Method and system for image annotation and segmentation

Abstract

Systems and methods for annotation and segmentation of 3D multi-volume imaging data are provided herein. In one example, a method includes obtaining three-dimensional (3D) multi-volume imaging data (302), the 3D multi-volume imaging data including a plurality of registered image volumes (504); determining one or more (306) positions of a user cursor within a first (2D) image of a first (2D) image volume of the plurality of registered image volumes; determining, for two or more (2D) of the plurality of registered image volumes, a contour point position (414) corresponding to each of one or more (2D) positions of the user cursor; generating two or more (2D) segmentation contours (424) corresponding to two or more (2D) of the plurality of registered image volumes; determining a selected segmentation profile (314) of the two or more (2D) segmentation profiles via the user input; a 3D segmentation mask (516) is generated based at least in part on the selected segmentation profile and saved to a memory.

Description

Technical Field

[0001] Embodiments of the subject matter disclosed herein relate to image processing, and more particularly to image annotation and segmentation. Background Art

[0002] Clinical decisions may be derived from the analysis of one or more sets of data. In the field of radiology, this may typically involve the analysis of regions of interest from medical image data, which may include 2D or 3D medical images, such as images of organs (kidneys, liver, spleen, etc.), blood vessels, bones, etc. Medical image analysis may be performed at the request of a referring physician to achieve a specific purpose; such purpose may include the detection, assessment and / or monitoring of the progression of anatomical abnormalities such as lesions, tumors, aneurysms, atrophy, etc.

[0003] In order to access these regions of interest of medical image data and perform the desired analysis, data corresponding to these regions may first be accurately and robustly separated from other data. A segmentation process may be employed to separate regions of interest for increased usability in diagnosis, treatment planning, and treatment monitoring. Summary of the invention

[0004] In one embodiment, a method includes obtaining three-dimensional (3D) multi-volume imaging data, the 3D multi-volume imaging data including a plurality of registered image volumes; determining one or more positions of a user cursor within a first image of a first image volume in the plurality of registered image volumes; determining, for two or more of the plurality of registered image volumes, a contour point position corresponding to each of the one or more positions of the user cursor; generating two or more segmentation contours corresponding to the two or more of the plurality of registered image volumes; determining, via user input, a selected segmentation contour of the two or more segmentation contours; generating a 3D segmentation mask based at least in part on the selected segmentation contour; and saving the 3D segmentation mask to a memory.

[0005] In another embodiment, a system includes: an imaging system configured to acquire three-dimensional (3D) multi-volume imaging data; and a computing device communicatively coupled to the imaging system, the computing device configured with instructions stored in a non-transitory memory executable by a processor, the instructions, when executed, causing the processor to: obtain a 3D multi-volume imaging dataset from an imaging archive, wherein the 3D multi-volume imaging dataset includes a plurality of registered image volumes, each of the plurality of registered image volumes including a plurality of image slices; display a first set of image slices of two or more of the plurality of registered image volumes in a graphical user interface (GUI) of a display device communicatively coupled to the computing device; The invention relates to an image slice in a slice of a first image volume; determining two or more positions of a user cursor within a first image slice of the first image volume in response to a user input; determining contour points within two or more image slices corresponding to two or more of the plurality of registered image volumes based on the two or more positions of the user cursor within the first image slice; generating two or more segmentation contours, wherein each of the two or more segmentation contours corresponds to one of the two or more image slices and includes a corresponding plurality of contour points; selecting a selected segmentation contour among the two or more segmentation contours in response to the user input; generating a 3D segmentation mask based on the selected contour; and saving the 3D segmentation mask to a non-volatile memory.

[0006] In yet another embodiment, a method includes determining a position of a user cursor within a first image of a first image volume among a plurality of image volumes of a 3D multi-volume imaging dataset; generating a geodesic propagation around the position of the user cursor within two or more images of two or more corresponding image volumes among the plurality of image volumes; determining one or more gradients within the geodesic propagation of each of the two or more images, wherein the one or more gradients are associated with an edge of a region of interest; defining the edge of the region of interest within the geodesic propagation of each of the two or more images based on the one or more gradients; determining a point of the edge that is closest to the position of the user cursor within each of the two or more images; and saving the point as one of a plurality of points of a segmentation contour for each of the two or more images.

[0007] It should be understood that the above brief description is provided to introduce in a simplified form selected concepts that are further described in the detailed description. It is not meant to identify key features or essential features of the claimed subject matter, the scope of which is uniquely defined by the claims that follow the detailed description. Furthermore, the claimed subject matter is not limited to implementations that solve any disadvantages noted above or in any part of this disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] This patent or patent 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.

[0009] The present disclosure will be better understood by reading the following description of non-limiting embodiments with reference to the accompanying drawings, in which:

[0010] Figure 1 A block diagram of an imaging system according to one embodiment of the present disclosure is shown.

[0011] Figure 2 A block diagram of an image processing system according to an embodiment of the present disclosure is shown.

[0012] Figure 3 A flow chart illustrating a method for computing segmentation contours is shown.

[0013] Figure 4 A flow chart illustrating a method for segmenting a 2D region of interest from one or more images is shown.

[0014] Figure 5 A flow chart illustrating a method for generating a 3D segmentation is shown.

[0015] Figure 6 Exemplary images of multi-parametric imaging data are shown.

[0016] Figure 7 Exemplary images of multi-phase imaging data are shown.

[0017] Figure 8 An exemplary graphical user interface (GUI) is shown with multi-parameter imaging and a user cursor displayed therein.

[0018] Fig. 9 shows a display with contour points Figure 8 GUI.

[0019] Fig.10 shows a diagram with a user cursor and outline points Figure 8 Region of interest for multi-parameter imaging.

[0020] Fig.11 shows a diagram with multiple user cursors and a contour point Figure 8 Region of interest for multi-parameter imaging.

[0021] Fig.12 Exemplary segmentation contours for each of a plurality of images of multi-phase imaging data are shown.

[0022] Fig.13A diagram of a flat area with a user cursor and the resulting contour points is shown.

[0023] Fig.14 An exemplary GUI is shown in which a user cursor is positioned in a first location within a medical image displayed in the exemplary GUI.

[0024] Fig.15 A GUI is shown with a user cursor positioned in a second location within the medical image. DETAILED DESCRIPTION

[0025] The following description relates to various embodiments of methods and systems for annotation and segmentation of three-dimensional (3D) multi-volume medical image data. Image segmentation is a subfield of image processing that seeks to group similar regions or segments of an image. Image segmentation may include dividing a digital image into different components (in other words, dividing a digital image into image segments, each of which is a collection of pixels or voxels), and may be utilized to simplify and / or change the representation of an image into something that is more meaningful and easier to analyze (e.g., by a referring physician). Image segmentation may be used for a variety of applications, including for medical imaging. During analysis of medical image data (which may be two-dimensional (2D), 3D, or in some examples, 3D data sets with multiple volumes (e.g., multi-parameter imaging data or multi-phase imaging data)), it may be useful to apply segmentation to the medical image data to allow a referring physician to more easily characterize certain anatomical features (e.g., organs, lesions, etc.) for diagnosis, analysis, etc. Segmentation is achievable on medical images captured using a variety of modalities including magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), ultrasound, and others.

