area) selected as a location for a target lesion is shown in example image 404 of FIG.

[0085] In some implementations, the computing device 104 may be configured to select individual regions (or zones) for targeting at least one lesion. For example, during a first cycle of performing operations of the example pipeline 200, the computing device 104 may select a first region for applying a first simulated lesion. In a second cycle, the computing device 104 may select a second region for applying a second simulated lesion, and so on.

[0086] At operation 214 , in response to selecting or identifying a location, the computing device 104 may simulate a shape (eg, including a size) of an example lesion (eg, for lesion generation). Figure 5 Depicts an example 500 of lesion masks generated in different shapes and sizes according to one or more implementations. To create variations in the shapes and sizes of example lesions, computing device 104 may generate shapes (e.g., 2D shapes) controlled by or according to one or more parameters (such as a major axis and / or a minor axis, etc.). In this case, the major axis may refer to the longest dimension or alignment of a structure or lesion. The minor axis may refer to a cross-sectional view of a structure perpendicular to the major axis. One or more parameters may define the size of the example lesion.

[0087] The shape may be an ellipse, a circle, a sawtooth, a ring, a crescent, etc. Computing device 104 may select the shape based on an indication or configuration by user 102. In some cases, computing device 104 may select the shape from the list of shapes using any suitable selection technique, such as random selection, sequential selection (e.g., for multiple simulations of a lesion mask), weighted selection, etc.

[0088] For example, the size of the example lesion may be within one or more parameters, which may indicate the maximum (or minimum) size of the example lesion. The one or more parameters may be based on the selected location (e.g., sampling area) to which the example lesion is to be applied. For example, the computing device 104 may determine the dimensions (including the major axis and / or minor axis) associated with the selected area or location of the body part. For example, these major axes and / or minor axes may be used to define the size of the example lesion so that the size of the example lesion does not exceed the size of the selected lesion. In some cases, the major axis and / or minor axis may be configured by the user 102.

[0089] Subsequently, the computing device 104 may simulate an example lesion (e.g., simulate a lesion mask) using, at least in part, the selected or determined shape and size (e.g., at operation 222, etc.). For example, column A (e.g., the first column) of example 500 may provide example images of lesion masks from the same area having different shapes and / or sizes. In various configurations, using the generated shapes and sizes (such as in column A of example 500, etc.), the computing device 104 may apply at least one type of distortion for creating a change in the shape and size of the example lesion. The computing device 104 may apply at least one type of distortion using at least one suitable image warping or image deformation technique (such as a mesh-based deformation, a thin plate spline, a free-form deformation, etc.). The type of distortion may include, but is not limited to, at least one of an elastic distortion, an elastic deformation, a field inhomogeneity, a ghosting artifact, a slice contour artifact, and / or a magnetically induced distortion, etc. Examples of distortions (e.g., elastic distortions) applied to shapes and sizes may be shown in at least the example images of columns B to C of example 500. By applying distortions, the shape and size of the example lesion may be further modified / changed.

[0090] In some configurations, to simulate a lesion mask using a region growing approach / technique / method, the computing device 104 may create / generate or simulate a lesion mask by using at least one suitable region growing technique (such as watershed, seed filling, region merging, etc.). By utilizing the region growing approach, the computing device 104 may maintain / keep the lesions within anatomically consistent regions (e.g., within the same region or associated regions of the body part) by isolating (or grouping) regions / locations of the body part with similar intensities. For example, at operation 210, the intensity may be determined based on a brain region classification.

[0091] For example, computing device 104 may initialize or place at least one seed within an input image (e.g., a brain MRI image) at a desired location (e.g., such as the selected location at operation 212). The seed may correspond to a starting point for lesion growth (e.g., growing a lesion mask). In some cases, the seed may represent a location from which an abnormality begins to grow. The number of seeds may be adjusted / updated based on the desired number of lesion locations / regions to grow. In some cases, for example, the seed may be represented by the intensity or contrast of a pixel or voxel within the image.

[0092] In response to placing a seed, computing device 104 may iteratively grow a region around the seed (e.g., start a region growing process). If there is more than one seed, computing device 104 may iteratively grow multiple regions associated with the corresponding seeds. Growing a region around a seed may include extending the region to voxels adjacent or neighboring the seed and other previously grown regions. The region may extend to neighboring voxels having similar intensities (e.g., voxel intensities) as the voxel in which the seed is placed. By growing the region to voxels having similar intensities, computing device 104 may ensure that an example lesion used to simulate a lesion mask is within at least one selected location or region. The number of iterations for growing the region may be predetermined or configured by user 102.

[0093] In some implementations, the number of iterations may be based on at least one of the selected region (such as the type of anatomical region or the size of the region, etc.) and predetermined parameters (such as the major axis and / or the minor axis, etc.). For example, the computing device 104 may iteratively grow the region around the seed until the size of the expanded region is or is approximately a certain percentage (such as 10%, 20%, or 30%) of the size of the region. In another example, based on the type of anatomical region selected, the computing device 104 may determine an associated size (e.g., a maximum size) or a number of iterations to expand the region around the seed. In yet another example, the computing device 104 may continue to grow adjacent regions until the major axis and / or the minor axis parameters are met. Specifying or configuring the number of iterations may allow control over the final size of the lesion mask.

[0094] The computing device 104 may apply at least one suitable region growing technique to separate growth regions (e.g., locations of body parts for growth regions) at boundaries where the change in intensity is equal to or above a threshold. The threshold may be predetermined by the user 102. The change in intensity may be compared between a first intensity at a seed location or at an expansion zone around the seed and a second intensity at a potential location for growing the region. For example, the intensity may be represented by a value ranging from 0 to 1, where 0 may represent the lowest intensity and 1 may represent the highest intensity, or vice versa. For example, if the threshold is set to 0.2 and the intensity at the seed is 0.5, the computing device 104 may separate growth regions at boundaries having an intensity above 0.7 or below 0.3.

[0095] By detecting changes in intensity, computing device 104 can prevent a growing region (e.g., as part of a simulated lesion mask) from spreading into regions of dissimilar intensity, thereby maintaining the region within the same anatomical region or regions. In response to iteratively growing the region, computing device 104 can generate a lesion mask (e.g., a binary lesion mask) based on the growing region around the seed (such as at operation 222, etc.). The lesion mask can simulate a lesion growing at, around, or within a body part while respecting / conforming to anatomical boundaries according to the intensity of the image.

[0096] In some configurations, in order to simulate a lesion mask using data-driven pathology mask generation, the computing device 104 may use historical (e.g., existing) lesion data to generate a lesion mask. In this case, the computing device 104 may make full use of or utilize at least one suitable ML technique (such as a generative adversarial network (GAN), a variational autoencoder (VAE), an autoregressive model, a deep Boltzmann machine (DBM), etc.), and a database with historical images. Depending on the type of structure to be simulated (e.g., healthy tissue or lesions), these historical images may include brain images with and / or without abnormalities (e.g., lesions). Using the techniques for generating lesion masks discussed herein can provide realistic representations of structures of body parts, such as healthy brain tissue or brain lesions, among others.

[0097] For example, the computing device 104 may train the model using at least one suitable machine learning technique using historical images of lesions. The computing device 104 may feed historical images as training data (or input data) used by the model. Using machine learning techniques, the model may learn various features, patterns, and / or characteristics associated with various structures of a body part, such as shapes, sizes, or other details associated with lesions. The model may be composed of or include a generator that is configured to generate / create at least one lesion mask based on the trained data. The generated lesion mask may mimic or simulate a real mask associated with the training data, such as having a shape, size, and / or other details similar to or comparable to a real lesion. In some cases, the model may perform at least one other machine learning technique for adjusting certain features of the simulated lesion mask.

[0098] In addition, the model can be composed of or include a discriminator that is configured to evaluate the fidelity associated with the generated lesion mask. For evaluation, the model can be configured with a metric or evaluation criterion, such as a structural similarity index (SSIM), mean square error (MSE), or peak signal-to-noise ratio (PSNR) associated with the lesion. The model can pre-process and / or normalize the data by configuring or defining the data range, removing outliers, and / or applying appropriate transformation techniques to ensure that the data is in a comparable format, such as between the actual / real lesions and the simulated example lesions. The model can perform at least one of visual inspection, statistical analysis, quantitative measurement, or other operations (not limited to those herein) to identify similarities and / or differences between the simulated example lesions and the real lesions (or real masks).

[0099] In some implementations, the computing device 104 may execute a model for refining the simulated lesion mask. For example, the model may apply at least one post-processing technique or operation to improve the quality, anatomical consistency, etc. of the generated lesion mask. Post-processing techniques for fidelity improvement may include, but are not limited to, texture mapping, post-processing filtering, lighting and shadow configuration, noise addition, or in some cases motion blur, among others. Post-processing techniques may, for example, enhance the realism of the simulated lesion mask by introducing characteristics (e.g., defects) generated by real-world data (such as when an image is captured via an imaging device, etc.), movement, or lighting effects.

[0100] In some implementations, such as in the lesion transfer stage 204, etc., the computing device 104 may transfer one or more historical lesions to the input image. In some implementations, the simulated lesion mask from the lesion transfer stage 204 may be used independently of the simulated lesion mask from the lesion generation stage 202. In some other implementations, the simulated lesion mask from the lesion transfer stage 204 may be used in combination with the simulated lesion mask from the lesion generation stage 202, such as including multiple lesion masks from both the lesion generation stage 202 and the lesion transfer stage 204, etc.

[0101] To transfer the lesion, at operation 220, the computing device 104 may obtain one or more historical images with annotated structures (e.g., annotated lesions and / or healthy tissue) from a data repository. The data repository may be local to the computing device 104 or remote from the computing device 104. The annotated structure may refer to data embedded in or associated with a corresponding historical image, which may indicate various details associated with the lesion presented in the historical image. For example, the computing device 104 may retrieve annotations of at least one of shape, size, location, color, etc. associated with the lesion.

