Anatomical segmentation of cross-sectional images containing anatomical abnormalities

Abstract

Disclosed herein is a medical system (100, 300, 400, 500) where execution of the machine-executable instructions causes a computing system to receive an input tomographic medical image showing an anatomical abnormality, receive a modified tomographic medical image without the anatomical abnormality in response to inputting the input tomographic medical image into an image processing neural network, receive an anatomical segmentation of the modified tomographic medical image in response to inputting the modified tomographic medical image into an image segmentation module, the anatomical segmentation dividing the modified tomographic medical image into a plurality of anatomical regions, and provide the anatomical segmentation as a segmentation of the input tomographic medical image.

Description

[Technical Field]

[0001] The present invention relates to medical tomographic imaging techniques, and more particularly to segmentation of tomographic medical images. [Background technology]

[0002] Various medical tomographic imaging techniques, such as magnetic resonance imaging (MRI), computed tomography, positron emission tomography, and single-photon emission tomography, allow for detailed visualization of a subject's anatomical structures. After a tomographic medical image is acquired, it is typically segmented or divided into multiple anatomical regions.

[0003] International Patent Application Publication WO2022057312A1 discloses a medical image processing method, apparatus, device, and storage medium. The method includes: extracting tissue abnormality masks from multiple tissue abnormality images using labeling information; randomly selecting one or more masks; generating a new tissue abnormality image on the selected one or more masks by using a Gaussian function; obtaining coordinate values ​​of a predicted abnormal tissue region in a normal tissue sample image; and overlaying the new tissue abnormality image on the image corresponding to the predicted abnormal tissue region according to the coordinate values ​​to obtain a simulated tissue abnormality image. The simulation of the tissue abnormality image is disclosed, and the morphology, size, grayscale distribution, and location of the lesion can be controlled.

[0004] US Patent Application US2020 / 364864 discloses a GAN-based method for mapping abnormal medical images to normative medical images, where regions of abnormality in the input image contain synthesized anatomical data. Summary of the Invention [Problem to be solved by the invention]

[0005] The present invention provides a medical system, a computer program and a method in the independent claims. Embodiments are given in the dependent claims.

[0006] A difficulty in segmenting tomographic medical images is that the presence of some anatomical abnormalities, such as tumors, can cause the image segmentation module to malfunction. Image segmentation modules can be implemented using different techniques and technologies, such as neural networks, shape deformable models, or anatomical atlas registration. In the presence of anatomical abnormalities, these image segmentation modules can output inaccurate anatomical segmentations. Embodiments can provide an improved means of segmenting tomographic medical images with anatomical abnormalities. First, an input tomographic medical image in the presence of an anatomical abnormality is input to an image processing neural network trained to output a modified tomographic medical image that is similar to the input tomographic medical image but with the anatomical abnormality removed. The modified tomographic medical image is essentially a simulated tomographic medical image. The modified tomographic medical image is then segmented using the image segmentation module to provide an anatomical segmentation. The modified anatomical segmentation of the tomographic medical image is provided as a segmentation of the input tomographic medical image. [Means for solving the problem]

[0007] In one aspect, the present invention provides a medical system comprising a memory storing machine-executable instructions and also storing an image processing neural network. The image processing neural network is a neural network that receives an image and outputs an image. The image processing neural network is configured to output a modified tomographic medical image describing the selected anatomical region of the subject in response to receiving an input tomographic medical image describing the selected anatomical region. The image processing neural network is further configured to output the modified tomographic medical image such that an anatomical abnormality depicted in the input tomographic medical image is removed. For example, a tumor or other anatomical abnormality may be depicted in the input tomographic medical image. This may then be removed by the image processing neural network, and a new image constructed in which the anatomical abnormality is not depicted.

[0008] The present invention relates to an image processing neural network that returns a corrected tomographic medical image from an input tomographic medical image depicting an anatomical abnormality. The anatomical abnormality is removed from the corrected tomographic image by the image processing neural network. An anatomical segmentation of the corrected tomographic image (i.e., without the anatomical abnormality) into multiple anatomical regions is received from a segmentation module. The segmentation is then applied to the input tomographic image (with the anatomical abnormality).

[0009] The medical system further comprises a computing system, wherein execution of the machine-executable instructions causes the computing system to receive an input tomographic medical image, wherein execution of the machine-executable instructions causes the computing system to receive a modified tomographic medical image in response to inputting the input tomographic medical image to an image processing neural network, and wherein execution of the machine-executable instructions further causes the computing system to receive an anatomical segmentation of the modified tomographic medical image in response to inputting the modified tomographic medical image to an image segmentation module.

[0010] The anatomical segmentation segments the corrected tomographic medical image into a plurality of anatomical regions. That is, the image segmentation module is configured to segment the corrected tomographic medical image into a plurality of anatomical regions. Execution of the machine-executable instructions further causes the computing system to provide the anatomical segmentation as a segmentation of the input tomographic medical image. This can be beneficial because a conventional image segmentation module can fail if an anatomical abnormality is present. For example, if it is a formal image segmentation module, the presence of an anatomical abnormality such as a tumor, growth, or abnormal anatomical structure can cause the segmentation module to fail and output an inaccurate anatomical segmentation. Using the anatomical segmentation of the corrected tomographic medical image can provide a better segmentation of the input tomographic medical image.

[0011] The image segmentation module can be implemented in various ways. For example, it can be done using a neural network such as a U-net or similar encoder-decoder network such as a V-net neural network. The image segmentation module can be a deformable shape model. In another example, image segmentation can be registration to an anatomical atlas.

[0012] In different examples, the computed tomography medical image may be, for example, a positron emission computed tomography (PET) image, a single photon emission computed tomography (SPECT), a computed tomography (CT) image, a magnetic resonance (MR) image, a combination of positron emission computed tomography and magnetic resonance image, a combination of computed tomography and magnetic resonance image, or other variations.

[0013] Image processing neural networks can be constructed in different ways, for example, they can be cycleGAN neural networks or variational autoencoders.

[0014] If the image processing neural network is a cycleGAN, it can be trained using the standard cycleGAN method based on unpaired training images with or without abnormalities / pathologies. The best approach is described in arXive paper 1810.10850v2, "An Adversarial Learning Approach to Medical Image Synthesis for Lesion Detection," by Sun et al., April 8, 2019.