[0026] Current tools for image segmentation of 3D imaging data, such as manual annotation workflow tools like "Paint on Slice", utilize user input to 2D slices of 3D imaging data, such as manually drawn contours around the region of interest, to segment the region of interest. In such examples, contours may be manually drawn onto each 2D slice of the 3D imaging data, and then an interpolation process is applied to generate a 3D segmentation from the multiple contours. Manually input contours for manual annotation workflow tools may result in a high degree of variability in annotations between contours, since only user input is used to determine contours and segmentations.

[0027] Due to the high variability of manually annotated segmentation methods, various methods have been developed to address this problem. In some examples, the closest gradient detection and / or edge detection algorithms are used in conjunction with manual input to generate contours on 2D slices of 3D imaging data. While these methods reduce the variability from manual input, variability may remain in the gradient and edge detection, which leads to inaccurate segmentation based on variability in scan acquisition, sequence parameters, etc. that affect the images and the regions of interest therein.

[0028] In addition, multi-volume imaging data, such as those generated by multi-parameter and / or multi-phase imaging methods that increase the available data of the region of interest, may require separate segmentation for each sequence or phase, although the same 3D imaging data set, wherein the data set includes multiple registered image volumes of the same field of view. Multi-parameter imaging data such as multi-parameter MRI, CT, ultrasound, etc. can be combined for multiple imaging parameters (e.g., sequences) of a set of 3D medical imaging data. For example, a multi-parameter MRI may include data of multiple sequences acquired from a specific field of view, such as a T1-weighted sequence, a T2-weighted sequence, a T1-contrast-enhanced (T1CE) sequence, a fluid-attenuated inversion recovery (FLAIR) sequence, a diffusion-weighted imaging (DWI) sequence, and the like. Multi-phase imaging data includes multiple images of target anatomical structures, typically taken at different time points as an intravenous contrast agent moves through the circulatory system. As an example, a multi-phase liver CT may include a non-contrast phase image, an arterial phase image (e.g., a late arterial phase), a portal vein phase image, and a delayed phase image, wherein each image is acquired at a different time when the contrast agent is in a specific enhancement for a specified area. For example, for the arterial phase, peak aortic attenuation with minimal liver enhancement can be seen, while for the portal venous phase, peak liver parenchyma as well as portal and hepatic venous enhancement can be seen.

[0029] In some examples, changes in the placement of anatomical structures within the field of view for different sequences / phases may occur due to periodic and / or non-periodic motion (e.g., breathing, skeletal muscle movement, etc.). Multiple volumes of a 3D imaging data set may be registered via a translational volume registration method, a rigid volume registration method, a non-rigid volume registration method, or other types of volume registration methods. Registration is the process of aligning different sets of spatial data (e.g., image volumes) by determining appropriate geometric transformations between them. Registration of multiple image volumes may allow compensation for periodic and / or non-periodic motion by adjusting position and orientation until the mutual information between the multiple image volumes is correlated.

[0030] In this manner, multi-volume imaging data including multiple image volumes of the same target anatomical structure with different parameters or at different times can provide a fourth dimension to the 3D imaging data set, whereby lesions or other regions of interest can be imaged in various ways (e.g., sequences of multi-parameter imaging) and / or at various times (e.g., phases of multi-phase imaging), thereby providing increased information and images of the region of interest. As noted, in order to segment and / or annotate the multi-volume imaging data, annotation and segmentation can be performed separately for each of the volumes, as in some examples, the lesions or regions of interest to be segmented may appear differently in different volumes, thus consuming an increased amount of time for the user and consuming increased processing power.

[0031] Therefore, embodiments of methods and systems for annotation and segmentation of 3D multi-volume medical imaging data that solve the aforementioned problems are disclosed herein. As will be explained in more detail below, embodiments disclosed herein may utilize images of 3D multi-volume (e.g., multi-parameter and / or multi-phase) imaging data, based on which a user may provide annotation input to indicate one or more regions to be included in a segmentation mask. A user cursor may be located within a first 2D slice of a first volume of the 3D multi-volume imaging data, and via determination of a gradient closest to the user cursor within a defined geodesic propagation region for each of two or more corresponding two or more image slices, a location of a point (e.g., a contour point) of a segmentation contour may be determined for each of the volumes of the multi-volume data. Multiple user cursor points for the first 2D slice of the first volume may define multiple contour points for each of the image volumes of the 3D multi-volume imaging data, and thus a segmentation contour for each volume may be defined. One or all of the generated segmentation contours may be saved for a particular slice, and a 3D segmentation mask may be generated using multiple selected segmentation contours for different slices.

[0032] Therefore, according to the embodiments disclosed herein, the above problems can be solved via a segmentation process, which includes determining a segmentation contour for each of the multiple volumes from a user cursor position within one of the multiple volumes of a 3D multi-volume imaging data set. In this way, the segmented region of interest can include data of multiple parameters or phases, thereby increasing the accuracy of annotation and segmentation. In addition, although examples involving segmentation of 3D images / volume data are presented herein, it should be understood that the segmentation process described herein can be applied to 2D images, 3D images / renderings, and other types of medical images.

[0033] Embodiments of the present disclosure will now be described by way of example with reference to the accompanying drawings. Figure 1 An exemplary imaging system that may be used to acquire 3D multi-volume imaging data is shown. Figure 2Shown is a block diagram of an image processing system that may be part of or otherwise communicatively coupled to an imaging system. Figures 3 to 5 Methods for determining a segmentation contour of multi-volume imaging data, determining points of a segmentation contour, and generating a 3D segmentation mask from a plurality of segmentation contours, respectively, are shown. Figure 6 to Figure 7 An exemplary image showing multi-volume imaging data, Figure 6 An example of multi-parameter imaging data is shown and Figure 7 An example of multi-phase imaging data is shown. Figure 8 An exemplary graphical user interface (GUI) displaying multiple images of multi-volume data to be segmented with a user cursor in a first position is shown. Fig. 9 shows a point for each volume determined from a first position of the user's cursor Figure 8 GUI. Fig.10 and Fig.11 The region of interest to be segmented using the user cursor and the corresponding point in the corresponding image volume are shown. Fig.12 A plurality of segmentation contours corresponding to a plurality of image volumes are shown. Fig.13 A diagram showing a flat region within the geodesic propagation of points used to determine a segmentation contour. Fig.14 and Fig.15 An exemplary GUI is shown with a user cursor positioned at various locations within an image.

[0034] Now turn to Figure 1 , shows a diagram of an exemplary imaging system that can be used to generate 3D multi-volume imaging data. As an example, the MRI system 10 includes a static magnetic field magnet unit 12, a gradient coil unit 13, an RF coil unit 14, an RF body or volume coil unit 15, a transmit / receive (T / R) switch 20, an RF driver unit 22, a gradient coil driver unit 23, a data acquisition unit 24, a controller unit 25, a patient bed or examination table 26, an image processing unit 31, an operation console unit 32, and a display device 33. In some embodiments, the RF coil unit 14 is a surface coil, which is a local coil that is usually placed near the anatomical structure of interest of the subject 16. Here, the RF body coil unit 15 is a transmission coil that transmits RF signals, and the local surface RF coil unit 14 receives MR signals. Therefore, the transmission body coil (e.g., the RF body coil unit 15) and the surface receiving coil (e.g., the RF coil unit 14) are independent but electromagnetically coupled components. The MRI system 10 transmits electromagnetic pulse signals to a subject 16 placed in an imaging space 18, wherein a static magnetic field is formed to perform scanning to obtain magnetic resonance signals from the subject 16. One or more images of the subject 16 may be reconstructed based on the magnetic resonance signals thus obtained by the scanning.