[0102] At operation 216, computing device 104 may transfer or extract the lesion shape from the historical image. At operation 218, computing device 104 may transfer or extract the location of the lesion from the historical image. By using the annotated lesions from the historical image, the lesions may be relatively similar or closely matched to the real lesions (e.g., real shape, size distribution, etc.). In some cases, computing device 104 may transfer the lesion shape, size, and / or location, as well as other details, from the historical image to the input image as part of the simulation mask (such as at operation 222, etc.). In order to compare between the volume of the real lesion and the volume of the simulated lesion from the lesion transfer stage 204, Figure 6 Example graphs 600, 602 of volume distribution for real lesions and simulated lesions with lesion transfers are depicted according to one or more implementations. For example, graph 600 can illustrate the size or volume distribution for real lesions from historical images. Graph 602 can illustrate the size or volume distribution for simulated lesions with lesion transfer operations. By comparison, for example, the volume distribution between real lesions and simulated lesions can be comparable because the simulation is based on annotation information of the real lesions.

[0103] In some configurations, the initial mask (e.g., lesion mask or normal mask) simulated by the computing device 104 at or as part of operation 222 may be in 2D. In this case, the computing device 104 may expand or enlarge the 2D simulated mask into a 3D simulated mask. Figure 7 Depicted are examples of 2D slices 700-706 (e.g., 2D simulated masks) including lesion generation for interpolation according to one or more implementations. For example, for a lesion generation path to extend the 2D simulated mask, computing device 104 may interpolate the 2D lesion mask across slices of 3D, such as extending the 2D mask across multiple adjacent slices of the 3D volume of the image, etc.

[0104] like Figure 7 As shown, at least one of the 2D slices 700 to 706 may correspond to a key slice (e.g., the first 2D simulation mask from the lesion generation stage 202 or the lesion metastasis stage 204). The computing device 104 may interpolate to at least one side (e.g., at least one of the left, right, top, and / or bottom) of the key slice to generate Figure 7 700 to 706 of the example 2D slices. At least one side may be configurable by the user 102 or according to the key slice. For example, if the key slice is simulated from the left side of the lesion, the computing device 104 may interpolate from left to right to generate the 3D simulation mask. If the key slice is simulated from the right side of the lesion, the computing device 104 may interpolate from right to left to generate the 3D simulation mask, and so on. In such a case, Figure 7The 2D slices 700 to 706 may represent various layers of at least a portion of a 3D simulation mask. For the purpose of providing an example, the 2D slices 700 to 706 may represent examples of 2D slices that are interpolated from left to right, respectively, to form at least a portion of a 3D lesion.

[0105] In some implementations, for the simulated lesion mask from the lesion transfer stage 204, the computing device 104 may simulate a 3D mask by transferring the shape, size, and / or location of the historical lesion. In such a case, the computing device 104 may not need to perform a 2D simulated mask for extending the lesion mask from the lesion transfer stage 204.

[0106] In response to the simulated lesion mask, the computing device 104 may enter the result stage 206. At operation 224 of the result stage 206, the computing device 104 may simulate the appearance of the simulated lesion mask. For example, the lesion mask from operation 222 may indicate the location and / or shape of the simulated lesion. Using the simulated lesion, the computing device 104 may determine the appearance of the lesion, including but not limited to the pixel or voxel intensity for the simulated lesion. In order to maximize or optimize the fidelity in the simulated lesion mask, as described herein, the computing device 104 may perform at least one of a feature engineering configuration (or approach) and a data-driven configuration for the appearance of the simulated lesion. For example, the appearance of the lesion may be part of the lesion mask.

[0107] Example Feature-Engineering Configuration

[0108] In various configurations, as part of a feature engineering configuration for simulating appearance, the computing device 104 may select an average (or mean, median, mode, etc.) pixel or voxel intensity of a lesion for simulating lesion appearance. For example, depending on the application context (e.g., the type of abnormality in a body part), the location (or anatomical region) associated with the lesion, etc., the lesion may be at least one or a combination of high signal, low signal, equal signal, etc. The computing device 104 may determine or select an average intensity to simulate for a lesion mask according to or based on the type of abnormality. Each type of abnormality for a particular body part may be stored in a lookup table or associated with a corresponding average intensity for various images. For example, in some cases, a white matter lesion on a fluid attenuated inversion recovery (FLAIR) may be presented as a high signal in an image, and a stroke lesion on an apparent diffusion coefficient (ADC) may be presented as a low signal.

[0109] Because different abnormalities can be associated with varying intensities (e.g., which can depend on the intensity of other parts of the image), computing device 104 can utilize the techniques or mechanisms discussed herein to provide customization of the intensity level for simulated lesions. For example, computing device 104 can consider the intensity distribution (e.g., mean, median, mode, etc.) of a 3D series of images. A 3D series can include consecutive 2D images or slices constructed to form a 3D image. The intensity distribution can be separated / divided into multiple intervals or categories, such as the following examples, etc.:

[0110] ·very_low:(min,low_1_percentile)

[0111] ·low:(low_1_percentile,low_2_percentile)

[0112] ·high:(top_2_percentile,top_1_percentile)

[0113] ·very_high:(top_1_percentile,volume_max)

[0114] The intensity distribution may be separated into one or more sources (e.g., one or more intervals, sets of values, or in some cases overlapping intervals). In some implementations, the intensity associated with the simulated lesion mask may be sampled (e.g., randomly) from one or more of the intervals, such as using a predetermined (e.g., user-defined) probability configuration. For example, to simulate a stroke lesion in DWI, the computing device 104 may obtain a probability configuration including a sampling probability of p=(0, 0, 0.05, 0.95) such that the likelihood / probability of giving the lesion a high signal relative to the rest of the brain tissue or other portions is relatively high. Depending on the selected probability configuration, the resulting stroke lesion intensity distribution for the appearance of the lesion may be similar to the corresponding real lesion.

[0115] For example, Figure 8An example bar graph 800 is depicted of an intensity distribution for a lesion on DWI according to one or more implementations. For example, the example bar graph 800 can be derived or constructed from real DWI data. The range of pixel or voxel intensities in the example bar graph 800 can be normalized to a range of 0 to 1. In various aspects, in the case of a lesion mask, intensity values ​​can be assigned to the lesion mask for insertion into an input image. To determine the intensity values ​​to assign, the computing device 104 can receive one or more intervals used for sampling, where each interval can include a set of intensities (e.g., intensity values). Using the set of intensities used for sampling, a sampling probability distribution (e.g., a resulting distribution) can be determined so that when calculating certain statistics, the same probability distribution as the lesion mask is used. Figure 8 The statistics from the real DWI lesions to the synthetic / simulated DWI lesions are shown to be similar, and the distribution (e.g., the mean of intensity and volume) can be the same between the real DWI lesions and the simulated DWI lesions. When the intensity distribution of the simulated lesions is similar to the intensity distribution of the real lesions, the computing device 104 can determine that the simulated lesions with (one or more than one) intensity values ​​have relatively high fidelity, and the corresponding interval set can be used to determine the intensity value. Otherwise, if the intensity distribution of the simulated lesions is not similar to the intensity distribution of the real lesions, the computing device 104 can determine that the simulated lesions with (one or more than one) intensity values ​​have relatively low fidelity. As shown, for real stroke lesions on DWI with a sampling probability of p = (0, 0, 0.05, 0.95), there may be a significantly increased DWI lesion volume between the region IDs ranging from 0 to 5 and 13 to 17 compared to other region IDs. The region ID can represent a number from a range representing different anatomical regions (e.g., in this case, 1 to 26), where each region ID can be associated with a corresponding region of the body part, such as 1 for the left ventricle, 2 for the right ventricle, and so on.

[0116] In some implementations, computing device 104 may add one or more patterns (e.g., edema patterns) to a lesion to simulate an appearance. Computing device 104 may additionally or alternatively add one or more patterns to enhance or simulate lesion intensity. For example, for certain types of abnormalities (such as edema, etc.), the effect of edema may be applied or added to the simulated example lesion as part of the appearance. The effect may be represented by a pattern. Fig. 9Depicting an example of a simulated edema effect around a lesion according to one or more implementations. As shown, computing device 104 can simulate the lesion mask in example image 902 for at least one anatomical region or area of ​​the body part in example image 900. As shown in example image 904, computing device 104 can simulate a pattern of edema, for example, by creating a region with relatively low intensity around at least a portion of the simulated lesion mask. For example, computing device 104 can provide the pattern of edema as a layer below the lesion mask or around the example lesion. Computing device 104 can employ different intensity levels from the lesion core, for example, additional simulated lesions (such as Fig. 9 In some embodiments, the example edema mask in 906 may include different intensity values ​​than the first lesion applied / inserted in the image. In response to modeling the pattern as part of the appearance, computing device 104 may apply the appearance to a lesion mask, which may be inserted or applied to the input image, for example, to generate example image 906.

[0117] In some implementations, computing device 104 may add or insert device acquisition noise as part of simulating the appearance of a lesion. Fig.10 Example images 1000-1004 depict acquisition noise in simulated lesions according to one or more implementations. For example, depending on the type of imaging device used to capture the image (e.g., a low-field MRI scanner, a CT scanner, etc.), environmental interference and / or other acquisition noise may be introduced into the image of the subject. Therefore, computing device 104 may simulate the appearance of the simulated lesion mask to have inherent non-uniformity (or certain types of noise) to improve the fidelity associated with the simulation.