[0015] When an image processing neural network is a variational autoencoder, the training data contains both anatomically normal and pathological images, both of which are mapped to latent space representations. In the autoencoder's latent space, the average locations of anatomically normal and pathological images can be determined and a difference vector can be calculated. When a tomographic medical image contains an anatomical abnormality, its representation in the latent space can be shifted by the difference vector to remove the anatomical abnormality.

[0016] In another embodiment, execution of the machine-executable instructions further causes the computing system to provide an image mask that identifies the location of anatomical abnormalities in the input magnetic resonance image by comparing the input tomographic medical image with the modified tomographic medical image. For example, the images can be subtracted from each other. The difference between the input tomographic medical image and the modified tomographic medical image can then be thresholded, which can provide an appropriate image mask.

[0017] In another embodiment, execution of the machine-executable instructions further causes the computing system to provide an image mask that identifies locations of anatomical abnormalities in the input tomographic medical image by receiving the image mask from an image processing neural network. The image processing neural network is further configured to output the image mask. During training, the locations of the anatomical abnormalities may be labeled in the ground truth image. The labels of the locations of the anatomical abnormalities in the ground truth image can provide a basis for training the image processing neural network to also output the image mask.

[0018] In another embodiment, execution of the machine-executable instructions further causes the computing system to determine, using the image mask and the anatomical segmentation, a geometric measure describing the anatomical abnormality in each of the plurality of anatomical regions. The geometric measure may be, for example, a quantitative penetration measure describing the volume fraction of the anatomical abnormality in each of the plurality of anatomical regions. In another example, the geometric measure may describe the distance of penetration of the anatomical abnormality into each of the plurality of anatomical regions. In yet another example, the geometric measure may describe the surface area of ​​the anatomical abnormality in each of the plurality of anatomical regions.

[0019] This embodiment can be beneficial because the image mask provides information about the size and location of the anatomical anomaly, and the anatomical segmentation provides information about what the anatomical region would be like without the anatomical anomaly. This provides a means to quantitatively measure the effect of the anatomical anomaly on various anatomical regions of a subject. Such measurements would likely not be possible using conventional segmentation techniques.

[0020] In another embodiment, execution of the machine-executable instructions further causes the computing system to render a composite image that overlays the anatomical segmentation on the input tomographic medical image. This may be beneficial because it may provide a means of visualizing anatomical regions affected by anatomical abnormalities. It may also provide a means of verifying or determining the validity of the anatomical segmentation.

[0021] In another embodiment, execution of the machine-executable instructions further causes the computing system to highlight regions of the anatomical segmentation in the image mask and in the composite image. The regions in the image mask are likely to be less reliable because they contain anatomical abnormalities. This may be a way to draw attention to regions of the anatomical segmentation that may be less accurate or inaccurate. They may also be provided to the user as a warning about the accuracy of the segmentation in the image mask or, for example, within a tumor. The highlighting of these regions may be considered a warning signal.

[0022] In another embodiment, execution of the machine-executable instructions further causes the computing system to receive a second segmentation of the input tomographic medical image in response to inputting the input tomographic medical image to the segmentation module. Execution of the machine-executable instructions further causes the computing system to identify at least one warning zone in the anatomical segmentation by determining an area of ​​the anatomical segmentation that differs from the second segmentation by more than a predetermined threshold. For example, a warning zone can be provided if the distance between the anatomical segmentation and the second segmentation at a particular point changes by more than a threshold distance. Identification of the at least one warning zone can be provided as a warning signal. Execution of the machine-executable instructions further causes the computing system to mark the warning zone on the composite image. For example, this can be such that the segmentation has a different appearance or a different color.

[0023] In another embodiment, the memory further stores an anatomical anomaly segmentation module configured to output an anomaly segmentation restricted to an image portion of the input tomographic medical image. Execution of the machine-executable instructions further causes the computing system to determine an input portion of the input tomographic medical image from locations of anatomical anomalies defined in the image mask. Execution of the machine-executable instructions further causes the computing system to receive the anomaly segmentation by applying the anomaly segmentation module to the input portion of the tomographic medical image. The anomaly segmentation module may be a segmentation module designed specifically to segment anatomical anomalies or particular types of anatomical anomalies. For example, the anomaly segmentation module may be designed to segment particular types of tumors or other medical conditions.

[0024] In another embodiment, the anatomical abnormality segmentation module is any one of a neural network, such as a deep learning trained neural network, threshold segmentation, fuzzy logic based image segmentation, region-based image segmentation, and edge-based image segmentation.

[0025] In another embodiment, execution of the machine-executable instructions further causes the computing system to modify the anatomical segmentation using the abnormality segmentation. For example, the abnormality segmentation can be spliced ​​or combined with the anatomical segmentation. Or, for example, a region of the anatomical segmentation outside the abnormality segmentation can be used normally and then simply combined with the abnormality segmentation. This can provide a better definition of the anatomical region, as well as a more accurate definition of the region of the anatomical abnormality.

[0026] In another embodiment, execution of the machine-executable instructions further causes the computing system to receive configuration data from a user interface configured to configure the anatomical abnormality segmentation module. Execution of the machine-executable instructions causes the computing system to modify the operation of the anatomical abnormality segmentation module before applying the anatomical abnormality segmentation module to the input portion of the tomographic medical image. For example, the user interface may be used to define seed points, gray value intervals, selection of a segmentation module type or mode, etc. This may be beneficial as it may provide a more accurate segmentation of the anatomical abnormality.

[0027] In another embodiment, the image processing neural network is a cycleGAN neural network.

[0028] In another embodiment, the image processing neural network is a variational autoencoder. In another embodiment, the medical system further comprises a tomographic medical imaging system. The memory further comprises imaging system control commands configured to control the medical imaging system to acquire medical imaging data describing a selected anatomical region of the subject. Execution of the machine-executable instructions further causes the computing system to acquire the medical imaging data by controlling the tomographic medical imaging system with the imaging system control commands. Execution of the machine-executable instructions further causes the computing system to reconstruct an input tomographic medical image from the medical imaging data.

[0029] In another embodiment, the memory further includes a radiation treatment planning module configured to output radiation treatment system control commands, the radiation treatment system control commands configured to control the radiation treatment system to irradiate one or more target zones within the selected anatomical region in response to receiving the locations of the one or more target zones, and to control anatomical segmentation and, preferably, the location of one or more protection regions within the selected anatomical region to minimize irradiation. Execution of the machine-executable instructions further causes the computing system to output radiation treatment system control commands in response to receiving the locations of the one or more target zones, the anatomical segmentation, and, preferably, the location of the one or more protection regions. This embodiment may be beneficial because the use of anatomical segmentation may provide more accurate and safer radiation treatments.