[0035] The static magnetic field magnet unit 12 includes, for example, an annular superconducting magnet installed in an annular vacuum container. The magnet defines a cylindrical space around the subject 16 and generates a constant main static magnetic field B 0 .

[0036] The MRI system 10 further includes a gradient coil unit 13, which forms a gradient magnetic field in an imaging space 18 so as to provide three-dimensional position information for a magnetic resonance signal received by an RF coil array. The gradient coil unit 13 includes three gradient coil systems, each of which generates a gradient magnetic field along one of three spatial axes perpendicular to each other, and generates a gradient field in each of a frequency encoding direction, a phase encoding direction, and a slice selection direction according to an imaging condition. More specifically, the gradient coil unit 13 applies a gradient field in a slice selection direction (or a scanning direction) of the subject 16 to select a slice; and the RF body coil unit 15 or the local RF coil array can transmit an RF pulse to the selected slice of the subject 16. The gradient coil unit 13 also applies a gradient field in a phase encoding direction of the subject 16 to phase encode a magnetic resonance signal from a slice excited by an RF pulse. Then the gradient coil unit 13 applies a gradient field in a frequency encoding direction of the subject 16 to frequency encode a magnetic resonance signal from a slice excited by an RF pulse.

[0037] The RF coil unit 14 is provided, for example, to surround a region to be imaged of the subject 16. In some examples, the RF coil unit 14 may be referred to as a surface coil or a receiving coil. 0 In the static magnetic field space or imaging space 18, the RF coil unit 15 transmits RF pulses as electromagnetic waves to the subject 16 based on a control signal from the controller unit 25, and thereby generates a high-frequency magnetic field B 1 This excites the proton spins in the slice to be imaged of the subject 16. The RF coil unit 14 receives the electromagnetic waves generated when the proton spins thus excited in the slice to be imaged of the subject 16 return to alignment with the initial magnetization vector as magnetic resonance signals. In some embodiments, the RF coil unit 14 may transmit RF pulses and receive MR signals. In other embodiments, the RF coil unit 14 may be used only to receive MR signals, and not to transmit RF pulses.

[0038] The RF body coil unit 15 is provided, for example, to surround the imaging space 18, and generates a main magnetic field B generated by the static magnetic field magnet unit 12 in the imaging space 18. 0Orthogonal RF magnetic field pulses are used to excite the nucleus. The RF coil unit 14 can be disconnected from the MRI system 10 and replaced with another RF coil unit, while the RF body coil unit 15 is fixedly attached and connected to the MRI system 10. In addition, although a local coil (such as the RF coil unit 14) can transmit or receive signals only to a local area of ​​the subject 16, the RF body coil unit 15 generally has a larger coverage area. For example, the RF body coil unit 15 can be used to transmit or receive signals to the whole body of the subject 16. The use of a receive-only local coil and a transmit body coil provides uniform RF excitation and good image uniformity at the expense of higher RF power deposited in the subject. For a transmit-receive local coil, the local coil provides RF excitation to the region of interest and receives MR signals, thereby reducing the RF power deposited in the subject. It should be understood that the specific use of the RF coil unit 14 and / or the RF body coil unit 15 depends on the imaging application.

[0039] The T / R switch 20 may selectively electrically connect the RF body coil unit 15 to the data acquisition unit 24 when operating in the receive mode, and may selectively electrically connect the RF body coil unit to the RF driver unit 22 when operating in the transmit mode. Similarly, the T / R switch 20 may selectively electrically connect the RF coil unit 14 to the data acquisition unit 24 when the RF coil unit 14 operates in the receive mode, and may selectively electrically connect the RF coil unit to the RF driver unit 22 when the RF coil unit operates in the transmit mode. When both the RF coil unit 14 and the RF body coil unit 15 are used for a single scan, for example, if the RF coil unit 14 is configured to receive MR signals and the RF body coil unit 15 is configured to transmit RF signals, the T / R switch 20 may direct the control signal from the RF driver unit 22 to the RF body coil unit 15 while directing the received MR signal from the RF coil unit 14 to the data acquisition unit 24. The coil of the RF body coil unit 15 may be configured to operate in a transmit-only mode or a transmit-receive mode. The coil of the local RF coil unit 14 may be configured to operate in a transmit-receive mode or a receive-only mode.

[0040] The RF driver unit 22 includes a gate modulator (not shown), an RF power amplifier (not shown), and an RF oscillator (not shown), which are used to drive an RF coil (e.g., the RF coil unit 15) and form a high-frequency magnetic field in the imaging space 18. The RF driver unit 22 modulates an RF signal received from the RF oscillator into a predetermined timing signal having a predetermined envelope based on a control signal from the controller unit 25 and using the gate modulator. The RF signal modulated by the gate modulator is amplified by the RF power amplifier and then output to the RF coil unit 15.

[0041] The gradient coil driver unit 23 drives the gradient coil unit 13 based on a control signal from the controller unit 25 and thereby generates a gradient magnetic field in the imaging space 18. The gradient coil driver unit 23 includes three driver circuits (not shown) corresponding to three gradient coil systems included in the gradient coil unit 13.

[0042] The data acquisition unit 24 includes a preamplifier (not shown), a phase detector (not shown), and an analog / digital converter (not shown) for acquiring a magnetic resonance signal received by the RF coil unit 14. In the data acquisition unit 24, the phase detector uses the output of the RF oscillator from the RF driver unit 22 as a reference signal, performs phase detection on the magnetic resonance signal received from the RF coil unit 14 and amplified by the preamplifier, and outputs the phase-detected analog magnetic resonance signal to the analog / digital converter to be converted into a digital signal. The digital signal thus obtained is output to the image processing unit 31.

[0043] The MRI apparatus 10 includes a table 26 for placing the subject 16 thereon. By moving the table 26 based on a control signal from the controller unit 25, the subject 16 can be moved inside and outside the imaging space 18.

[0044] The controller unit 25 includes a computer and a recording medium on which a program to be executed by the computer is recorded. The program, when executed by the computer, causes the various parts of the device to perform operations corresponding to a predetermined scan. The recording medium may include, for example, a ROM, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, or a non-volatile memory card. The controller unit 25 is connected to the operation console unit 32 and processes an operation signal input to the operation console unit 32, and also controls the examination bed 26, the RF driver unit 22, the gradient coil driver unit 23, and the data acquisition unit 24 by outputting control signals to them. The controller unit 25 also controls the image processing unit 31 and the display device 33 based on the operation signal received from the operation console unit 32 to obtain a desired image.

[0045] The operation console unit 32 includes a user input device such as a touch screen, a keyboard, and a mouse. The operation console unit 32 is used by the operator to input such data as an imaging protocol, and to set the area where the imaging sequence will be executed. The data on the imaging protocol and the imaging sequence execution area are output to the controller unit 25.