[0118] To simulate the appearance of acquisition noise, the computing device 104 may drop or remove pixels or voxels with a probability (e.g., labeled “drop_prob”). By dropping certain pixels or voxels based on the probability, the simulated lesion may appear with a relatively small intensity. The drop probability may range from 0 to 1, where 0 may indicate that no pixels or voxels are to be dropped and 1 may indicate that all pixels and voxels associated with the simulated lesion are to be dropped. The computing device 104 may then apply a smoothing technique (e.g., Gaussian smoothing and other techniques) to the simulated mask (e.g., labeled “σ (sigma)” for the configuration value of the smoothing). By applying the smoothing technique, the computing device 104 may reduce blurry sharp transitions between pixels or voxels of the image. The computing device 104 may apply other processing techniques to mimic the acquisition noise of the imaging device or other noise potentially captured in the image. As Fig.10As shown, the example image 1000 may include a simulated lesion 1006 where drop_prob is 0 (e.g., no pixel dropping) and σ is 2 (e.g., a twice wider blur of pixels, where corresponding 2 pixels in each direction are aggregated or blended to create a smooth transition). In some cases, σ being 0 may provide a step-like boundary between pixels (e.g., no smoothing between pixels). In some other cases, a relatively high σ number (such as 20, etc.) may provide a relatively smooth transition (e.g., a blend of pixel contrast or intensity) between corresponding numbers of pixels in each direction. In various aspects, the computing device 104 may create a smooth transition between healthy tissue and an inserted pathological area. When the drop probability is increased (such as increased to 0.2, etc. in the example image 1002), the simulated lesion 1006 may be considered to have a relatively lower intensity compared to the example image 1000. When the rejection probability is further increased to 0.8 in the example image 1004, the simulated lesion 1006 can be considered to have an even lower overall intensity than the other example images 1000, 1002, thereby blending with other parts of the image (e.g., healthy tissue surrounding the lesion) in a relatively more natural manner. By utilizing the rejection probability, the computing device 104 can simulate a variety of different types of textures. For example, different types of abnormalities can include different textures. By using different rejection probabilities, the computing device 104 can simulate various types of abnormalities used for detection, such as but not limited to stroke, hemorrhage, tumors, etc.

[0119] In some implementations, computing device 104 may add Gaussian noise to the simulated lesion as part of the lesion appearance. Fig.11 Depicted are examples of different noise levels (such as different levels of Gaussian noise, etc.) for an image according to one or more implementations. For example, the computing device 104 may sample Gaussian noise for each lesion mask pixel or voxel (e.g., generate values ​​from a Gaussian distribution) to create a noise signal. The value of the pixel or voxel may be multiplied or calculated according to a constant (e.g., the value of the Gaussian noise) to attenuate or amplify an indication of the noise. Applying Gaussian noise may produce / generate a lesion with a relatively granular appearance (e.g., a lesion that appears granular). In this case, applying Gaussian noise may reduce the smoothness of the simulated lesion mask (e.g., as opposed to applying a discard mask of simulated acquisition noise). The parameters of the Gaussian noise may be based on a maximum pixel value (e.g., a pixel intensity or brightness value of a pixel). For example, the Gaussian noise may be in the range of [0, pixel_value_max]. The computing device 104 may enhance and / or attenuate the simulated lesion intensity, which may change the overall image contrast when scaling the overall image contrast based on the minimum and maximum pixel values ​​(or from the minimum to the maximum pixel value) after lesion insertion.

[0120] For example, Fig.11As shown in the example image 1100 of , for the simulated lesion mask, the noise level can be zero. In this case, the simulated lesion 1106 may be affected by Gaussian noise with a base standard deviation (such as 1, etc.). The base standard deviation can be multiplied by a constant (e.g., 100, 250, etc.) (which can refer to the noise level). The upper bound / upper limit of the constant can correspond to pixel_value_max (e.g., the highest pixel value in the image). As in the example image 1102, the computing device 104 can receive a configuration from the user 102 or obtain a predetermined configuration indicating that the noise level is increased to 100. Increasing the noise level (e.g., the Gaussian noise level) can increase the variability or randomness in the pixel values, thereby resulting in reduced clarity, such as less clear simulated lesions, reduced details, reduced signal-to-noise ratio (SNR) or in some cases overall intensity values ​​and others. In this case, various pixels of the simulated lesions of the example image 1102 may be affected by Gaussian noise. As shown, higher Gaussian noise can result in a reduction in the overall clarity (or intensity in some cases) of the simulated lesions. As in example image 1004, computing device 104 may further increase the noise level to 250. As shown, pixels associated with simulated lesions (e.g., in this case for a noise level of 250) may be further affected by Gaussian noise, thus further reducing the overall clarity of the simulated lesions. The higher the noise level, the lower the overall intensity of the simulated lesions.

[0121] In some implementations, computing device 104 may simulate a mass effect as part of the appearance of a simulated lesion. A mass effect may refer to how an anatomical structure or region of an image deforms or distorts in response to the growth of a lesion or the type of abnormality. For example, for certain abnormalities associated with a body part (such as a tumor, hemorrhage, etc.), in addition to or as an alternative to intensity changes at a target location (or region) of a simulated lesion, a simulated lesion may push or displace tissue surrounding the simulated region due to a mass effect.

[0122] To simulate the mass effect, the computing device 104 may use the contour (e.g., boundary or outline) of the lesion mask (e.g., sometimes referred to as the abnormality mask) to calculate / measure the gradient around the structure and / or determine / estimate the direction of the deformation. The gradient around the structure and the direction of the deformation may be part of the deformation field. The mass effect may reflect the impact of the lesion on the surrounding healthy tissue at a certain point in time, for example, pushing the healthy tissue outward due to the newly formed lesion occupying the space. For example, the computing device 104 may create a deformation field, for example, by sampling the displacement vector value of each voxel from a Gaussian distribution. The computing device may smooth the deformation field by applying a Gaussian kernel spatially. The computing device may adjust or modify the intensity / level / amplitude of the deformation and / or the proportion of the size change to achieve a realistic appearance or effect caused by abnormal growth. For example, the computing device may constrain the deformation within the simulation mask of the anatomical region so that the deformation affects soft tissue but not hard tissue. In various cases, the computing device 104 may use the gradient information of the image to determine the direction of the mass effect. The image gradient direction may be defined as a local difference (e.g., |Img[x,y]-Img[x-1,y]|, |Img[x,y]-Img[x,y-1]|), which may be similar to a local tangent in 2D. The computing device 104 may determine the gradient using other approaches or techniques, not limited to those described herein. Thus, based on the gradient information, the computing device 104 may determine the deformation direction and / or strength of the pixels used for simulation. Thus, for a space occupancy effect simulation, in response to applying a space occupancy (e.g., a simulated lesion) at a selected location to simulate a space occupancy effect, the computing device 104 may calculate the distortion direction and strength for various local pixels or voxels (e.g., neighboring pixels or voxels) near the lesion mask, and apply such a deformation field to the image to simulate the distortion caused by the space occupancy effect.

[0123] Fig.12 An example of a simulated mass effect around a mask according to one or more implementations is depicted. As shown, example image 1200 may present a body part without a simulated lesion. In response to applying a simulated lesion to a selected location, computing device 104 may generate example image 1202, which may include simulated lesion 1204 (e.g., lesion mask). In example image 1202, soft tissue around the simulated lesion may be affected by a mass effect caused by an abnormal growth.

[0124] Sample data-driven configuration

[0125] In some configurations, the computing device 104 may utilize a data-driven configuration to simulate the appearance of a simulated lesion. By utilizing data-driven appearance simulation, the computing device 104 may be configured to inject a simulated lesion into an image (e.g., an input image or a first image) using at least one ML technique. For example, the computing device 104 may execute a model to receive input data including, but not limited to, at least one image (e.g., an input image or an image patch) to be injected with a simulated lesion and a simulated lesion mask (e.g., sometimes referred to as a pathology mask), wherein the simulated lesion mask is obtained using at least one of the techniques for simulating lesion masks discussed herein. The model may be configured to generate an output image, such as an input image injected with a simulated lesion, and the like.

[0126] In some implementations, to perform lesion injection on the input image, the computing device may utilize one or more ML models, such as but not limited to a first model, a second model, and / or a third model, etc. In some cases, these models may be independent models trained or operated using corresponding ML techniques. In some other cases, these models may be, for example, part of a single model (e.g., cyclicGAN, denoised diffusion probability model (DDPM), etc.) trained or operated using at least one ML technique (e.g., at least one of deep learning, neural networks, convolutional neural networks, GANs, etc.).

[0127] For example, the first model can operate as a conditional generator. For example, the first model can take an input image (e.g., labeled "P") and a simulated lesion mask (e.g., labeled "L"). The first model can inject the simulated lesion mask into the input image to generate an output image (e.g., labeled "Z"). For example, to perform the injection, a neural network (e.g., labeled N1, which can be part of the first model) can be trained to take the input image P as input. The neural network N1 can be a convolutional network. In response to obtaining the input image P, the input image P can, for example, traverse a series of convolutional layers to generate one or more images with simulated lesions. Processing for training the network can be described herein. In some cases, during the training of the neural network N1 (or the first model), other models (e.g., discriminator models) can be trained, such as simultaneously or using similar training processes. In some implementations, during inference (e.g., when a data-driven approach is to be used to inject simulated lesions), the first model (e.g., generator) can be used without at least one other model (such as without a discriminator model, etc.).

[0128] The second model can operate as a discriminator configured to distinguish between images. For example, the second model can obtain an output image Z from the first model and obtain at least one reference image from a data set including real (non-simulated) lesions. The second model can perform various operations to distinguish between the output image Z and the real distribution, such as distinguishing between a simulated lesion mask L in the output image Z and real lesions distributed in the reference image. The distinction can be part of the training process. For example, a standard GAN framework can be used to distinguish, such as considering a set of healthy images (e.g., blocks) {L_H1...L_HN} and a set of pathological images (e.g., blocks) {L_A1...L_AN}, the healthy blocks can be fed to the neural network N1 (which can output {Z1...ZN}), etc. The discriminator (e.g., the second model) can obtain the outputs {Z1...ZN} and {L_A1...L_AN} from the neural network N1 in any order. The discriminator can classify whether the image block is generated by the generator or whether the image block is a real sample (e.g., a block from L_A). If the discriminator provides an incorrect classification (e.g., comparing the classification to an expected result), the discriminator can be trained (e.g., further) using backpropagation. Otherwise, the computing device 104 can determine that the discriminator has been trained for deployment. In various cases, the first model and the second model can be trained simultaneously.