[0030] In some cases, the location of the anatomical abnormality may be identical to one or more target zones.

[0031] In another embodiment, the medical system further comprises a radiation therapy system. Execution of the machine-executable instructions further causes the computing system to irradiate one or more target zones by controlling the radiation therapy system with the radiation therapy control commands. This embodiment may be beneficial because it may provide more accurate irradiation of the one or more target zones and may also provide a higher degree of safety for the subject.

[0032] In another embodiment, the radiation therapy system is a LINAC system.

[0033] In another embodiment, the radiation therapy system is a gamma knife.

[0034] In another embodiment, the radiation therapy system is an X-ray radiation therapy system.

[0035] In another aspect, the present invention provides a computer program comprising machine-executable instructions and an image processing neural network. Both the machine-executable instructions and the image processing neural network may be executable by a computing system. The image processing neural network is configured to output a modified tomographic medical image describing the selected anatomical region of the subject in response to receiving an input tomographic medical image describing the selected anatomical region. The image processing neural network is further configured to output the modified tomographic medical image such that anatomical abnormalities depicted in the input tomographic medical image are removed. Execution of the machine-executable instructions causes the computing system to receive the input tomographic medical image.

[0036] Execution of the machine-executable instructions causes the computing system to receive a modified tomographic medical image in response to inputting the input tomographic medical image into an image processing neural network. Execution of the machine-executable instructions further causes the computing system to receive an anatomical segmentation of the modified tomographic medical image in response to inputting the modified tomographic medical image to an image segmentation module. The anatomical segmentation divides the modified tomographic medical image into a plurality of anatomical regions. Execution of the machine-executable instructions further causes the computing system to provide the anatomical segmentation as a segmentation of the input tomographic medical image.

[0037] In another aspect, the present invention provides a method of medical imaging. The method includes receiving an input tomographic medical image. The input tomographic medical image describes a selected anatomical region of a subject and selects an anatomical abnormality. The method further includes receiving a modified tomographic medical image in response to inputting the input tomographic medical image into an image processing neural network. The image processing neural network is configured to output a modified tomographic medical image describing the selected anatomical region of the subject in response to receiving the input tomographic medical image describing the selected anatomical region. The image processing neural network is further configured to output the modified tomographic medical image such that the anatomical abnormality depicted in the input tomographic medical image is removed. The method further includes receiving an anatomical segmentation of the modified tomographic medical image in response to inputting the modified tomographic medical image into an image segmentation module. The anatomical segmentation divides the modified tomographic medical image into a plurality of anatomical regions. The method further includes providing the anatomical segmentation as a segmentation of the input tomographic medical image.

[0038] It is understood that one or more of the above-described embodiments of the present invention may be combined, provided that the combined embodiments are not mutually exclusive.

[0039] As will be appreciated by one of skill in the art, aspects of the present invention may be embodied as an apparatus, a method, or a computer program product. Accordingly, aspects of the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment (including firmware, resident software, microcode, etc.), or an embodiment combining software and hardware aspects, all of which may be referred to generally herein as a "circuit," "module," or "system." Further, aspects of the present invention may take the form of a computer program product embodied in one or more computer-readable medium(s) having computer-executable code embodied thereon.

[0040] Any combination of one or more computer-readable media may be utilized. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. As used herein, "computer-readable storage medium" encompasses any tangible storage medium capable of storing instructions executable by a processor or computing system of a computing device. The computer-readable storage medium may also be referred to as a computer-readable non-transitory storage medium. The computer-readable storage medium may also be referred to as a tangible computer-readable medium. In some embodiments, the computer-readable storage medium may also be capable of storing data that can be accessed by the computing system of a computing device. Examples of computer-readable storage media include, but are not limited to, floppy disks, magnetic hard disk drives, solid-state hard disks, flash memory, USB thumb drives, random access memory (RAM), read-only memory (ROM), optical disks, magneto-optical disks, and computing system register files. Examples of optical disks include compact discs (CDs) and digital versatile discs (DVDs), such as CD-ROM, CD-RW, CD-R, DVD-ROM, DVD-RW, or DVD-R discs. The term computer-readable storage medium also refers to various types of storage media that can be accessed by a computer device over a network or communications link. For example, data may be retrieved via a modem, over the Internet, or over a local area network. Computer-executable code embodied on a computer-readable medium may be transmitted using any suitable medium, including, but not limited to, wireless, wireline, fiber optic cable, RF, etc., or any suitable combination of the foregoing.

[0041] A computer-readable signal medium may include, for example, a propagated data signal having computer-executable code embodied therein, in baseband or as part of a carrier wave. Such a propagated signal may take any of a variety of forms, including, but not limited to, electromagnetic, optical, or any suitable combination thereof. A computer-readable signal medium is not a computer-readable storage medium but may be any computer-readable medium that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0042] "Computer memory" or "memory" is an example of a computer-readable storage medium. Computer memory is any memory directly accessible to a computing system. "Computer storage" or "storage" is a further example of a computer-readable storage medium. Computer storage is any non-volatile memory computer-readable storage medium. In some embodiments, computer storage may be computer memory, and vice versa.

[0043] As used herein, a "computing system" encompasses electronic components capable of executing programs, machine-executable instructions, or computer-executable code. References to a computing system, including examples of a "computing system," should be interpreted as including two or more computing systems or processing cores, as the case may be. A computing system may be, for example, a multi-core processor. A computing system may also refer to a collection of computing systems within a single computing system or distributed among multiple computing systems. The term computing system should also be interpreted as referring to a collection or network of computing devices, possibly each comprising a processor or computing system. Machine-executable code or instructions may be executed by multiple computing systems or processors, which may be within the same computing device or distributed across multiple computing devices.

[0044] Machine-executable instructions or computer-executable code may comprise instructions or programs that cause a processor or other computing system to perform aspects of the present invention. Computer-executable code for performing operations for aspects of the present invention may be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as the "C" programming language or similar programming languages, and compiled into machine-executable instructions. In some cases, the computer-executable code may be in the form of a high-level language or in pre-compiled form and may be used in conjunction with an interpreter that generates machine-executable instructions on the fly. In other cases, the machine-executable instructions or computer-executable code may form a program for a programmable logic gate array.