[0046] The image processing unit 31 includes a computing device and a recording medium on which a program to be executed by the computing device to perform predetermined data processing is recorded. The image processing unit 31 is connected to the controller unit 25, and performs data processing based on a control signal received from the controller unit 25. The image processing unit 31 is also connected to the data acquisition unit 24, and generates spectral data by applying various image processing operations to the magnetic resonance signals output from the data acquisition unit 24.

[0047] The display device 33 may display one or more images within the GUI on a display screen of the display device based on a control signal received from the controller unit 25. The display device 33 displays, for example, images regarding input items for inputting operation data by an operator from the operation console unit 32. The display device 33 also displays a 2D slice image or a 3D image of the subject 16 generated by the image processing unit 31.

[0048] The MRI system 10 can be configured for multi-volume imaging, such as multi-parameter and / or multi-phase imaging, in which multiple imaging sequences and / or phases are imaged during a single imaging session. The resulting MRI imaging data may include images from each of the imaged sequences and / or phases, wherein the MRI imaging data is subdivided into specified sequences and / or phases. Each of the specified sequences and / or phases may define multiple 2D slices thereof, each of which is specific to the z coordinate of the MRI imaging data. The z coordinate may therefore define multiple 2D slices, each of which is from each of the specified sequences and / or phases.

[0049] Although an MRI system is described by way of example, it should be understood that the present techniques may be applied to images acquired using other imaging systems capable of multi-parameter, multi-phase, or other types of multi-volume imaging, such as CT, tomosynthesis, PET, ultrasound, etc. The present discussion of the MRI imaging modality is provided merely as an example of one suitable imaging modality.

[0050] refer to Figure 2, an image processing system 200 is shown. In some embodiments, the image processing system 200 can be incorporated into an imaging system such as an MRI system 10. For example, the image processing system 200 can be the image processing unit 31 of the MRI system 10. However, in other embodiments, the image processing system 200 can be disposed at a device (e.g., a server, an edge device, etc.) that can be communicatively coupled to the imaging system via a wired and / or wireless connection. In some examples, at least a portion of the image processing system 200 can be disposed at a separate device (e.g., a workstation) that can receive images from the imaging system or from a storage device that stores images generated by the imaging system and / or other additional imaging systems. The image processing system 200 can be communicatively coupled to a user input device 234, a display device 236, and an imaging archive 238. In some examples, the user input device 234 can be included in Figure 1 The operator console unit 32 and the display device 236 may be Figure 1 Display device 33.

[0051] The image processing system 200 may include a processor 202 configured to execute machine-readable instructions stored in a non-transitory memory 204. The processor 202 may be a single-core or multi-core processor, and the program executed thereon may be configured for parallel processing or distributed processing. In some embodiments, the processor 202 may optionally include separate components spread across two or more devices, which may be remotely located and / or configured for collaborative processing. In some examples, one or more aspects of the processor 202 may be virtualized and executed by a remotely accessible networked computing device configured in a cloud computing configuration.

[0052] In addition to images provided directly by the image processing system 200, images may also originate from an imaging archive 238 that is communicatively coupled to the image processing system 200. The imaging archive 238 may include, for example, a picture archiving and communication system (PACS), a vendor neutral archive (VNA), or other suitable medical image database. The medical imaging archive may be hosted on a remote server that is configured to allow the image processing system 200 to access a plurality of medical images and patient data hosted thereon. In some examples, the plurality of medical images stored in the imaging archive 238 may be of different types, such as MRI images, CT images, ultrasound images, and more medical images may all be stored in the imaging archive 238 for one or more patients.

[0053] The non-transitory memory 204 can store a segmentation module 206, a flat region characterization module 208, and a contour point location module 210. The segmentation module 206 can obtain or otherwise determine one or more contours for a multi-volume imaging data set, each of the one or more contours corresponding to a specified 2D slice of the imaging data set. The segmentation module 206 can also interpolate between the one or more contours to define a 3D segmentation mask for a region of interest.

[0054] The flat region characterization module 208 may analyze one or more images (e.g., 2D slices of 3D data, thick slices of 3D data, renderings of 2D projections, etc.) for pixel intensity, gradients, tonal distribution, etc. The flat region characterization module 208 may determine geodesic propagation around the user cursor location and determine pixel intensities within the geodesic propagation region. An image gradient may be a change in intensity in an image, and a highest gradient may be where the largest change occurs between pixels. In some examples where the gradient is sharp, the highest gradient may indicate an edge of a region of interest. In other examples where the gradient is gradual or less sharp, one or more gradients associated with an edge of a region of interest may not be the highest gradient within the geodesic propagation. The flat region characterization module 208 may therefore store instructions to determine a gradient associated with an edge of a region of interest within the geodesic propagation for each displayed image for a given location of a user cursor, whether the gradient is the highest gradient or another gradient. If one or more gradients are detected within the geodesic propagation region, a designated region approximating an edge may be determined based on the detected gradients. If no gradient is detected, the region may be considered a flat region, where all pixels (or voxels) in the flat region have the same or relatively the same (e.g., ± a specified noise) pixel intensity. If a flat region is determined, the region may be circular (for 2D slices) or spherical (for 3D data). When a gradient is detected or otherwise determined, the region may not be circular, as will be described with respect to Fig.13 Flat region characterization and gradient determination may be performed by the flat region characterization module 208 for each of the multiple volumes of the 3D multi-volume imaging dataset based on a user cursor position provided to an image slice of a single image volume.

[0055] The contour point location module 210 may obtain data of flat area characterizations and gradients, and store instructions for determining the location of a point (also referred to herein as a "contour point") corresponding to the user cursor position for each of the plurality of volumes. In some examples, the point may be one of a plurality of points of a segmentation contour corresponding to the respective image volume. The determined point may be a point where the edge of a specified area around the user cursor is closest to the position of the user cursor. In some examples, the closest point may be one of the determined associated gradients. In addition, the contour point location module 210 may include data of previously determined contour points to stabilize subsequent contour points of a particular segmentation contour.

[0056] In some embodiments, non-transitory memory 204 may include components disposed at two or more devices that may be remotely located and / or configured for collaborative processing. In some embodiments, one or more aspects of non-transitory memory 204 may include a remotely accessible networked storage device configured in a cloud computing configuration.

[0057] Now turn to Figure 3 , a flow chart illustrating a method 300 for determining and displaying a plurality of segmentation contours is shown. Figure 1 to Figure 2 The method 300 is described with reference to the system depicted in the present disclosure, but it should be understood that similar methods can be used with other systems without departing from the scope of the present disclosure. The method 300 can be executed via instructions stored in a non-transitory memory of one or more computing devices. In some examples, the method 300 can be executed by a computing device (e.g., including Figure 2 The computing device may be implemented by a processor 202 of the image processing system 200 and a group of remote servers (also referred to herein as a cloud computing system) in communication with the computing device. The computing device may communicate with an imaging archive (eg, imaging archive 238).