[0129] The third model may operate as a segmentation component configured to segment (or attempt to segment) a simulated lesion mask L from the output image Z. By segmenting the simulated lesion mask L, the computing device 104 may provide the segmented simulated lesion mask L as an annotation for the purpose of training or improving the performance of a certain model (e.g., a model used for segmentation, which may be trained using simulated data). The segmentation process may be an extension of the discriminator process (e.g., training for classification of real images compared to simulated images). The various training discussed herein may be used to ensure that the model inserts a shape specified by the lesion mask L. In some implementations, in order to force the model to insert the specified shape, for example, a loss function may be added to penalize the model if the inserted lesion shape is different from the lesion mask L. The additional segmenter may be used for this training, for example, by attempting to segment the lesion from the inserted simulated lesion of the output image Z. If the output segmentation of the third model (e.g., labeled Lz) is different from the lesion mask L, the generator (e.g., the first model) may be trained using back propagation so that the generator may generate a block that more closely resembles the lesion mask L in a subsequent execution cycle. The loss function may include at least CE(L, segmentor(generator(P))), where CE corresponds to cross entropy, among others. Additionally or alternatively, a DDPM may be trained. In response to training the DDPM, the DDPM may be configured to perform feature generation conditioned on an input patch (e.g., an input image P). The DDPM may be trained using, among others, image denoising-based training objectives.

[0130] In some implementations, these models (e.g., the first model, the second model, and the third model) can be jointly trained, for example, the models are trained simultaneously using shared information and / or learning from each other during the training process. In some cases, a historical data set for segmenting lesions from other healthy tissues within the image can be used to pre-train a third model for segmenting the simulated mask L from the output image Z. In some cases, one or more of these models can be trained on at least one other external computing device. In this case, the computing device 104 can obtain / receive (one or more than one) trained model from the external computing device. In some aspects, for example, the computing device 104 can utilize a fourth model to perform a mapping operation from the output image Z to the input image P. For example, by incorporating the fourth model, the computing device 104 can obtain the ability to perform image-to-image conversion using a certain framework (e.g., the cycleGAN framework).

[0131] In some implementations, the computing device 104 may replace the third model with at least one or more different types of loss functions (e.g., CE(L, segmentor(Z)), etc.), such as to ensure that the simulated lesion region can be identified in the generated image, etc. In this case, instead of using a segmenter (e.g., the third model), the computing device 104 may perform another approach, such as calculating a local correlation to ensure that the simulated lesion portion of the simulated lesion image (e.g., the output image Z) has a relatively high correlation with the input lesion mask L, etc. For example, the type of loss function may include, but is not limited to, at least one of a clustering loss, a local cross-correlation, a dice loss, a focal loss, etc. For example, the computing device 104 may utilize a clustering loss function to identify values ​​of pixels or voxels, where the values ​​within the simulated lesion are relatively close to the mean (pixel or voxel) and relatively far from the mean in a region outside the simulated lesion. In another example, by utilizing a local cross-correlation function, the computing device 104 may identify or determine a local correlation between the input image P and the simulated mask L. For example, the local correlation may be in the range from 0 to 1. A relatively high local correlation (such as at least 0.7, etc.) may indicate that the simulated mask L has a pattern, pixel or voxel value (e.g., contrast or intensity feature), and other features that are similar or related to the input image P (e.g., surrounding tissue). A relatively high local correlation may represent or indicate a relatively high fidelity simulation, which may allow the computing device 104 to use the output image Z for model training and performance improvement purposes. A relatively low local correlation (such as less than 0.7, etc.) may indicate that the simulated mask L does not appear to have a pattern or feature that is related to certain tissues from the input image P. A relatively low local correlation may indicate a relatively low fidelity simulation. In such a case, for example, the computing device 104 may not use the output image Z as part of improving model performance.

[0132] In some configurations, the computing device 104 may be configured to simulate hydrocephalus or other types of abnormalities. Fig.13 An example of a hydrocephalus simulation according to one or more implementations is depicted. For example, as shown in example image 1300, computing device 104 can apply / insert / add a simulation of hydrocephalus (e.g., enlarged ventricles) to an image. The input can include the ventricles (e.g., unless missing from the subject). To insert the hydrocephalus, computing device 104 can distort the current ventricular lesions, such as to synthetically enlarge them, etc. The enlarged ventricular lesions can represent or resemble a case of hydrocephalus. Such as incorporating at least Fig.12As described, in response to applying simulated hydrocephalus, the computing device 104 can initiate a space effect simulation operation. The space effect simulation can be shown in the example image 1302. By performing the space effect simulation, the computing device 104 can simulate the abnormality associated with the shape deformation of the lateral ventricle of the hydrocephalus in this case, for example. In a further example, the computing device 104 can calculate the deformation based on historical data including historical patterns observed for the corresponding type of abnormality (for example, in this case, hydrocephalus). The computing device 104 can apply the deformation based on the gradient around the contour of the annotated lateral ventricle (which can be part of a separate area of ​​the brain). In various cases, the deformation for each ventricle can be calculated / determined and applied independently and / or sequentially to simulate the space effect. As shown, as a result of the space effect, hydrocephalus can grow from part 1304 of the example image 1300 to part 1306 of the example image 1302.

[0133] In some cases, computing device 104 may apply healthy tissue similar to the hydrocephalus simulation. For example, instead of simulating hydrocephalus, computing device 104 may inject healthy tissue at the location, determine the shape of the pathological distortion, and simulate the mass effect to distort the local voxels.

[0134] In various configurations, computing device 104 may simulate contrast for an image. Fig.14 An example of an MR image 1400 with different contrasts according to one or more implementations is depicted. For example, using segmentation maps of various anatomical regions (or sub-regions) of a body part and one or more sequence parameters, the computing device 104 can simulate images of different contrasts. Examples of sequence parameters may be described herein. As shown in the example MR image 1400, different anatomical regions may be represented with different contrasts. Furthermore, depending on the sequence parameters, various tissues may be represented with different contrasts, thereby simulating changes in images captured by one or more imaging devices using different configurations / settings, for example.

[0135] Potential regions associated with an image may include, but are not limited to, at least one of the following: background of the image, CSF cerebrospinal fluid, gray matter (GM), white matter (WM), fat, muscle, skull, skin, blood vessels, dura mater, bone marrow, left cerebral white matter, left cerebral cortex, left lateral ventricle, left inferior lateral ventricle, left cerebellar white matter, left cerebellar cortex, left thalamus, left caudate nucleus, left putamen, left globus pallidus, third ventricle, fourth ventricle, brainstem, left hippocampus, left amygdala, left accumbens, left ventral diencephalon (DC), right cerebral white matter, right cerebral cortex, right lateral ventricle, right inferior lateral ventricle, right cerebellar white matter, right cerebellar cortex, right thalamus, right caudate nucleus, right putamen, right globus pallidus, right hippocampus, right amygdala, right accumbens, right ventral DC, etc. In some cases, the sub-region may be any pathology sub-region. Each brain region can be represented as a 2D image or a portion of a 3D image of a probability map, where each pixel can correspond to a corresponding value from 0 to 1 (e.g., normalized). A value of 0 can represent the lowest contrast and a value of 1 can represent the highest contrast, or vice versa. The size of the image can range from [32, 32, 32] to [512, 512, 512] and other sizes.

[0136] Fig.15 An example of a tissue region probability map 1500 for GM, WM, and CSF is depicted according to one or more implementations. In some aspects, such as in conjunction with at least Fig.15 The probability map may indicate the probability that a particular brain region or structure is present at a particular location in the image. In some cases, probability maps from different regions (e.g., from at least one of the GM, WM, or CSF regions) may overlap with each other in terms of pixel locations.

[0137] In some implementations, each tissue marker may have one or more tissue-specific parameters, such as T1, T2, T2*, and / or PD, among others. Tissue parameters may vary according to the magnetic field strength associated with an image (e.g., an input image or a simulated image). For example, tissue-specific parameters for an image associated with a magnetic field strength of 64 mT may be different from tissue-specific parameters for the same image associated with a magnetic field strength of 1.5 T. Each sequence may have at least one sequence-specific parameter, including but not limited to relaxation time (TR), echo time (TE), inversion time (TI), flip angle, and the like. Examples of certain sequence-specific parameters may include, for example, a FLAIR sequence, TR=3000 ms, TE=80 ms, flip angle=90 degrees, and / or TI=1700 s.

[0138] The computing device 104 may generate or simulate image contrast using at least one or a combination of signal formulas / equations. The signal equations may include, but are not limited to, spin echo, spoiled gradient echo, rewrap gradient, time reversal gradient echo, and / or random. These example signal equations may be provided in the following examples:

[0139] Spin echo:

[0140] S=PD(1-2e^(-TI / T1)+e^(-TR / T1))e^(-TE / T2)

[0141] Spoiled Gradient Echo:

[0142] S=PD*sin(a)(1-e^(-TR / T1))e^(-TE / T2*)

[0143] ----------------------------------------------------------

[0144] 1-cos(a)e^(-TR / T1)

[0145] Rewind Gradient:

[0146] S=PD*sin(a)(1-e^(-TR / T1))e^(-TE / T2*)

[0147] ---------------------------------------------------------------------

[0148] [(1-cos(a)e^(-TR / T1)-e^(-TR / T2))(e^(-TR / T1)-cos(a))]

[0149] Time-reversal gradient echo:

[0150] S=pd e^(-(2*TR-TE) / T2)

[0151] random:

[0152] S = pd*uniform(0,1)

[0153] Computing device 104 may generate images with different contrasts as follows. Fig.16 Depicted are examples of the same brain generating different contrasts according to one or more implementations using the spin echo signal equation and according to different TR, TE and / or TI, and other parameters. Fig.16The example image 1600 includes the same brain image with different contrasts generated by the computing device 104. For example, the computing device 104 can select or use an input image including an image volume to generate different contrasts. The computing device 104 can select multiple tissue regions to be used (e.g., regions such as WM, GM, CSF, etc., where one or more of the regions can be selected or randomly selected by the user 102). In some cases, the tissue region can include an anatomical mask or be similar to an anatomical mask. In some cases, a contrast simulation can be used to determine pixel values ​​for each of these regions. The computing device 104 can generate a final image (e.g., an output image) by adding these region masks to the supplied / provided intensity values ​​based on the contrast simulation. The computing device 104 can select at least one signal equation and one or more sequence parameters, such as TR, TE, TI and / or flip angle and others. In some cases, the computing device 104 can randomly select a signal equation. In some other cases, the computing device 104 can sequentially select a signal equation from a list of signal equations. In this case, different signal equations can be selected for each cycle of the image generation process. In some other cases, the computing device 104 may receive instructions from the user 102 to utilize at least one predetermined signal equation. For the purpose of providing an example, the echo signal equation may be used to generate the example image 1600, although other signal equations may be similarly utilized herein. For each tissue, the computing device 104 may load at least one of a 2D or 3D segmentation probability map and tissue parameters. For each pixel within the probability map, the computing device 104 may calculate a signal value (e.g., labeled "S") based on the signal equation, tissue parameters, and sequence parameters. In response to the calculation, the computing device 104 may generate an S map for each tissue, wherein the S map may be the same size as the input segmentation map. For various tissues, the computing device 104 may generate multiple S maps. Therefore, the computing device 104 may combine or aggregate the S maps for the tissue to generate or produce (e.g., ultimately) a combined image. The computing device 104 may repeat the process to, for example, generate other contrasts for the input image using different values ​​for the sequence parameters or different sequence parameters.