[0045] The computer executable code may run entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer, partially on a remote computer, or entirely on a remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be to an external computer (e.g., via the Internet using an Internet Service Provider).

[0046] Aspects of the present invention will be described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block or portion of a block in the flowcharts, diagrams, and / or block diagrams, where applicable, can be implemented by computer program instructions in the form of computer-executable code. Furthermore, it should be noted that combinations of blocks in different flowcharts, diagrams, and / or block diagrams can be combined, if not mutually exclusive. These computer program instructions can be provided to a general-purpose computer, special-purpose computer, or other programmable data processing device computing system to produce a machine, such that the instructions, executed via the computer or other programmable data processing device computing system, create means for performing the functions / acts specified in the flowchart and / or block diagram block or blocks.

[0047] These machine-executable instructions or computer program instructions may be stored on a computer-readable medium that can instruct a computer, other programmable data processing apparatus, or other device to function in a particular manner, such that the instructions stored on the computer-readable medium produce an article of manufacture including instructions that implement the function / acts specified in a block or blocks of the flowcharts and / or block diagrams.

[0048] The machine-executable instructions or computer program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps executed on the computer, other programmable apparatus, or other device to produce a computer-implemented process, such that the instructions executing on the computer or other programmable apparatus provide a process for implementing the function / operation specified in the flowchart and / or block diagram block or blocks.

[0049] As used herein, a "user interface" is an interface that allows a user or operator to interact with a computer or computer system. A "user interface" is sometimes referred to as a "human interface device," and a user interface can provide information or data to an operator and / or receive information or data from an operator. A user interface can allow input from an operator to be received by a computer and can provide output from the computer to a user. In other words, a user interface can allow an operator to control or manipulate a computer, and an interface can allow a computer to show the effects of the operator's control or manipulation. The display of data or information on a display or graphical user interface is an example of providing information to an operator. Receiving data via a keyboard, mouse, trackball, touchpad, pointing stick, graphics tablet, joystick, gamepad, webcam, headset, pedals, wired gloves, remote control, and accelerometer are all examples of user interface components that allow information or data to be received from an operator.

[0050] As used herein, a "hardware interface" encompasses an interface that allows a computer system to interact with and / or control external computing devices and / or equipment. A hardware interface may allow a computer system to send control signals or instructions to external computing devices and / or equipment. A hardware interface may also allow a computing system to exchange data with external computing devices and / or equipment. Examples of hardware interfaces include, but are not limited to, a universal serial bus, an IEEE 1394 port, a parallel port, an IEEE 1284 port, a serial port, an RS-232 port, an IEEE-488 port, a Bluetooth connection, a wireless local area network connection, a TCP / IP connection, an Ethernet connection, a control voltage interface, a MIDI interface, an analog input interface, and a digital input interface.

[0051] As used herein, "display" or "display device" encompasses an output device or user interface adapted to display images or data. A display can output visual, audio, and / or tactile data.

[0052] Examples of displays include, but are not limited to, computer monitors, television screens, touch screens, tactile electronic displays, Braille screens, cathode ray tubes (liquid), memory tubes, bi-stable displays, electronic paper, vector displays, flat panel displays, vacuum fluorescent displays (VF), light emitting diode (LED) displays, electroluminescent displays (ELD), plasma display panels (PDP), liquid crystal displays (LCD), organic light emitting diode displays (OLED), projectors, and head mounted displays.

[0053] Medical imaging data is defined herein as recorded measurements made by a tomographic medical imaging system that describe a subject. Medical imaging data may be reconstructed into a medical image. A tomographic medical image is defined herein as a reconstructed two-dimensional or three-dimensional visualization of anatomical data contained within the medical imaging data. This visualization may be performed using a computer.

[0054] K-space data is defined herein as the recorded measurements of radio frequency signals emitted by atomic spins using the antenna of a magnetic resonance machine during a magnetic resonance imaging scan. Magnetic resonance data is an example of tomographic image data.

[0055] A magnetic resonance imaging (MRI) image or MR image is defined herein as a reconstructed two-dimensional or three-dimensional visualization of anatomical data contained within magnetic resonance imaging data. This visualization can be performed using a computer. A magnetic resonance image is an example of a tomographic image.

[0056] In the following, preferred embodiments of the invention will be described, by way of example only, with reference to the drawings, in which: [Brief explanation of the drawings]

[0057] [Figure 1] 1 shows an example of a medical system. [Figure 2] 2 shows a flowchart illustrating a method of using the medical system of FIG. 1. [Figure 3] 1 illustrates a further example of a medical system. [Figure 4] 4 shows a flowchart illustrating a method of using the medical system of FIG. 3. [Figure 5] 1 illustrates a further example of a medical system. [Figure 6] 1 illustrates a further example of a medical system. [Figure 7] 7 shows a flowchart illustrating a method of using the medical system of FIG. 6. [Figure 8]1 shows an example of an input tomographic medical image and a modified tomographic medical image. [Figure 9] 1 illustrates an example of a model-based segmentation module. DETAILED DESCRIPTION OF THE INVENTION

[0058] Like numbered elements in these figures are equivalent elements or perform the same function. An element as described above is not necessarily discussed in a subsequent figure if there is functional equivalence.

[0059] 1 illustrates an example medical system 100. The medical system 100 is shown as including a computer 102. The computer 102 may represent one or more computers in one or more locations. For example, the computer 102 may be a virtual machine available within a cloud-based system.

[0060] The computer 102 is shown as including a computing system 104. The computing system 104 may represent one or more computing systems or processing cores located in one or more locations. The computing system 104 is shown in communication with an optional hardware interface 106. If present, the hardware interface 106 may be used to control other components, such as a medical imaging system or a magnetic resonance imaging system. The computing system 104 is further shown in communication with a user interface 108. If present, the user interface 108 may be used by an operator to control and interact with the medical system 100. The computing system 104 is further shown in communication with a memory 110. The memory 110 is intended to represent various types of memory accessible to the computing system 104. For example, the memory 110 may be a non-transitory storage medium. In other examples, the memory 110 may be conventional computer RAM.

[0061] The memory 110 is shown as including machine-executable instructions 120. The machine-executable instructions 120 enable the computing system 104 to perform various data manipulation, image manipulation, and computational tasks. In some examples, the machine-executable instructions 120 may also enable control of a medical imaging system via the hardware 106. The memory 110 is further shown as including an image processing neural network 124. The memory 110 is further shown as including an input tomographic medical image 124. The memory 110 is further shown as including a corrected tomographic medical image 126. The image processing neural network 122 is configured to receive the input tomographic medical image 124 and, in response, output a corrected tomographic medical image 126. The input tomographic medical image 124 may have anatomical abnormalities, which may be removed so that they are no longer present in the corrected tomographic medical image 126.