[0058] At 302, method 300 includes obtaining a plurality of images of a 3D multi-volume imaging dataset from an imaging archive. The imaging system (e.g., Figure 1 The MRI system 10 of the present invention acquires a multi-volume imaging data set, and the imaging system is configured to acquire multi-volume data, such as multi-parameter imaging data and / or multi-phase imaging data. In some examples, the multi-volume imaging data set to be segmented can be selected via user input. For example, the user can select the desired multi-volume imaging data set to be segmented, and the segmentation of the selected 3D multi-volume imaging data set can be requested via user input (e.g., by selecting a segmentation label, tool, etc.). The 3D multi-volume imaging data set may include multiple registered image volumes, each image volume from a different imaging sequence or phase taken from the same field of view and registered with each other as described above. In some examples, each of the multiple registered image volumes may be divided into multiple 2D slices. Each of the 2D slices may be specific to a specific coordinate according to the axis from which they are obtained (e.g., the z coordinate of the axial view). In some examples, multiple 2D slices for each of the multiple registered image volumes may have the same set of coordinates. In other examples, due to sorting parameters, acquisition parameters, etc., a set of coordinates for multiple 2D slices of one or more of the multiple registered image volumes may be different from other multiple 2D slices. The multiple images obtained may be the entire imaging volume.

[0059] At 304, method 300 includes displaying image slices of one or more of the plurality of images in a GUI. In some examples, the displayed image slices may be 2D slices. In other examples, the displayed image slices are renderings of 2D projections, such as maximum intensity projections (MIPs), minimum intensity projections (MinIPs), mean intensity projections (meanIPs), etc., where pixels of 3D data are projected onto a 2D plane based on pixel intensities (maximum intensity, minimum intensity, mean intensity, etc.). The GUI may be displayed on a display device (e.g., Figure 2 The GUI may be displayed on a display device 236 of the image and may include one or more selectable elements and / or tools by which a user can interact with the image, annotate the image, or perform other actions. For example, the GUI may include a segmentation element that, when selected via user input, launches a segmentation tool menu. The 3D multi-volume imaging dataset may include images obtained from one or more orientations, and thus the images may be displayed in one of various views (e.g., axial view, sagittal view, coronal view, etc.). In some examples, each of the one or more images displayed may be displayed from the same view.

[0060] In addition, the plurality of images may additionally include images that are not displayed in the GUI as well as those one or more images that are displayed. In some examples, the GUI may include a menu through which a user may select which of the plurality of images are displayed. For example, a 3D multi-parameter imaging data set may include volumes from a T1 sequence, a T1CE sequence, a FLAIR sequence, a DWI sequence, and a T2-weighted sequence. A user may select, for example, a T1 sequence, a FLAIR sequence, and a T2-weighted sequence to be displayed in the GUI. DWI sequences and T1CE sequences may be stored in transient and / or non-transient memory and may be displayed in the GUI when a user selects via a menu. In other examples, each of the plurality of images of the 3D multi-volume imaging data set may be displayed in the GUI.

[0061] At 306, method 300 includes receiving user input of a position of a user cursor within a first image slice of a displayed image slice corresponding to one or more images. The first image slice may have a first image volume. The user cursor may be adjustable via user input (e.g., movement of a user input device such as a mouse), and the position of the user cursor may be determined via user input (e.g., a mouse click). The position of the user cursor may be determined relative to the displayed image / 2D slice. In some examples, a coordinate system may be predefined for a 3D multi-volume imaging dataset and its volume, and thus, the coordinate system may be the same for each of the volumes. The position of the user cursor within the first image slice may therefore be known, and because the coordinate system is the same for each of the volumes, the position of the user cursor may be transferred to another of the multiple images. In some examples, more than one position of the user cursor may be determined, for example, when the user moves the user cursor from an initial position around a region of interest. One or more positions may be defined along a line, the first position being the position of the user cursor when an initial user input (e.g., a mouse click) is received, and one or more second positions may be defined (e.g., by dragging in the case of a clicked mouse) until the initial user input is released (e.g., the mouse click is released). The user input defining the position of the user cursor may be an annotation input.

[0062] At 308, method 300 includes calculating a segmentation contour for each of the displayed image slices based on the position of the user cursor. Figure 4 Described in more detail, for each position of the user cursor, the position of a point (e.g., a contour point position) may be determined simultaneously within each of the displayed image slices. The calculated point may be one of a plurality of points of a segmentation contour. Thus, for a given set of user cursor positions input to a first image slice, a segmentation contour may be defined for each of a plurality of images of a 3D multi-volume imaging dataset.

[0063] Determining contour points from the user cursor position may be performed in real time or near real time (e.g., without intentional delay), whereby when an annotation input defining the user cursor position is received, contour points are added to the corresponding image slice in real time. In this manner, a user may be able to effectively visualize contour points as they are determined. Additionally, processing efficiency may be improved by the real-time nature of the annotation. For example, since a single annotation input may generate multiple contour points in real time, processing time and power may be reduced because repeated annotations are mitigated and determination of contour points may be performed based on a single acquisition of the position of the user cursor. In this manner, annotation and segmentation of multiple volumes of data may be performed more efficiently because determination of contour points and generation of corresponding segmentation contours need not be performed based on separate annotation inputs for each corresponding volume.

[0064] In this way, annotation input that defines the user cursor position to a single image slice can produce contour points for output of multiple images (e.g., displayed 2D slices), thereby reducing the time spent by the user in annotating images and processing power as the number of user inputs is reduced. In addition, calculating the segmentation contours for each of multiple images of the same region of interest increases the amount of data available at the time of annotation and segmentation, thereby increasing the accuracy of the resulting segmentation, as the region of interest can be visualized and analyzed from various sequences that differently show the internal anatomical structure.

[0065] At 310, method 300 includes displaying the corresponding segmentation contours as an overlay on the corresponding image slices. Each of the segmentation contours calculated at 308 may be different from each other because each position of the user cursor corresponds to an independently calculated contour point position in each of the image slices. As an example, a first segmentation contour calculated for a first 2D slice may be displayed as an overlay on a displayed first image slice in a GUI. A second segmentation contour calculated for a second 2D slice may be displayed as an overlay on a displayed second image slice in a GUI. In some examples, the first segmentation contour may be displayed in a first color, and the second segmentation contour may be displayed in a different second color. In other examples, the first segmentation contour and the second segmentation contour may be displayed in the GUI in the same color.

[0066] In some examples, each of the corresponding segmentation contours may be editable. For example, via user input such as a mouse click, the position of a point of the segmentation contour may be changed without having to repeat the calculation of the point, thereby reducing processing power by avoiding repeated calculations.

[0067] At 312, method 300 optionally includes displaying all of the calculated contours as an overlay on each of the image slices. In some examples, all of the calculated contours may be displayed as an overlay on each of the image slices so that a user can visualize the contours in comparison to each of the one or more images. In this manner, a user may be able to determine which of the calculated contours best approximates the shape of the region of interest ("best" is a subjective determination by the user when comparing the segmented contours). As noted above, each of the displayed image slices may belong to a set of image slices for particular coordinates, and the selected contour may be defined relative to the set of image slices for the particular coordinates. In this manner, multiple contours may be selected, each for a different coordinate / group of image slices, such as by selecting a contour for each of the image slices. Figure 5 Described.

[0068] At 314, method 300 includes receiving a user input selecting one of the superimposed contours. In some examples, each of the superimposed contours may be assigned a number to allow for differentiation thereof. Each of the superimposed contours may be a selectable element that, when selected via user input such as a mouse click or keyboard touch (e.g., a keyboard touch corresponding to the assigned number), indicates that the selected segmentation contour for one or more displayed images (e.g., of a particular set of image slices displayed) will be saved. Alternatively or additionally, all segmentation contours may be selected.