[0154] The computing device 104 may use another signal equation to generate an image with different contrast. For example, Fig.17 Depicts an example of contrast generated using a random signal equation according to one or more implementations. In this case, computing device 104 may utilize, for example, a combination of at least Fig.16 Similar operations as described above are used to generate example image 1700. For example, instead of Fig.16The computing device 104 may utilize a random signal equation to generate the example image 1700 based on the spin echo equation of FIG. 1700. The use of the signal equation may not be limited to any particular brain region, such as shown in the example image 1700 including various regions of the brain. For example, the computing device 104 may use the random signal equation to generate different contrasts for any region. In addition, the computing device 104 may use other signal equations (e.g., involving tissue parameters and / or sequence parameters) for image generation, not limited to those described above.

[0155] For purposes of example, contrast simulation may involve the injection of simulated lesions. In some other examples, contrast simulation may involve the injection of healthy tissue segments into an image of a body part. In such cases, the computing device 104 may rewrite the values ​​of certain pixels or voxels based on at least one signal equation (e.g., sometimes referred to as a contrast equation). In some cases, the computing device 104 may use simulated contrast for lesion appearance simulation, where, for example, the signal equation may be used to determine the intensity value of the simulated lesion. In this case, the computing device 104 may use these pixels or voxels with or without the intensity values ​​of the pixels or voxels of an image (e.g., a first image or an input image). In certain aspects of using simulated contrast, the computing device 104 may use metadata information (such as T1 relaxation time (T1) and / or T2 relaxation time (T2) values, etc.) of the region / segment of the simulated lesion for lesion appearance simulation. For example, each tissue may include corresponding tissue properties, such as proton density (PD), T1, T2, T2* relaxation, etc. The metadata may include one or more than one of the tissue properties. These constants can, for example, be provided to a signal equation to determine pixel values ​​for various pixels within a tissue region (eg, a simulated lesion) or, in some cases, as part of a healthy tissue region.

[0156] In some configurations, the computing device 104 may perform pediatric simulations to simulate images of a subject's body parts at various stages of development, such as from infant to adult. Fig.18Depicted are examples of pediatric simulations 1800, 1802 according to one or more implementations. For example, to perform a pediatric simulation, the computing device 104 may obtain a dataset of various pairs of adult and neonatal images (e.g., brain MRI scans). These image pairs may be references for transformations of adult scans to neonatal scans. The computing device 104 may apply contrast matching (or history matching) between the neonatal scans and the adult scans, such as matching the neonatal scans to the adult scans, or vice versa, etc. Applying histogram matching may reduce the GM and / or WM contrast in the adult scans (or increase the GM and / or WM contrast in the neonatal scans) to improve the similarity of the contrast shown in the neonatal scans (or adult scans). Subsequently, the computing device 104 may be able to resize a body part (e.g., brain) by compressing (or squeezing) the adult scans to match the size of the neonatal scans. Additionally or alternatively, for example, the computing device 104 may resize the brain by stretching the neonatal scans to match the size of the adult scans. By performing the resizing, the computing device 104 can replicate the size and / or shape of a neonatal brain (or an adult brain, depending on the starting image). In response to performing contrast matching and brain resizing, the computing device 104 can generate examples of transformations from adult scans to neonatal scans, or vice versa, as shown in the example images of the pediatric simulation 1800, 1802.

[0157] In some cases, computing device 104 may simulate healthy tissue using similar operations as pediatric simulation. For example, computing device 104 may perform pediatric simulation by injecting at least one of shape (e.g., of healthy tissue), coverage values, and applying global distortions (such as space occupancy effects, contrast simulation, etc.) to achieve contrast and size matching or consistency.

[0158] As discussed herein, the computing device 104 may perform structural or image simulations for various pathology types (not limited to specific pathology types). Some examples of pathologies that may be simulated using the techniques described herein may include, but are not limited to, stroke (e.g., large vessel occlusion (LVO)), hemorrhage or hematoma (e.g., intracerebral hemorrhage (ICH), intraparenchymal hemorrhage (IPH), subarachnoid hemorrhage (SAH), subdural hematoma (SDH), or epidural hematoma (EDH)), tumors, high signal lesions (e.g., white matter high signal or periventricular high signal), low signal lesions, trauma (e.g., traumatic encephalopathy (TBI)), multiple sclerosis (MS), hydrocephalus, enlargement of any specific brain region, mass effect / edema, or surgical effects (e.g., resection or removal of the skull or any other anatomical structure). In some cases, computing device 104 may utilize the techniques discussed herein to simulate characteristics of Alzheimer's disease (e.g., brain atrophy, brain volume loss, such as in the hippocampus and / or temporal lobes), Parkinson's disease (e.g., atrophy of the substantia nigra and / or other brainstem structures), brain abscesses (e.g., round lesions with hyperintense borders and / or hypointense centers), or meningitis (e.g., hyperintense thin layers along the brain surface), among others. In various cases, certain pathologies (e.g., tumors) may have a relatively complex appearance, such as necrosis, tumor cores, edema, and the like.

[0159] In some configurations, one or more of the techniques described herein (e.g., lesion mask simulation and / or lesion appearance simulation) may be applied multiple times / iterations such that multiple lesions are injected one on top of another. By applying multiple simulated lesions or appearances, the computing device 104 may create an image with simulated lesions that include relatively more complex and diverse appearances. Thus, the computing device 104 may provide simulated structures or images as training data to train (or further train) one or more models for separating healthy structures from abnormal structures, thereby improving the performance of the models to diagnose and treat various health conditions. The computing device 104 may provide the simulated data to other processing devices for improving model performance, or use the simulated data to train at least one local model.

[0160] refer to Fig.19 , depicts a flow chart of an example method 1900 for simulating a structure or image according to one or more implementations. The method 1900 may be performed using any suitable computing system (e.g., Figure 1 The computing device 104, the controller 106, Fig. 20 It will be appreciated that certain steps of method 1900 may be performed in parallel (e.g., concurrently) or sequentially while still achieving useful results. As described herein, method 1900 may be performed to simulate one or more structures or images.

[0161] At operation 1902, method 1900 may include obtaining a first image of a subject, such as by using a computing device (e.g., computing device 104) or a controller (e.g., controller 106). The first image may be a first MR image or other types of images captured by an imaging device, such as a CT image, an ultrasound image, etc.

[0162] At operation 1904, method 1900 may include: determining, using a computing device, a location within the first image for simulating a structure. The structure may include or correspond to healthy tissue or a lesion. For purposes of the examples herein, the structure to be simulated or the structure used to simulate the image may include a lesion. In some other aspects, the structure to be simulated may be healthy tissue. In some implementations, to determine the location, the computing device may identify at least one anatomical region associated with a body part of the subject. For purposes of example, the anatomical region associated with the first image may be at least a portion of the subject's brain. The computing device may use the identified at least one anatomical region or extract (or separate) various regions associated with the first image from the identified at least one anatomical region. For example, such as in combination with Figure 2 As described in at least operation 210 of the present invention, extracting regions may refer to classifying regions from the first image. Based on the extracted regions, the computing device may select at least one first region associated with the first image as a location for a mask (e.g., a lesion mask, a structure mask, an abnormality mask, or a tissue mask). The mask may be a first mask simulated by the computing device. As discussed herein, the computing device 104 may at least partially apply the mask to the first image to generate another image (e.g., a second image), wherein the second image includes at least a portion of the first image with the simulated mask.

[0163] In some implementations, after selecting at least one first region as a location, the computing device may select at least one second region associated with the first image as a second location for another mask. In such a case, for example, the computing device may apply another mask (e.g., a second mask) to the first image to generate a third image. The third image may include at least a portion of the first image with a simulated mask (e.g., a mask in the first region and the second region). In some cases, the second mask may be a portion of the first mask (e.g., an extension of the first mask).

[0164] In some implementations, the computing device may receive an indication of at least one region associated with the first image, such as from a user (eg, user 102). In this case, computing device 104 may use the received at least one region as a location to simulate a lesion.

[0165] At operation 1906, method 1900 may include, using a computing device, simulating a shape for the structure based on the location. For example, the computing device may simulate the shape of the structure based on or based on a location (e.g., at least one region) associated with the first image. The shape and / or size of the structure may be based on an associated region or location of the body part.

[0166] In some implementations, to simulate the shape, the computing device may generate an elliptical shape based on one or more parameters. For example, the one or more parameters may include a major axis or a minor axis of the first image that defines the size of the structure. In some cases, the one or more parameters may be based on the location of the shape to be simulated. The computing device may apply an elastic distortion to the elliptical shape to simulate the shape of the structure. The shape simulation may be combined with, but not limited to, Figure 6 to describe.