[0062] The memory 110 is further shown to include an image segmentation module 128 that is used to segment the corrected tomographic medical image 126. The memory 110 is further shown to include an anatomical segmentation 130 that is received from the image segmentation module 128 in response to receiving the corrected tomographic medical image 126. The anatomical segmentation 130 can be used as a segmentation of the input tomographic medical image 124. This may have the advantage that the anatomical segmentation 130 is not affected by the presence of anatomical abnormalities. For example, this may allow the input tomographic medical image 124 to be accurately segmented into different anatomical regions, although anatomical abnormalities such as tumors may be present.

[0063] FIG. 2 shows a flowchart illustrating a method for operating the medical system 100 of FIG. 1. First, in step 200, an input tomographic medical image 124 is received. This may be received, for example, by retrieving it from memory 110, via a network connection, or by acquiring or reconstructing the input tomographic medical image 124. Next, in step 202, a modified tomographic medical image 126 is received in response to inputting the input tomographic medical image 124 to an image processing neural network. The image processing neural network is configured to output the modified tomographic medical image 126 such that anatomical abnormalities depicted in the input tomographic medical image 124 have been removed. The method further includes, in step 206, providing an anatomical segmentation 130 as a segmentation of the input tomographic medical image 124. Providing the anatomical segmentation may take different forms in different examples. In one example, the anatomical segmentation may be overlaid on the input tomographic medical image 124. In another example, the anatomical segmentation may be associated with the input tomographic medical image 124, for example, the input tomographic medical image 124 may be stored in a so-called DICOM file, in which case the anatomical segmentation 130 may be added to this DICOM file.

[0064] Figure 3 shows a further example of a medical system 300. The system 300 is similar to the system 100 shown in Figure 1, except that it further comprises a tomographic medical imaging system 302. The tomographic medical imaging system 302 is intended to represent various system types. For example, it may be a magnetic resonance imaging system, a positron emission tomography system, a computed tomography system, a single photon emission tomography system, or a combined modality such as a combined CT and MR system or a combined CT and positron emission tomography system.

[0065] The tomographic medical imaging system 302 includes a subject support 308 that supports a subject 306 within an imaging zone 304 .

[0066] The memory 110 is further shown as containing imaging system control commands 320 used to control the operation and functionality of the tomographic medical imaging system 302. The memory 110 is further shown as containing collected medical imaging data 322 describing the imaging zone 304. The medical imaging data 322 was collected or measured by controlling the tomographic medical imaging system using the imaging system control commands 320.

[0067] The memory 110 is further shown to include an image mask 324 determined by comparing the input tomographic medical image 124 and the modified tomographic medical image 126. For example, one can be subtracted from the other and the difference can be thresholded. This can identify the location of anatomical abnormalities. In another example, the image processing neural network 122 can be designed or trained to also output the image mask 324. The memory 110 is further shown to include a geometric measure 326 that describes anatomical abnormalities in one or more of the multiple anatomical regions using the image mask 324 and the anatomical segmentation 130. For example, this may be a measure of the volume fraction of anatomical abnormalities within some or all of the multiple anatomical regions relative to others, such as penetration distance or surface area.

[0068] The memory is further shown as containing a composite image 328, which is a combination of the anatomical segmentation 130 and the input tomographic medical image 124. The memory 110 is further shown as containing enhancement regions 330 determined for the composite image 328. These may be regions of the anatomical segmentation 130 that fall within the location of the image mask 324. They may be enhanced, for example, by changing their appearance, color, or other optical properties so that the operator recognizes that the anatomical segmentation 130 falls within an anatomical abnormality.

[0069] FIG. 4 shows a flowchart illustrating a method for operating the medical system 300 of FIG. 3. First, in step 400, medical imaging data 322 is acquired by controlling the tomographic medical imaging system 302 with image system control commands 320. Next, in step 402, an input tomographic medical image 124 is reconstructed from the medical imaging data 322. The method then repeats steps 200, 202, 204, and 206 as shown in FIG. 2. After step 206 is performed, the method proceeds to step 404. In step 404, an image mask 324 is provided. The image mask 324 provides the location of anatomical abnormalities in the input tomographic medical image 124. This can be done, for example, by subtracting the input tomographic medical image 124 from the corrected tomographic medical image 126. Next, in step 406, geometric measurements 326 describing anatomical abnormalities in at least one of the plurality of anatomical regions are determined using the image mask 324 and the anatomical segmentation 130. Next, in step 408, a composite image 328 is rendered. The composite image 328 overlays the anatomical segmentation 130 onto the input tomographic medical image 124. Finally, in step 410, regions 330 of the anatomical segmentation 130 within the image mask 324 are highlighted in the composite image 328.

[0070] 5 shows a further example of a medical system 500. The medical system 500 is similar to that shown in FIG. 3, except that in this case the tomographic medical imaging system 302 is a magnetic resonance imaging system 502.

[0071] The magnetic resonance imaging system 502 includes a magnet 504. The magnet 504 is a superconducting cylindrical magnet with a bore 506. Different types of magnets can be used, including both split cylindrical magnets and so-called open magnets. Split cylindrical magnets are similar to standard cylindrical magnets, except that the cryostat is divided into two sections to allow access to the magnet's isosurface. Such magnets can be used, for example, in conjunction with charged particle beam therapy. Open magnets have two magnet sections, one above the other, with a space between them large enough to accommodate the subject, an arrangement of the two sections similar to the area of ​​a Helmholtz coil. Open magnets are popular because they provide less subject confinement. Inside the cryostat of a cylindrical magnet is a collection of superconducting coils.

[0072] Within the bore 506 of the cylindrical magnet 504 is the imaging zone 304, where the magnetic field is strong and sufficiently uniform to perform magnetic resonance imaging. A field of view 508 is shown within the imaging zone 304. Magnetic resonance data typically acquired for the field of view 508. A subject 306 is shown supported by a subject support 308 such that at least a portion of the subject 306 is within the imaging zone 304 and the field of view 508.