[0069] At 316, method 300 includes saving the selected segmentation contour to a memory. In response to the user selection of the selected segmentation contour, the selected segmentation contour may be saved to the memory for access and / or use at a later date. For example, the selected segmentation contour may be used to interpolate a 3D segmentation mask, such as Figure 5 As further described. In this way, a 3D segmentation mask may be generated at least partially on the selected segmentation contour. The steps of method 300, such as from 304 to 316, may be performed for two or more groups of image slices of the volume, so that a plurality of segmentation contours are defined for each volume (e.g., each volume at different coordinates) and the selected segmentation contour may be selected for each corresponding coordinate. The plurality of selected contours may then be used to form a segmentation mask.

[0070] Now turn to Figure 4 , a flow chart illustrating an exemplary method 400 for determining contour point locations and calculating segmentation contours is shown. Figure 1 to Figure 2 The method 400 is described with reference to the system depicted, but it should be understood that similar methods can be used with other systems without departing from the scope of the present disclosure. The method 400 can be executed via instructions stored in a non-transitory memory of one or more computing devices. In some examples, the method 400 can be executed by a computing device (e.g., including Figure 2 The method 400 may be implemented as a computing device (e.g., a processor 202 of the image processing system 200) and a set of remote servers (also referred to herein as a cloud computing system) in communication with the computing device. The computing device may communicate with an imaging archive (e.g., the imaging archive 238). Figure 3 The exemplary method implemented as a portion of method 300 (specifically at 308) of FIG. It should be understood that other methods of calculating the segmented contours and determining the contour point locations are possible without departing from the scope of the present disclosure.

[0071] At 402, method 400 includes obtaining a plurality of images of a 3D multi-volume imaging dataset from an imaging archive. The imaging system (e.g., Figure 1A multi-volume imaging data set is acquired by an MRI system 10 configured to acquire multi-volume data, such as multi-parameter imaging data and / or multi-phase imaging data. In some examples, the multi-volume imaging data set to be segmented may be selected via user input. For example, a user may select a desired multi-volume imaging data set to be segmented, and may request segmentation of the selected 3D multi-volume imaging data set via user input (e.g., by selecting a segmentation label, tool, etc.). The 3D multi-volume imaging data set may include multiple registered image volumes, each image volume from a different imaging sequence or phase of the same field of view. As described above, multiple image volumes may be registered with each other. In some examples, each of the multiple registered image volumes may be divided into multiple image slices. In some examples, multiple image slices for each of the multiple registered image volumes may have the same set of image slice coordinates (e.g., slice positions for a given axis). The multiple images obtained may be multiple registered image volumes.

[0072] At 404, method 400 includes displaying image slices of one or more of the plurality of registered image volumes in a GUI. The GUI may be displayed on a display device (e.g., Figure 2 The GUI may be displayed on a display device 236 of the image and may include one or more selectable elements and / or tools by which a user may interact with the image, annotate the image, or perform other actions. For example, the GUI may include a segmentation element that, when selected via user input, launches a segmentation tool menu. The 3D multi-volume imaging dataset may include images obtained from one or more orientations (e.g., axes), and thus the images may be displayed in one of a variety of views (e.g., axial views, sagittal views, coronal views, etc.). In some examples, each of the one or more images displayed may be displayed from the same view.

[0073] In addition, as regards Figure 3 As described, the plurality of images may include images that are not displayed within the GUI. In some examples, the GUI may include a menu through which a user may select which of the plurality of images to display. In other examples, each of the plurality of images of the 3D multi-volume imaging dataset may be displayed within the GUI.

[0074] At 406, method 400 includes receiving a user input of a position of a user cursor within a first image slice in the displayed image slices. The first image slice may correspond to a first imaging volume. The user cursor may be adjustable via user input (e.g., movement of a user input device such as a mouse), and the position of the user cursor may be selected via user input. The position of the user cursor may be determined relative to the displayed image slice. In some examples, a coordinate system may be predefined for the 3D multi-volume imaging dataset and its volumes, and therefore, the coordinate system may be the same for each of the volumes. The position of the user cursor within the first image slice may therefore be known, and because the coordinate system is the same for each of the volumes, the position of the user cursor may be transferred to another of the multiple images.

[0075] At 408, method 400 includes defining a geodesic propagation around the position of the user cursor for each of two or more image slices corresponding to the two or more image volumes. The geodesic propagation may be circular, where the radius of the circle depends on the tonal distribution within the propagation region, as indicated at 410. For example, a larger propagation region may be defined for a smaller tonal distribution range (e.g., where the pixel intensities are relatively the same), and a smaller propagation region may be defined for a larger tonal distribution range. The geodesic propagation may define an area of ​​influence of the user cursor, whereby pixels within the propagation region are considered for gradients, edges, and / or contour points.

[0076] At 412, method 400 includes determining a gradient within a geodesic propagation region for each respective image slice. As noted above, the geodesic propagation region around the location of the user cursor may define an area of ​​influence. The determined gradient may be a change in pixel intensity between adjacent pixels determined to be associated with an edge of the region of interest, such as determined by processor 202 based on information stored in Figure 2 The flat region characterization module 208 may be determined by instructions in the flat region characterization module 208. In some examples, when all pixel intensities of a region within the geodesic propagation are the same or relatively the same (e.g., ± a specified amount of noise) and no gradients are detected, the region may be considered a flat region. If one or more gradients are detected, the edges of the region of interest may be detected. Edge detection may define a specified region around the user cursor. The specified region may combine the geodesic propagation region with the detected edges, and the specified region may therefore not be a circle.

[0077] At 414, method 400 includes determining a contour point based on the position of the user cursor and the determined gradient points for each corresponding image slice. In some examples, a roundness criterion may be applied, as indicated at 416. The roundness criterion provides an indication of how round the contour to which the contour point belongs is, wherein the roundness value ranges from 0 to 1, wherein values ​​closer to 0 are less round and values ​​closer to 1 are more round. In some examples, the contour point position may be the closest edge position relative to the position of the user cursor, as indicated at 418. In this way, the contour point may be located along a ray of the user cursor. Thus, if two user cursors are located along the same ray, the same contour point may be proposed for both of the two user cursor positions for a given image slice.

[0078] In some examples, the contour point or the position of the given image slice may be stabilized based on the previous contour point position of the given image slice, as indicated at 420. In the case where the contour point is the first contour point of the segmented contour, stabilization may not be performed. In the case where the contour point is not the first contour point, stabilization may be performed. The method for stabilization may include a low pass filter, where the contour point is equal to the previous contour point plus a defined displacement, the defined displacement being the difference between the contour point and the previous contour point.

[0079] At 422, method 400 includes determining whether additional contour points are needed to generate the corresponding segmentation contour. Each segmentation contour may be generated based on multiple contour point locations. Figure 3 As described, a user input such as a left mouse click and drag may indicate an initial user cursor position and one or more subsequent user cursor positions. A second user input such as a release of a mouse click may indicate that no additional contour points are to be added to the segmented contour (e.g., the segmented contour is complete). In some examples, the segmented contour may be considered complete when the final contour points approximate the initial contour points.