[0167] At operation 1908, method 1900 may include: using a computing device, generating a mask (eg, a first mask) based on the location and shape of the structure. In some implementations, computing device 104 may generate a mask having a combination of, for example, but not limited to, Figure 6 The mask may be a 2D mask or a 3D mask including a plurality of 2D masks.

[0168] In some implementations, to generate a mask, such as described in conjunction with at least a region growing technique / approach, the computing device may place or set a seed to the first image at the location. The seed may represent a starting point for growth of the mask. The computing device may use at least one region growing algorithm to grow at least one region around the seed. The region growing algorithm may include, but is not limited to, at least one of the following: region growing, region merging, splitting and merging, watershed transform, connected component labeling, and graph cut segmentation. The computing device may iteratively grow the region around the seed (e.g., to neighboring pixels or voxels) until at least one condition (such as a maximum size of the mask, a number of iterations, etc.) is met. Based on the grown seed (or the region around the seed), the computing device may generate a mask for lesion simulation.

[0169] In some implementations, a seed may be set at at least one voxel having a first intensity (e.g., voxel intensity). The computing device may grow a region around the seed to at least one neighboring voxel having a second intensity, where the second intensity is substantially similar to the first intensity. The computing device may, for example, prevent region growth to other regions having intensities substantially different from the first intensity to maintain the lesion in the same anatomical region. In some cases, multiple seeds may be set to multiple voxels as starting points for growth of at least one mask.

[0170] In some implementations, to generate a mask, such as described in conjunction with at least a data-driven pathology mask generation process, a computing device may obtain a historical mask associated with historical images from one or more subjects. The historical mask may refer to a real lesion captured by at least one imaging device. In this case, the computing device may train a model using at least one ML technique based on the historical mask. The model may learn patterns, features, characteristics, or other details associated with the historical mask. Using the model, the computing device may generate a mask for a location based on the historical mask. The computing device may refine, modify, or perform any other updates to the mask for simulating a structure based on a comparison between the generated mask and the historical mask associated with the location. For example, based on similarities or differences in features between the mask and (one or more than one) historical masks, the computing device may provide one or more corrections to the generated mask to increase the fidelity of the simulation.

[0171] In some implementations, to generate the mask, the computing device may determine the appearance of the mask (e.g., of a simulated lesion). The appearance of the mask may include, but is not limited to, the intensity of one or more voxels of the first image. The intensity may refer to pixel or voxel intensity. The appearance may include other features discussed herein, such as contrast, noise, pattern, etc.

[0172] In some implementations, to determine the appearance, the computing device may select an aggregate (e.g., average, mean, median, etc.) pixel intensity of the mask. The computing device may apply at least one pattern to the mask. The at least one pattern may include, but is not limited to, at least one of the following: an edema pattern, a hemorrhagic pattern, a necrosis pattern, a cystic pattern, an inflammatory pattern, a tumor pattern, and / or an ischemic pattern. In addition to or in lieu of applying at least one pattern, the computing device may simulate at least one noise for the mask. The at least one noise may include, but is not limited to, acquisition noise, Gaussian noise, etc. The aggregated pixel intensity, at least one pattern, and / or at least one noise may be part of the appearance of the mask. Such as in combination with at least one Figure 2 After simulating the lesion appearance of operation 224, the computing device may perform other operations.

[0173] In some implementations, to generate a mask, such as by combining but not limited to at least Fig.13 In the hydrocephalus simulation described above, the computing device may simulate a shape deformation associated with the shape. In this case, the simulated shape deformation may correspond to simulating hydrocephalus for a body part of the subject. The computing device may generate a mask based on the position and shape deformation of the structure.

[0174] At operation 1910, method 1900 may include, using a computing device, applying a mask to a first image (e.g., an input image) to generate or simulate a second image (e.g., an output image) that simulates the structure. The second image may be a second MR image or other type of image similar to the first image.

[0175] In some implementations, to determine the appearance and apply the mask, the computing device may perform, for example, the features or operations described in conjunction with at least a portion of the lesion appearance simulation. For example, the computing device may provide at least a portion of the first image for applying the mask and the generated mask as input to a model trained using at least one machine learning technique. Using the model, the computing device may generate a third image, the third image including at least a portion of the first image and the generated mask (e.g., injected into the first image). The computing device may use the model to compare the third image with at least one historical image. At least one historical image may have a second mask (e.g., corresponding to or including a historical structure or a real lesion) at a location, the second mask having a second shape similar to the shape of the mask. In some cases, the second mask may have a different shape than the mask. Based on the comparison, the computing device may identify features or details of structures in the third image that are comparable to historical structures (e.g., real lesions) of the historical image and / or features that do not represent historical structures. Based on the comparison, the computing device may update the third image to enhance the fidelity of the simulated structure and / or image (e.g., in this case, the third image).

[0176] In some implementations, to apply a mask, such as in conjunction with but not limited to Figure 12 to Figure 13 In at least one of the above, the computing device can simulate a spacer effect around the mask. In this case, the computing device can apply the mask to the first image with the simulated spacer effect to generate a second image with the simulated structure. In some cases, the spacer effect can be part of the appearance of the mask.

[0177] In some implementations, to apply the mask, the computing device may identify one or more anatomical regions associated with a body part of the subject. The individual anatomical regions may be represented as a 2D image or a 3D image. The computing device may simulate contrast for the first image based on at least one of the anatomical regions and one or more sequence parameters. The computing device may simulate contrast using at least one signal equation (e.g., sometimes referred to as a contrast equation), the at least one signal equation including but not limited to at least one of the following: spin echo, spoiled gradient echo, rewind gradient, time reversal gradient echo, and / or a random equation. One or more sequence parameters may include at least one of a relaxation time, an echo time, an inversion time, and / or a flip angle, etc. Different sequence parameters may be used to generate different contrast details for the image. The computing device may apply the mask and the simulated contrast to the first image to generate a second image of the simulated structure. In some cases, the contrast may indicate at least one of the GM, WM, and / or CSF, and others, associated with the body part of the subject. In various cases, the contrast may be part of the appearance of the mask. In this case, generating the mask may include the computing device determining the appearance of the mask based on the simulated contrast. It may be combined with, but not limited to Figures 14 to 17 At least one of the examples for simulating contrast is used to illustrate.

[0178] In some implementations, the computing device can perform pediatric simulations, such as simulating pediatric images or switching between adult scans and neonatal scans. Fig.18 To illustrate the operations used in the pediatric simulation. For example, a computing device may obtain multiple images of a subject, the images including at least a fourth image and a fifth image associated with a body part of at least one subject (e.g., an adult and / or a neonatal subject). The fourth image may be a neonatal image (e.g., a neonatal scan) of the body part of the at least one subject, such as from a neonatal subject. The fifth image may be a developmental image (e.g., an adult scan) of the body part of the at least one subject.

[0179] For purposes of example, the computing device may, for example, be configured to transform an adult scan of a fifth image into a simulated neonatal scan to simulate a sixth image. In this case, the sixth image may be another neonatal image simulated based on a developmental image of a body part of at least one subject. For example, the computing device may conform the fifth image to the fourth image. Conforming the fifth image to the fourth image may include at least one of conforming a first contrast of the fifth image to a second contrast of the fourth image (e.g., contrast matching) and / or conforming a first size of the fifth image to a second size of the fourth image (e.g., size matching). Conforming the first contrast to the second contrast and conforming the first size to the second size may include at least one of: injecting one or more shapes into the fifth image based on the fourth image; overwriting one or more values ​​associated with pixels or voxels of the fifth image based on the fourth image; and applying one or more distortions to the fifth image based on the fourth image. For example, by conforming one or more features of the fifth image to the fourth image, the computing device may transform the fifth image into a sixth image representing a neonatal scan simulated based on an adult scan (or a simulated sixth image).

[0180] Fig. 20 Component diagram illustrating an example computing system suitable for use in various implementations described herein according to an example implementation. In a non-limiting example, computing system 2000 may implement Figure 1 computing device 104 or controller 106, or various other example systems and devices described in this disclosure.

[0181] The computing system 2000 includes a bus 2002 or other communication component for communicating information, and a processor 2004 coupled to the bus 2002 for processing information. The computing system 2000 also includes a main memory 2006 (such as a RAM or other dynamic storage device) coupled to the bus 2002 for storing information and instructions to be executed by the processor 2004. The main memory 2006 can also be used to store location information, temporary variables, or other intermediate information during the execution of instructions by the processor 2004. The computing system 2000 can also include a ROM 2008 or other static storage device coupled to the bus 2002, which is used to store static information and instructions used by the processor 2004. A storage device 2010, such as a solid-state device, magnetic disk, or optical disk, is coupled to the bus 2002 for persistent storage of information and instructions.

[0182] The computing system 2000 may be coupled via the bus 2002 to a display 2014, such as a liquid crystal display or an active matrix display, for displaying information to a user. An input device 2012, such as a keyboard including alphanumeric and other keys, may be coupled to the bus 2002 for communicating information and command selections to the processor 2004. In another implementation, the input device 2012 has a touch screen display. The input device 2012 may include any type of biometric sensor or cursor control, such as a mouse, trackball, or cursor direction keys, for communicating direction information and command selections to the processor 2004 and for controlling cursor movement on the display 2014.

[0183] In some implementations, the computing system 2000 may include a communication adapter 2016, such as a networking adapter. The communication adapter 2016 may be coupled to the bus 2002 and may be configured to enable communication with a computing or communication network or other computing systems. In various exemplary implementations, the communication adapter 2016 may be used to implement any type of networking configuration, such as wired (e.g., via Ethernet), wireless (e.g., via Wi-Fi, Bluetooth), satellite (e.g., via GPS), pre-configured, ad-hoc, LAN, WAN, etc.