[0073] Also within the magnet bore 506 are a set of magnetic field gradient coils 510 used for preliminary magnetic resonance data acquisition to spatially encode magnetic spins within the imaging zone 304 of the magnet 504. The gradient coils 510 are connected to a gradient coil power supply 512. The magnetic field gradient coils 510 are intended to be representative. Typically, the magnetic field gradient coils 510 include three separate coil sets for spatial encoding in three orthogonal spatial directions. The gradient power supply supplies current to the gradient coils. The current supplied to the magnetic field gradient coils 510 is controlled as a function of time and can be ramped or pulsed.

[0074] Adjacent to the imaging zone 304 is a radio frequency coil 514 for manipulating the orientation of magnetic spins within the imaging zone 304 and for receiving radio transmissions from the spins within the imaging zone 304. A radio frequency antenna may include multiple coil elements. A radio frequency antenna may also be referred to as a channel or antenna. The radio frequency coil 514 is connected to a radio frequency transceiver 516. The radio frequency coil 514 and the radio frequency transceiver 516 may be replaced by separate transmit and receive coils and separate transmitters and receivers. It is understood that the radio frequency coil 514 and the radio frequency transceiver 516 are representative. The radio frequency coil 514 is also intended to represent a dedicated transmit antenna and a dedicated receive antenna. Similarly, the transceiver 516 may also represent a separate transmitter and receiver. The radio frequency coil 514 may also have multiple receive / transmit elements, and the radio frequency transceiver 516 may have multiple receive / transmit channels. For example, if a parallel imaging technique such as sensing is performed, the radio frequency 514 may have multiple coil elements.

[0075] The transceiver 516 and the tilt controller 512 are shown connected to the hardware interface 506 of the computing system 502 .

[0076] In this example, the imaging system control commands 320 are pulse sequence commands 320'. The pulse sequence commands 320' are either commands or data that can be converted into commands used to control the magnetic resonance imaging system 502 to acquire k-space data. In this example 110, the medical imaging data is k-space data 322'.

[0077] 6 shows a further example of a medical system 600 similar to the medical system 500 shown in FIG. 5, except that it further comprises a radiation therapy system 602. In this example, a magnetic resonance imaging system 502 is used to guide the radiation therapy system 602. This is intended to be exemplary. For example, the magnetic resonance imaging system 502 could be replaced with a computed tomography system.

[0078] The radiation treatment system 602 includes a gantry 604 and a radiation treatment source 606. The gantry 604 is for rotating the radiation treatment source 606 about a gantry axis of rotation or axis of rotation 614. Adjacent to the radiation treatment source 606 is a collimator 608.

[0079] The magnet 504 shown in this embodiment is a standard cylindrical superconducting magnet. The magnet 504 has a cryostat 610 with an internal superconducting coil 612. Alternatively, a segmented magnet may be used.

[0080] Within the subject 306 is a target zone 618. The gantry axis of rotation 614 is, in this example, coaxial with the cylindrical axis of the magnet 504. The radiation source 606 is aimed at the axis of rotation 614 such that the radiation source has a target volume about the axis of rotation 614.

[0081] The subject support 308 is positioned such that a target zone 618 is located on an axis 614 of gantry rotation. The radiation source 606 is shown passing through a collimator 608 and generating a radiation beam 616 that passes through the target zone 618. As the radiation source 606 rotates about the axis 614, the target zone 618 is always targeted by the radiation beam 616. The radiation beam 616 passes through a cryostat 610 of the magnet 604. The gradient coils may have a gap separating them into two sections. If present, this gap reduces attenuation of the radiation beam 616 by the gradient coils 510. In some embodiments, the radio frequency coil 514 may have a gap or be separated to reduce attenuation of the radiation beam 616.

[0082] Memory 110 is further shown to include a radiation therapy planning module 620. Radiation therapy planning module 620 is configured to output radiation therapy system control commands configured to control radiation therapy system 602 in response to receiving one or more target zones 618, anatomical segmentation 130, and optionally, the location of one or more protection regions within the anatomical region selected to minimize irradiation. Memory is further shown to include radiation therapy system control commands 622 generated using radiation therapy planning module 620.

[0083] Figure 7 illustrates a method of operating the medical system of Figure 6. First, steps 400, 402, 200, 202, 204, and 206 are performed as shown in Figure 4. Next, in step 700, in response to receiving the location of one or more target zones, the anatomical segmentation, and preferably the location of one or more protection regions, radiation therapy system control commands are output, and in 702, the one or more target zones are irradiated by controlling the radiation therapy system with the radiation therapy system control commands.

[0084] The performance of anatomical image segmentation algorithms is typically trained on non-pathological imaging and is therefore often plagued by the presence of abnormal imaging or anatomical abnormalities, such as tumors, fractures, or other types of lesions.

[0085] An embodiment can provide a system that can improve the segmentation of pathological scans by utilizing artificial intelligence. Its main component is a component that estimates the normal-looking correspondence of abnormal images, upon which anatomical segmentation algorithms are expected to perform more reliably. This segmentation result is then mapped to the original abnormal image and visualized along with reliability information derived from the mapping.

[0086] Many segmentation methods that learn geometric and / or gray value characteristics from data (e.g., some model-based segmentation or other machine and deep learning-based methods) are trained on healthy controls and are therefore optimized for healthy controls, and may fail in the case of abnormal contrast caused by lesions or pathologies. For example, segmentation of subcortical brain structures may fail in the case of brain tumors or heavy white matter lesion burdens. This is because such pathological contrasts are highly variable in terms of size, location, and appearance, making it difficult to fully cover their expected variations in training. Nevertheless, typical expected gray values ​​may be significantly altered by such pathologies, which can significantly hinder correct segmentation.

[0087] An example may address the above-mentioned problem by using artificial intelligence to artificially "remove" any pathological contrast, abnormality, or lesion in the test image to be segmented. The abnormal test image is then converted and modified into an image of normal contrast, along with a mask indicating which image regions contain abnormalities. A segmentation algorithm is then applied to the converted image with normal contrast, where it is expected to produce accurate and robust segmentation results. This result is then transferred to the original abnormal image. Any portions of the segmentation result that fall in areas of the image mask, i.e., areas containing abnormalities, are marked (e.g., displayed in a more transparent manner) to indicate that they are unreliable in the presence of abnormalities. Furthermore, the segmentation results can also be applied to the original abnormal test image, comparing the segmentations of the abnormal image and the converted normal-appearing image, and marking areas where both segmentations differ significantly (even outside the image mask) as "unreliable."