[0080] If additional contour points are to be added to generate the segmentation contour, method 400 returns to 406 to receive user input of the position of the user cursor within the first image slice. As previously described, the user input may be a mouse click or a drag of the mouse. When the determined contour points are different from the previously determined contour points, new contour points may be added as the cursor is moved. If no additional contour points are to be added, method 400 proceeds to 424.

[0081] When the annotation input is added to the first image slice, the contour points can be added in real time (e.g., without intentional delay). In the GUI, this can be demonstrated by adding the contour points at or near the same time as the corresponding annotation input is entered. For example, as the user drags the user cursor, the contour points within each corresponding image slice can be added and displayed for viewing. In this way, efficiency and accuracy can be improved because real-time interpretation can be done by the user. In addition, due to the real-time nature of the annotation, processing can be more efficient.

[0082] At 424, method 400 includes generating a segmentation contour based on the plurality of contour points. In some examples, interpolation may be performed between adjacent contour points to generate contour lines therebetween. For example, a plurality of contour points that approximate a circle may generate a circular segmentation contour via interpolation between the points. When the contour points approximate the edge of the region of interest, the segmentation contour may approximate the shape of the region of interest. After 424, method 400 ends.

[0083] Now turn to Figure 5 , a flow chart illustrating an exemplary method 500 for generating a 3D segmentation mask from a plurality of 2D segmentation contours is shown. Figure 1 to Figure 2 The method 500 is described with reference to the system depicted, but it should be understood that similar methods can be used with other systems without departing from the scope of the present disclosure. The method 500 can be executed via instructions stored in a non-transitory memory of one or more computing devices. In some examples, the method 500 can be executed by a computing device (e.g., including Figure 2 The method 500 may be implemented by a computing device (e.g., a processor 202 of the image processing system 200) and a set of remote servers (also referred to herein as a cloud computing system) in communication with the computing device. The computing device may communicate with an imaging archive (e.g., the imaging archive 238). The method 500 may be performed in conjunction with the methods 300 and 400.

[0084] At 502, method 500 includes obtaining a sample from an imaging archive (e.g., Figure 2 The imaging archive 238) obtains multiple images of the 3D multi-volume imaging data set. Figure 3 As described, each of the plurality of images may be a 3D volumetric image (e.g., an image volume) and may include a plurality of image slices, as indicated at 504. As previously described, the plurality of images may include 3D volumetric images of different sequences, phases, and / or parameters. Furthermore, as also described with respect to method 300, in some examples, the 3D multi-volume imaging dataset to be segmented may be selected via user input.

[0085] At 506, method 500 includes generating, for a selected first set of image slices, a plurality of first segmentation contours, one contour in the plurality of first segmentation contours corresponding to each of the first set of image slices. Each of the image slices in the first set of image slices may correspond to a respective image volume of the plurality of images. The plurality of first segmentation contours may correspond to a region of interest imaged by the imaging system. The selected first set of image slices may be selected via user input, such as scrolling through the plurality of image slices to select one of the plurality of 2D slices. Figure 3 and Figure 4 As described, a user cursor position may be determined within one of a plurality of images and may be used to calculate corresponding contour point positions for each of a plurality of images of a selected first set of image slices. For example, calculating the corresponding contour point positions may include definition of geodesic propagation, determination of one or more gradients, wherein the one or more gradients are associated with an edge of a region of interest, and determination of a point at the gradient edge that is closest to the user cursor position. A plurality of user cursor positions input to one of the first set of image slices and a corresponding plurality of contour points may define a first segmentation contour for each of the selected first set of image slices. As previously described, each of the plurality of segmentation contours may be different from one another because the calculation of the contour point positions is performed independently for each of the image slices.

[0086] At 508, method 500 includes receiving user input selecting a first selected contour of the plurality of first segmented contours. Figure 3 As described, a user input such as a mouse click or a keyboard touch may select a first selected contour for a selected first set of image slices. The first selected contour may be saved to a memory for the corresponding selected first set of image slices.

[0087] At 510, method 500 includes generating a corresponding plurality of contours for one or more selected second sets of image slices. In some examples, image slices in the one or more selected second sets of image slices may correspond to the same image volume as the first set of image slices. One or more selected second sets of image slices may be selected similarly to the selected first set of image slices. For example, after selecting the first selected contour, a user input (such as scrolling through the plurality of image slices) may select the second set of image slices. The user cursor position may be determined within one of the image slices of each of the one or more second sets of image slices and may be used to calculate a corresponding contour point position for each of the image slices of the corresponding second set of image slices. The calculation of the contour point positions and the corresponding segmentation contours may be as described at 504. For each of the one or more selected second sets of image slices, a plurality of second contours may be generated. Each of the plurality of second contours may be specific to one of the plurality of images of the 3D multi-volume imaging data set.

[0088] Similar to About Figure 3 As described, each of the calculated contours may be displayed as an overlay on the corresponding image slice. In addition, each of the calculated contours and the displayed contours may be adjustable to change the position of one or more contour points if the user desires so. For example, via user input, the points of a specified contour may be adjusted from an initial position to a second position, thereby changing the specified contour.

[0089] At 512, method 500 includes receiving a user input selecting a profile for each of a plurality of second profiles to define a selected profile for each of one or more second groups of image slices. The selection of the first and one or more selected second profiles may define a set of selected profiles for the 3D multi-volume imaging dataset. In some examples, the set of selected profiles may include a selected profile for each group of image slices of the 3D multi-volume imaging dataset. In other examples, the set of selected profiles may include a selected profile for each of a subset of a group of image slices of the 3D multi-volume imaging dataset.

[0090] At 514, method 500 includes interpolating between selected contours to generate a 3D segmentation mask. As noted, each of the selected contours may be 2D. In some examples, each of the selected contours may have known coordinates, for example, in an axial view, each slice may have a known z coordinate. Multiple selected contours may include a stack of 2D contours, each 2D contour having different coordinates along a particular axis. In this context, interpolation may include estimating point locations between the selected contours to define a 3D shape from the multiple 2D contours. The defined 3D shape may approximate the region of interest to be segmented.

[0091] At 516, method 500 includes generating a 3D segmentation mask based on the interpolation. The 3D shape defined by the interpolation can define a 3D segmentation mask of the region of interest. As an example of a method for generating a 3D segmentation mask, the data points of the interpolated 2D segmentation contour can be projected or mapped into a 3D space in which the 3D multi-volume imaging data set is initially obtained. The resulting 3D segmentation mask can define pixels included in the region of interest and pixels excluded from the region of interest. The 3D segmentation mask can be saved to a memory of a computing device and can be accessible for later use. For example, for diagnosis, evaluation, or other purposes, other users can access the 3D segmentation mask to view the 3D segmentation mask. As described herein, the 3D segmentation mask can combine data from multiple volume images of the region of interest, and thus can provide a 3D segmentation mask with a more accurate representation of the region of interest.

[0092] Therefore, the processing requirements for segmentation can be reduced by allowing segmentation and difference of 2D slices to a 3D mask rather than segmenting the 3D data independently. The segmentation of 3D data itself is inefficient in terms of time and processing power. Segmenting one or more 2D image slices and then difference back to 3D space can be more time efficient and can reduce processing power.