[0184] According to various implementations, the processing of the illustrative implementations described herein may be implemented by the computing system 2000 in response to the processor 2004 executing the implementation of the instructions contained in the main memory 2006. Such instructions may be read into the main memory 2006 from another computer-readable medium, such as the storage device 2010, etc. The execution of the implementation of the instructions contained in the main memory 2006 causes the computing system 2000 to perform the illustrative processing described herein. One or more processors in a multi-processing implementation may also be employed to execute the instructions contained in the main memory 2006. In alternative implementations, hard-wired circuitry may be used in place of or in combination with software instructions to implement the illustrative implementations. Thus, the implementation is not limited to any specific combination of hardware circuitry and software.

[0185] Potential examples include, but are not limited to:

[0186] Embodiment AA: A method for simulating a structure in an image, comprising: obtaining a first image of a subject; determining a position within the first image for simulating a structure; simulating a shape for the structure based on the position; generating a mask based on the position and the shape for the structure; and applying the mask to the first image to generate a second image simulating the structure.

[0187] Embodiment AB: A method according to embodiment AA, wherein determining the position within the image includes: identifying at least one anatomical region associated with a body part of the subject; using the identified at least one anatomical region, extracting multiple regions associated with the first image; and selecting at least one first region associated with the first image as the position for the mask.

[0188] Embodiment AC: The method of Embodiment AB further comprising: after selecting the at least one first region, selecting at least one second region associated with the first image as a second position for another mask for generating a third image.

[0189] Embodiment AD: A method according to any one of embodiments AA to AC, wherein determining the position and simulating the shape includes: receiving an indication of at least one area associated with the first image as the position; and simulating the shape of the structure based on the at least one area associated with the first image.

[0190] Embodiment AE: A method according to any one of embodiments AA to AD, wherein generating the mask comprises: setting a seed to the first image at the position; using at least one region growing algorithm to grow at least one region around the seed; and generating the mask based on the grown seed.

[0191] Embodiment AF: A method according to embodiment AE, wherein the at least one region growing algorithm includes at least one of the following items: region growing, region merging, split and merge, watershed transform, connected component labeling, and graph cut segmentation.

[0192] Embodiments AG: A method according to embodiments AE, wherein the seed is set at at least one voxel having a first intensity, and wherein a region around the seed is grown to at least one neighboring voxel having a second intensity, wherein the second intensity is substantially similar to the first intensity.

[0193] Embodiment AH: A method according to any one of embodiments AA to AG, wherein simulating the shape includes: generating an elliptical shape based on one or more parameters, wherein the one or more parameters include the major axis or minor axis of the first image defining the size of the structure; and applying elastic distortion to the elliptical shape to simulate the shape of the structure.

[0194] Embodiment AI: A method according to any one of embodiments AA to AH, wherein generating the mask comprises: obtaining multiple historical masks associated with multiple images from one or more subjects; training a model based on the multiple historical masks using at least one machine learning technique; using the trained model, generating a mask for the location based on the multiple historical masks; and refining the mask for simulating the structure based on a comparison between the generated mask and the multiple historical masks associated with the location.

[0195] Embodiment AJ: The method of any one of Embodiments AA to AI, wherein generating the mask comprises determining an appearance of the mask associated at least with an intensity of one or more voxels of the first image.

[0196] Embodiment AK: The method of embodiment AJ, wherein determining the appearance comprises: selecting aggregate pixel intensities of the mask; applying at least one pattern to the mask; and simulating at least one noise to the mask.

[0197] Embodiment AL: The method of embodiment AK, wherein the at least one pattern comprises at least one of: an edema pattern, a hemorrhagic pattern, a necrotic pattern, a cystic pattern, an inflammatory pattern, a tumor pattern, and an ischemic pattern.

[0198] Embodiment AM: A method according to embodiment AJ, wherein determining the appearance and applying the mask includes: providing at least a portion of the first image for applying the mask and the generated mask as input to a model trained using machine learning techniques; using the model to generate a third image, the third image including the at least a portion of the first image and the generated mask; and updating the third image based on a comparison of the third image with at least one historical image, the at least one historical image having a second mask at the location, the second mask having a second shape similar to the shape of the mask.

[0199] Embodiment AN: A method according to any one of embodiments AA to AM, wherein applying the mask comprises: simulating a space-occupying effect around the mask; and applying the mask to a first image having the simulated space-occupying effect to generate the second image simulating the structure.

[0200] Embodiment AO: The method of any one of Embodiments AA to AN, wherein the first image is a first magnetic resonance (MR) image and the second image is a second MR image.

[0201] Embodiment AP: The method of any one of embodiments AA to AO, wherein generating the mask comprises: simulating a shape deformation associated with the shape; and generating the mask based on the position and the shape deformation for the structure.

[0202] Embodiment AQ: The method of embodiment AP, wherein simulating the shape deformation corresponds to simulating hydrocephalus of a body part of the subject.

[0203] Embodiment AR: A method according to any one of embodiments AA to AQ, wherein applying the mask further comprises: identifying multiple anatomical regions associated with a body part of the subject; simulating contrast for the first image based on at least one of the multiple anatomical regions and one or more sequence parameters; and applying the mask and the simulated contrast to the first image to generate the second image of the structure.

[0204] Embodiment AS: The method of embodiment AR, wherein the one or more sequence parameters include at least one of relaxation time, echo time, inversion time, and flip angle.

[0205] Embodiment AT: The method of embodiment AR, wherein each anatomical region of the plurality of anatomical regions is represented as a two-dimensional (2D) image or a three-dimensional (3D) image.

[0206] Embodiment AU: The method of embodiment AR, wherein the contrast indicates at least one of gray matter (GM), white matter (WM), and cerebrospinal fluid (CSF) associated with the body part of the subject.

[0207] Embodiment AV: A method according to embodiment AR, wherein simulating the contrast includes: using at least one signal equation, the at least one signal equation including at least one of the following terms: spin echo, spoiled gradient echo, rewind gradient, time reversal gradient echo and random.

[0208] Embodiment AW: The method of embodiment AR, wherein generating the mask includes determining an appearance of the mask based on a simulated contrast.

[0209] Embodiment AX: According to the method described in any one of embodiments AA to AW, it includes: obtaining multiple images of multiple subjects, the multiple images including at least a fourth image and a fifth image associated with a body part of at least one subject; making the fifth image conform to the fourth image; and simulating a sixth image based on the conforming fifth image.

[0210] Embodiment AY: A method according to embodiment AX, wherein making the fifth image conform to the fourth image includes at least one of the following items: making the first contrast of the fifth image conform to the second contrast of the fourth image; and making the first size of the fifth image conform to the second size of the fourth image.

[0211] Embodiment AZ: A method according to embodiment AY, wherein the fourth image is a neonatal image of a body part of the at least one subject, the fifth image is a developmental image of a body part of the at least one subject, and the sixth image is another neonatal image simulated based on the developmental image of the body part of the at least one subject.

[0212] Embodiment AAa: A method according to embodiment AY, wherein making the first contrast conform to the second contrast and making the first size conform to the second size includes at least one of the following items: injecting one or more shapes into the fifth image based on the fourth image; overwriting one or more values ​​associated with pixels or voxels of the fifth image based on the fourth image; and applying one or more distortions to the fifth image based on the fourth image.

[0213] Embodiment BA: A system for simulating a structure in an image, comprising one or more processors, wherein the one or more processors are configured to: obtain a first image of a subject; determine a position within the first image for simulating a structure; simulate a shape for the structure based on the position; generate a mask based on the position and the shape for the structure; and apply the mask to the first image to generate a second image simulating the structure.

[0214] Embodiment CA: A non-transitory computer-readable storage medium storing processor-executable instructions which, when executed by at least one processor, cause the at least one processor to: obtain a first image of a subject; determine a position within the first image for simulating a structure; simulate a shape for the structure based on the position; generate a mask based on the position and the shape for the structure; and apply the mask to the first image to generate a second image simulating the structure.

[0215] The implementations described herein have been described with reference to the accompanying drawings. The drawings illustrate certain details of specific implementations that implement the systems, methods, and programs described herein. Describing the implementations using the drawings should not be construed as imposing any limitations on the disclosure that may be present in the drawings.

[0216] It should be understood that claim elements herein are not to be interpreted under the provisions of 35 USC §112(f) unless the element is expressly recited using the phrase "means for."

[0217] As used herein, the term "circuit" may include hardware configured to perform the functions described herein. In some implementations, each corresponding "circuit" may include a machine-readable medium for configuring the hardware to perform the functions described herein. A circuit may be embodied as one or more circuit system components, including but not limited to processing circuit systems, network interfaces, peripheral devices, input devices, output devices, sensors, etc. In some implementations, a circuit may take the form of one or more analog circuits, electronic circuits (e.g., integrated circuits (ICs), discrete circuits, system-on-chip (SOC) circuits), telecommunications circuits, hybrid circuits, and any other type of "circuit". In this regard, a "circuit" may include any type of component for implementing or facilitating the implementation of the operations described herein. In a non-limiting example, a circuit as described herein may include one or more transistors, logic gates (e.g., NAND, AND, NOR, OR, XOR, NOT, XNOR), resistors, multiplexers, registers, capacitors, inductors, diodes, wiring, etc.

[0218] A "circuit" may also include one or more processors communicatively coupled to one or more memories or memory devices. In this regard, one or more processors may execute instructions stored in the memory, or may execute instructions that are otherwise accessible to one or more processors. In some implementations, one or more processors may be embodied in various ways. One or more processors may be constructed in a manner sufficient to perform at least the operations described herein. In some implementations, one or more processors may be shared by multiple circuits (e.g., circuit A and circuit B may include or otherwise share the same processor, which, in some example implementations, may execute instructions stored or otherwise accessed via different areas of the memory). Alternatively or additionally, one or more processors may be constructed to perform or otherwise perform certain operations independently of one or more coprocessors.