[0088] The example is: A fully automatic statistical segmentation algorithm (image segmentation module 128) (e.g., model-based segmentation, voxel-based segmentation, etc.); an automatic algorithm (image processing neural network 122) that automatically removes pathological, abnormal and lesion contrasts, for example based on artificial intelligence / generative adversarial networks, and converts the abnormal image into a normal-looking image without pathological contrasts, abnormalities and lesions, along with a (binary) image mask indicating which image regions contain abnormalities and are therefore corrected; a graphical display of the segmentation results, where portions of the segmentation corresponding to masked image regions are visually modified to indicate that they are unreliable in the presence of anomalies; A comparator,compares the segmentation results of the original abnormal image with,the segmentation results of the transformed healthy-appearing,image, and the results are fed into a graphical display of the,segmentation results, which further indicates areas of significant,differences between both segmentation results, even outside the,mask. Further segmentation methods and user interfaces specifically tailored to segment abnormal tissue within a marked volume of interest, wherein the volume of interest is set to correspond to a region of an image mask, and the user interface is used to manually select an appropriate segmentation method and manually set further boundary conditions (e.g., seed points, gray value intervals, etc.). The present invention may include one or more of the following features:

[0089] By using automatic algorithms from artificial intelligence to artificially remove abnormalities, pathological contrasts, and lesion contrasts, the abnormal image (input tomographic medical image 124) is transformed into an image with normal-looking contrasts (corrected tomographic medical image 124).

[0090] Several images are shown in Figure 8. Image 124 is an example of an input tomographic medical image, which is a magnetic resonance image, in which an anatomical abnormality 700, in this case a tumor, is clearly present. Image 126 is an example of a modified tomographic medical image 126 generated by inputting input tomographic medical image 124 into image processing neural network 122. Tumor 700 has been removed from the image. Image 702 is a difference image between images 124 and 126. In this case, there is a high contrast region 704 indicating the location of anatomical abnormality 700 or tumor. Image mask 324 can be constructed by thresholding image 702. This identifies the region labeled 704 within the mask.

[0091] After the corrected tomographic medical image 126 is generated, a statistical segmentation algorithm (image segmentation module 128), such as model-based segmentation, is applied to the transformed normal-appearing image (see FIG. 9 below). Due to the absence of pathological contrast, abnormalities, and lesions, the segmentation results are expected to be more accurate and robust.

[0092] 9 shows several views of a modified tomographic medical image 126 to which an anatomical segmentation 130 has been applied, showing a model-based segmentation of subcortical structures and cortical tissue classes (gray matter, white matter, cerebrospinal fluid) in a (normal-appearing) MR volume.

[0093] The generated segmentation (anatomical segmentation 130) is then added back to the original anomaly image (input tomographic medical image 124). The generated image mask defines the regions where the original and transformed images differ, and the portions of the segmentation result that fall within the image mask are displayed in a modified manner (e.g., with greater transparency) to indicate that they are unreliable due to the presence of an anomaly.

[0094] Optionally, the segmentation result (anatomical segmentation 130) can also be applied to the original abnormal image (input tomographic medical image 126). Comparing the segmentations of the abnormal image and the normal-appearing image, all regions with significant deviations (even outside the image mask) define unreliable segmentations and can be visually indicated (again, using transparency or any other suitable visual indicator).

[0095] Furthermore, if a dedicated segmentation algorithm for a particular type of abnormality is available (this could be a simple region-growing algorithm in a spatially restricted region for lesion segmentation), the second segmentation step can be triggered by a corresponding user interface, which involves manual selection of the appropriate segmentation algorithm and manual definition of additional constraints, e.g., defining a volume of interest or a seed point.

[0096] While the invention has been illustrated and described in detail in the drawings and foregoing description, such illustration and description are to be considered illustrative or exemplary and not restrictive, and the invention is not limited to the disclosed embodiments.

[0097] Other variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure, and the appended claims. In the claims, the word "comprising" does not exclude other elements or steps, and the indefinite articles "a" or "an" do not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. A computer program can be stored / distributed on a suitable medium, such as an optical storage medium or a solid-state medium, together with or supplied as part of other hardware, but can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. Any reference signs in the claims should not be construed as limiting the scope. [Explanation of symbols]

[0098] 100 Medical Systems 102 Computer 104 Computing Systems 106 Hardware Interface 108 User Interface 110 memory 120 machine-executable instructions 122 Image Processing Neural Networks 124 input tomographic medical images 126 Corrected Tomographic Medical Images 128 Image Segmentation Module 130 Anatomical Segmentation Receive 200 input tomography medical images 202 receiving a modified tomographic medical image in response to inputting the input tomographic medical image into the image processing neural network. 204 receiving an anatomical segmentation of the modified tomographic medical image in response to inputting the modified tomographic medical image to the image segmentation module. 206 Provides anatomical segmentation as a segmentation of input tomographic medical images 300 Medical Systems 302 Tomographic Medical Imaging System 304 Imaging Zone 306 Subject 308 Support 320 Imaging System Control Command 320' Pulse sequence command 322 Medical Imaging Data 322' k-space data 324 Image Mask 326 Geometric Measurements 328 composite images 330 Priority Areas 400 Medical imaging data is acquired by controlling the tomographic medical imaging system with imaging system control commands. 402 Reconstruction of input tomographic medical images from medical imaging data 404 Provides an image mask that identifies the location of anatomical abnormalities in an input tomographic medical image. 406 Using the image mask and anatomical segmentation to determine geometric measurements describing anatomical abnormalities in at least one of a plurality of anatomical regions. 408 Rendering a composite image that overlays anatomical segmentation onto an input tomographic medical image 410 Highlighting regions of anatomical segmentation within image masks in composite images 500 Medical Systems 502 Magnetic resonance imaging equipment 504 Magnet 506 Magnet Bore 508 Field of view 510 Magnetic Gradient Coil 512 Gradient magnetic field coil power supply 514 High Frequency Coil 516 Transceiver 600 Medical Systems 602 Radiation Therapy System 604 Gantry 606 Radiation therapy source 608 Collimator 610 Cryostat 612 Superconducting Coil 614 Rotational Axis 616 Radiation Beam 618 Target Zone 620 Radiation Treatment Planning Module 622 Radiation Therapy System Control Command 700. Outputting radiation therapy system control commands in response to receiving the location and anatomical segmentation of one or more target zones. 702 Radiation therapy system control commands control a radiation therapy system to irradiate one or more target zones. 800 Anatomical Abnormalities 802 difference images 804 High-contrast regions for generating image masks