[0219] In other example implementations, two or more processors may be coupled via a bus to enable independent, parallel, pipeline or multi-threaded instruction execution. Each processor may be implemented as one or more general-purpose processors, ASICs, FPGAs, GPUs, TPUs, digital signal processors (DSPs) or other suitable electronic data processing components configured to execute instructions provided by a memory. One or more processors may take the form of a single-core processor, a multi-core processor (e.g., a dual-core processor, a triple-core processor or a quad-core processor), a microprocessor, etc. In some implementations, one or more processors may be external to the device, and in a non-limiting example, one or more processors may be remote processors (e.g., cloud-based processors). Alternatively or additionally, one or more processors may be internal or local to the device. In this regard, a given circuit or its components may be arranged locally (e.g., as part of a local server, a local computing system) or remotely (e.g., as part of a remote server such as a cloud-based server). To this end, a "circuit" as described herein may include components distributed across one or more locations.

[0220] An example system for implementing the entire system or part of the implementation may include a general computing device in the form of a computer, including a processing unit, a system memory, and a system bus that couples various system components including the system memory to the processing unit. Each memory device may include a non-transitory volatile storage medium, a non-volatile storage medium, a non-transitory storage medium (e.g., one or more volatile or non-volatile memories), etc. In some implementations, the non-volatile medium may take the form of a ROM, a flash memory (e.g., a flash memory such as NAND, 3D NAND, NOR, 3D NOR, etc.), an EEPROM, an MRAM, a magnetic storage unit, a hard disk, an optical disk, etc. In other implementations, the volatile storage medium may take the form of a RAM, a TRAM, a ZRAM, etc. The above combination is also included in the scope of the machine-readable medium. In this regard, in a non-limiting example, the machine executable instructions include instructions and data that cause a general-purpose computer, a special-purpose computer, or a special-purpose processing machine to perform a certain function or group of functions. According to example implementations described herein, each respective memory device may be operable to maintain or otherwise store information related to operations performed by one or more associated circuits, including processor instructions and related data (e.g., database components, object code components, script components).

[0221] It should also be noted that, as used herein, the term "input device" may include any type of input device, including but not limited to a keyboard, keypad, mouse, joystick, or other input device that performs similar functions. In contrast, as used herein, the term "output device" may include any type of output device, including but not limited to a computer monitor, printer, fax machine, or other output device that performs similar functions.

[0222] It should be noted that although the figures herein can show the specific order and composition of method steps, it should be understood that the order of these steps may be different from the order depicted. In a non-limiting example, two or more than two steps may be performed concurrently or partially concurrently. In addition, some method steps performed as discrete steps may be combined, steps performed as combined steps may be separated into discrete steps, the order of certain processing may be reversed or otherwise changed, and the nature or quantity of discrete processing may be changed or changed. The order or sequence of any element or device may be changed or replaced according to alternative implementations. Therefore, all such modifications are intended to be included in the scope of the present disclosure as defined in the appended claims. Such changes will depend on the selected machine-readable medium and hardware system and designer selection. It should be understood that all such changes are within the scope of the present disclosure. Similarly, the software and web implementations of the present disclosure may be implemented using standard programming techniques with rule-based logic and other logic to implement various database search steps, correlation steps, comparison steps, and decision steps.

[0223] Although this specification includes many specific implementation details, these should not be interpreted as limitations on the scope of any invention or the scope of the content that may be claimed, but should be interpreted as a description of the specific features of the specific implementation of the system and method described herein. Certain features described in the context of separate implementations in this specification may also be implemented in a single implementation in combination. On the contrary, the various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations. In addition, although the features may be described as working in certain combinations and even initially claimed as such, one or more than one feature from the claimed combination may be removed from the combination in some cases, and the claimed combination may involve a deformation of a sub-combination or a sub-combination.

[0224] In some cases, multitasking and parallel processing may be advantageous. In addition, the separation of various system components in the implementation described above should not be understood as requiring such separation in all implementations, and it should be understood that the program components and systems described generally can be integrated together in a single software product or packaged in multiple software products.

[0225] Now that some illustrative implementations and implementations have been described, it is apparent that the foregoing is illustrative rather than restrictive and has been presented by way of example. In particular, although many of the examples presented herein relate to specific combinations of method actions or system elements, those actions and those elements may be combined in other ways to achieve the same purpose. Actions, elements, and features discussed only in conjunction with one implementation are not intended to be excluded from similar roles in other implementations.

[0226] The words and terms used herein are for illustrative purposes and should not be considered limiting. The use of "including," "having," "having," "comprising," "involving," "characterized by," "characterized by," and variations thereof herein is intended to encompass the items listed thereafter and their equivalents and additional items, as well as alternative implementations consisting solely of the items listed thereafter. In one implementation, the systems and methods described herein consist of one, various combinations of more than one, or all of the described elements, acts, or components.

[0227] Any reference to an implementation or element or action of a system and method mentioned herein in the singular may also include an implementation comprising a plurality of these elements, and any plural reference to any implementation or element or action herein may also include an implementation comprising only a single element. References in singular or plural form are not intended to limit the currently disclosed system or method, its components, actions or elements to a single or multiple configurations. References to any action or element based on any information, action or element may include an implementation where the action or element is based at least in part on any information, action or element.

[0228] Any implementation disclosed herein may be combined with any other implementation, and references to "implementations," "some implementations," "alternative implementations," "various implementations," or "an implementation," etc., are not necessarily mutually exclusive and are intended to indicate that a particular feature, structure, or characteristic described in connection with an implementation may be included in at least one implementation. These terms as used herein do not necessarily all refer to the same implementation. Any implementation may be combined, inclusively or exclusively, with any other implementation in any manner consistent with the aspects and implementations disclosed herein.

[0229] References to "or" may be construed as inclusive such that any term described using "or" may indicate any of a single, more than one, and all of the described terms.

[0230] Where reference numerals are present after technical features in the drawings, detailed description or any claims, the sole purpose of including such reference numerals is to increase the intelligibility of the drawings, detailed description and claims. Therefore, reference numerals and their absence shall not have any limiting effect on the scope of any claim element.

[0231] The foregoing description of implementations is presented for purposes of illustration and description. The foregoing description of implementations is not intended to be exhaustive or to limit the disclosure to the precise form disclosed, and modifications and variations may be made in accordance with the above teachings, or modifications and variations may be obtained from the disclosure. These implementations are selected and described to explain the principles of the disclosure and its practical application, so that those skilled in the art can utilize various implementations and utilize various modifications suitable for the intended specific use. Other substitutions, modifications, changes and omissions may be made in the design, operating conditions and implementation of the implementations without departing from the scope of the disclosure as expressed in the appended claims.

Claims

1. A method for modeling a structure in an image, comprising: obtaining a first image of a subject; determining a location within the first image for simulating a structure; modeling a shape for the structure according to the position; generating a mask based on the location and a shape for the structure; as well as The mask is applied to the first image to generate a second image that models the structure.

2. The method according to claim 1, wherein: Determining the location within an image involves: identifying at least one anatomical region associated with a body part of the subject; extracting a plurality of regions associated with the first image using the identified at least one anatomical region; and At least one first region associated with the first image is selected as a location for the mask.

3. The method according to claim 2, further comprising: After selecting the at least one first region, At least one second region associated with the first image is selected as a second location for another mask used to generate a third image.

4. The method according to claim 1, wherein: Determining the position and simulating the shape includes: receiving an indication of at least one area associated with the first image as the location; and A shape of the structure is modeled based on the at least one region associated with the first image.

5. The method according to claim 1, wherein: Generating the mask comprises: seeding the first image at the location; growing at least one region around the seed using at least one region growing algorithm; and The mask is generated based on the grown seeds.

6. The method according to claim 5, wherein: The at least one region growing algorithm comprises at least one of: region growing, region merging, split and merge, watershed transform, connected component labeling, and graph cut segmentation.

7. The method according to claim 5, wherein: The seed is disposed at at least one voxel having a first intensity, and wherein a region around the seed is grown to at least one neighboring voxel having a second intensity, wherein the second intensity is substantially similar to the first intensity.

8. The method according to claim 1, wherein: Simulating the shape includes: generating an elliptical shape according to one or more parameters including a major axis or a minor axis of the first image defining a size of the structure; and An elastic warp is applied to the elliptical shape to simulate the shape of the structure.

9. The method according to claim 1, wherein: Generating the mask comprises: obtaining a plurality of history masks associated with a plurality of images from one or more subjects; Based on the plurality of historical masks, training a model using at least one machine learning technique; generating a mask for the location based on the plurality of historical masks using the trained model; and The mask is refined for simulating the structure based on a comparison between the generated mask and a plurality of historical masks associated with the location.

10. The method according to claim 1, wherein: Generating the mask includes determining an appearance of the mask associated with at least an intensity of one or more voxels of the first image.

11. The method according to claim 10, wherein: Determining the appearance includes: selecting an aggregate pixel intensity of the mask; applying at least one pattern to the mask; and At least one noise is simulated for the mask.

12. The method according to claim 11, wherein: The at least one pattern includes at least one of: an edema pattern, a hemorrhagic pattern, a necrotic pattern, a cystic pattern, an inflammatory pattern, a tumor pattern, and an ischemic pattern.

13. The method according to claim 10, wherein: Determining the appearance and applying the mask comprises: providing at least a portion of the first image for applying the mask and the generated mask as input to a model trained using machine learning techniques; generating a third image using the model, the third image comprising the at least a portion of the first image and the generated mask; and The third image is updated based on a comparison of the third image with at least one historical image having a second mask at the location, the second mask having a second shape similar to a shape of the mask.

14. A system for modeling a structure in an image, comprising one or more processors, the one or more processors being configured to: obtaining a first image of a subject; determining a location within the first image for simulating a structure; modeling a shape for the structure according to the position; generating a mask based on the location and a shape for the structure; as well as The mask is applied to the first image to generate a second image that models the structure.

15. A non-transitory computer-readable storage medium storing processor-executable instructions that, when executed by at least one processor, cause the at least one processor to: obtaining a first image of a subject; determining a location within the first image for simulating a structure; modeling a shape for the structure according to the position; generating a mask based on the location and a shape for the structure; as well as The mask is applied to the first image to generate a second image that models the structure.

Images