Claims

1. 1. A medical system comprising: a memory storing executable instructions and an image processing neural network configured to, in response to receiving an input tomographic medical image describing a selected anatomical region, output a modified tomographic medical image describing the selected anatomical region of the subject, the image processing neural network further configured to output the modified tomographic medical image such that an anatomical abnormality shown in the input tomographic medical image is removed; a computing system, wherein execution of the machine-executable instructions causes the computing system to: receiving the input tomographic medical image; receiving the corrected tomographic medical image in response to inputting the input tomographic medical image into the image processing neural network; receiving an anatomical segmentation of the corrected tomographic medical image in response to inputting the corrected tomographic medical image into an image segmentation module, the anatomical segmentation dividing the corrected tomographic medical image into a plurality of anatomical regions; providing the anatomical segmentation as a segmentation of the input tomographic medical image; A medical system configured to:

2. Execution of the machine-executable instructions further comprises: comparing the input tomographic medical image with the corrected tomographic medical image; receiving an image mask from the image processing neural network, the image processing neural network being configured to output the image mask; 2. The medical system of claim 1, further comprising: a step of providing an image mask that identifies the location of an anatomical abnormality in the input tomographic medical image by any one of the steps of:

3. 3. The medical system of claim 2, wherein execution of the machine-executable instructions is further configured to cause the computing system to perform the step of using the image mask and the anatomical segmentation to determine geometric measurements that describe the anatomical abnormality in at least one of the plurality of anatomical regions.

4. 4. The medical system of claim 2 or 3, wherein execution of the machine-executable instructions is configured to cause the computing system to further perform the step of rendering a composite image that overlays the anatomical segmentation on the input tomographic medical image.

5. The medical system of claim 4 , wherein execution of the machine-executable instructions is configured to cause the computing system to further perform the step of highlighting regions of anatomical segmentation within an image mask in the composite image.

6. Execution of the machine-executable instructions further causes the computing system to: receiving a second segmentation of the input tomographic medical image in response to inputting the input tomographic medical image into the image segmentation module; identifying at least one warning zone in the anatomical segmentation by determining areas of the anatomical segmentation that differ from the second segmentation by more than a predetermined threshold; marking the warning zone on the composite image; The medical system according to claim 4 or 5, configured to execute the following:

7. The memory further stores an anatomical anomaly segmentation module configured to output an anomaly segmentation restricted to an image portion of the input tomographic image, and execution of the machine-executable instructions further causes the computing system to: determining an input portion of the input tomographic medical image from the location of an anatomical abnormality defined in the image mask; receiving the anomaly segmentation by applying the anatomical anomaly segmentation module to an input portion of the tomographic medical image; 7. The medical system according to claim 2, configured to execute the following:

8. 8. The medical system of claim 7, wherein execution of the machine-executable instructions is further configured to cause the computing system to perform the step of correcting the anatomical segmentation using the abnormal segmentation.

9. Execution of the machine-executable instructions further causes the computing system to: receiving configuration data from a user interface configured to configure the anatomical abnormality segmentation module; modifying the operation of the anatomical anomaly segmentation module prior to applying the anatomical anomaly segmentation module to the input portion of the tomographic medical image; 9. The medical system according to claim 7 or 8, configured to cause the system to execute the following:

10. The medical system of any one of claims 1 to 9, wherein the image processing neural network is any one of a cyclic GAN neural network, a variational autoencoder, and a U-Net.

11. The medical system further comprises a tomographic medical imaging system, the memory further comprising imaging system control commands configured to control the tomographic medical imaging system to acquire medical imaging data describing a selected anatomical region of the subject, and execution of the machine-executable instructions causes the computing system to: acquiring the medical imaging data by controlling the tomographic medical imaging system using the imaging system control commands; reconstructing the input tomographic medical image from the medical imaging data; The medical system of claim 1 , configured to:

12. 12. The medical system of claim 1, wherein the memory further comprises a radiation treatment planning module configured to output radiation treatment system control commands configured to control a radiation treatment system to irradiate one or more target zones in the selected anatomical region in response to receiving the locations of the one or more target zones, the anatomical segmentation, and preferably the locations of one or more protection regions within the selected anatomical region that minimize irradiation, and wherein execution of the machine-executable instructions further causes the computing system to perform the step of outputting the radiation treatment system control commands in response to receiving the locations of the one or more target zones, the anatomical segmentation, and preferably the locations of the one or more protection regions.

13. 13. The medical system of claim 12, further comprising the radiation therapy system, and wherein execution of the machine-executable instructions is configured to cause the computing system to further perform the step of irradiating the one or more target zones by controlling the radiation therapy system with the radiation therapy system control commands.

14. 1. A computer program comprising: machine-executable instructions; and an image processing neural network; wherein the image processing neural network is configured to, in response to receiving an input tomographic medical image describing the selected anatomical region, output a modified tomographic medical image describing the selected anatomical region of a subject; and the image processing neural network is further configured to output the modified tomographic medical image such that an anatomical abnormality depicted in the input tomographic medical image is removed; and execution of the machine-executable instructions causes a computing system to: receiving the input tomographic medical image; receiving the corrected tomographic medical image in response to inputting the input tomographic medical image into an image processing neural network; receiving an anatomical segmentation of the corrected tomographic medical image in response to inputting the corrected tomographic medical image into an image segmentation module, the anatomical segmentation dividing the corrected tomographic medical image into a plurality of anatomical regions; providing the anatomical segmentation as a segmentation of the input tomographic medical image; A computer program configured to cause the computer to execute

15. 1. A method of medical imaging, said method comprising: receiving an input tomographic medical image; receiving a modified tomographic medical image in response to inputting the input tomographic medical image into an image processing neural network, the image processing neural network being configured to output the modified tomographic medical image describing a selected anatomical region of a subject in response to receiving the input tomographic medical image describing a selected anatomical region, the image processing neural network being further configured to output the modified tomographic medical image such that an anatomical abnormality depicted in the input tomographic medical image is removed; receiving an anatomical segmentation of the corrected tomographic medical image in response to inputting the corrected tomographic medical image into an image segmentation module, the anatomical segmentation dividing the corrected tomographic medical image into a plurality of anatomical regions; providing the anatomical segmentation as a segmentation of the input tomographic medical image; A method comprising:

Images