[0093] In some examples, a representation of the 3D segmentation mask can be displayed on a display device, as an overlay on one or more images, or in a separate window. The representation of the 3D segmentation mask can show the shape of the region of interest, and in some examples can be interactive, such that a user can change the orientation, rotate, move, etc. of the representation via user input to visualize different angles.

[0094] Now go to Figure 6 , shows an exemplary image of a first multi-volume imaging dataset of a patient. The first multi-volume imaging dataset may be a multi-parameter imaging dataset comprising a plurality of registered image volumes of various imaging sequences acquired from the same field of view. For example, the first multi-volume imaging dataset may be a multi-parameter MRI scan of the brain, such as Figure 6 shown.

[0095] The first multi-volume imaging data set may include a first image 602 of the first volume, a second image 604 of the second volume, and a third image 606 of the third volume. The first volume, the second volume, and the third volume may include multiple images or 2D slices thereof. The first image 602, the second image 604, and the third image 606 may include data of the same anatomical structure of the patient. For example, a region of interest 610, which may be a lesion, a tumor, an organ region, or other feature, may be present in each of the first volume, the second volume, and the third volume. In some examples, the first volume may be a T1 sequence, the second volume may be a T1CE sequence, and the third volume may be a FLAIR sequence. Additional sequences may be included in the first multi-volume imaging data set not shown here. Each of the sequences of the first multi-volume imaging data set may be acquired by the imaging system during a single scan acquisition. Therefore, the region of interest 610 may be imaged by each of the sequences during the same time period.

[0096] Each of the sequences of the first multi-volume imaging data set can focus, highlight, or otherwise show a different portion of the imaged anatomical structure. For example, a T1CE sequence can show more enhanced regions than a T1 sequence. In this way, the region of interest 610 can be viewed using multiple parameters, thereby providing increased information about the region of interest that can be used to more accurately segment the region of interest from surrounding anatomical structures.

[0097] Figure 7Similarly, examples of images of a second multi-volume imaging dataset are shown, including a first image 702 of a first volume, a second image 704 of a second volume, a third image 706 of a third volume, and a fourth image 708 of a fourth volume. In some examples, the second multi-volume imaging dataset may be a multi-phase imaging dataset, such as a multi-phase CT scan of a liver, e.g. Figure 7 shown.

[0098] Similar to the first multi-volume imaging dataset, the volumes of the second multi-volume imaging dataset may include multiple images, and the first image 702, the second image 704, the third image 706, and the fourth image 708 may each be one of the corresponding multiple images. For example, the first volume may be a first phase, such as a non-contrast phase, the second volume may be a second phase, such as an arterial phase, the third volume may be a third phase, such as a portal vein phase, and the fourth volume may be a fourth phase, such as a delayed phase. In some examples, these phases may be obtained at different times during scan acquisition as the contrast agent travels through the vascular system. Also similar to the first multi-volume imaging dataset, each of the volumes of the second multi-volume imaging dataset may include data for the same region of interest, such as the region of interest 710. Due to the different phases, the region of interest 710 may appear differently in each of the volumes. In this way, additional information about the region of interest 710 may be provided, and thus the segmentation of the region of interest 710 may be more accurate.

[0099] Images of both the first multi-volume imaging dataset and the second multi-volume imaging dataset may be acquired by the imaging system, stored in an imaging archive, and displayed within the GUI. Figures 3 to 5 The described method segments a region of interest such as the regions of interest 610 and / or 710 .

[0100] Now turn to Figure 8 , shows an exemplary GUI 800. The GUI 800 may be displayed on a computer associated with an image processing system such as Figure 2 A display device (such as an image processing system 200) that communicates with Figure 2 GUI 800 may display one or more images of a multi-volume imaging dataset of a patient acquired by an imaging system and stored in an imaging archive (such as imaging archive 238) in communication with an image processing system.

[0101] As noted, the GUI 800 may display one or more images of a multi-volume imaging dataset. For example, the GUI 800 may display a first image 802, a second image 804, and a third image 806. In some examples, the first image 802, the second image 804, and the third image 806 may each belong to a different volume of the multi-volume imaging dataset. For example, the different volumes may be from different sequences or from different phases. As previously explained, a region of interest 810 may be imaged in each of the volumes. In some examples, the GUI 800 may display one image slice (e.g., a 2D slice) of the respective image volume at a time.

[0102] In some examples, the GUI 800 may include a menu 820 of available sequences or phases for a multi-volume imaging dataset. Each of the available sequences may be displayed as a selectable element in a list, such as sequence 822. Two or more of the listed sequences may be selected at a time. When selected, images of the volumes corresponding to the selected two or more sequences (e.g., slices of the corresponding image volumes) may be displayed within the GUI 800.

[0103] The GUI 800 may also include a plurality of tool icons 824 that, when selected via user input, activate corresponding tools, such as a ruler tool, a brightness adjustment tool, a segmentation tool, etc., through which a user can interact with one or more images. The user cursor 812 may be moved via user input utilizing a user input device (such as a mouse) in communication with a display device. When the segmentation tool is selected, a first type of user input (such as a mouse click) may indicate a position of the user cursor 812, as described with respect to method 400. In some examples, the position of the user cursor 812 may determine a position of a contour point (e.g., a point of a segmentation contour) within each of the one or more displayed images, as described with respect to Figure 4 In other examples, the position of the user cursor 812 may determine the position of a contour point within each of the images of the multi-volume imaging dataset, even those images that are not displayed within the GUI 800 at the time of the user input.

[0104] Now go to Fig. 9 , showing the position of multiple contour points after determining Figure 8800 of the present invention. Each of the plurality of contour points may be displayed in a corresponding position as an overlay within the respective image. For example, the first contour point 900 may be displayed within the first image 802, the second contour point 902 may be displayed within the second image 804, and the third contour point 904 may be displayed within the third image 806. Each of the first contour point 900, the second contour point 902, and the third contour point 904 may be displayed simultaneously. The positions of the first contour point 900, the second contour point 902, and the third contour point 904 may be independently determined based on the position of the user cursor 812 and the flat area characterization, tonal distribution, and gradient of the corresponding image, as described with respect to the method 400. In some examples, the positions of additional contour points of the image that are not displayed within the GUI 800 when the user input is received may also be determined based on the position of the user cursor 812.

[0105] Now go to Fig.10 and Fig.11 , showing an enlarged view 1000 of a portion of the first image 802 of the GUI 800 and an enlarged view 1002 of a portion of the second image 804 of the GUI 800. As described in detail with respect to the method 400, the user cursor 812 may be positioned within a specified distance of an edge of the region of interest 810 so that when geodesic propagation is generated and gradients therein are determined, gradients may be detected within the propagation region. A ray may be defined to determine the closest point of the edge of the region of interest 810 to the user cursor 812. The determined edge of the region of interest 810 may be based on the tonal distribution and the detected gradient, and thus may be different for each of the first image 802 and the second image 804. Thus, the positions of the first contour point 900 and the second contour point 902 may be different.

[0106] Due to the rays defined during the determination of the contour points, the position of the user cursor within the same ray may generate a single contour point in each image. Fig.11 As shown, the first user cursor position 1104 may generate a first contour point 1108. The second user cursor position 1106 may also generate the same first contour point 1108 as the first user cursor position 1104 when it is in the same ray relative to the edge of the region of interest 810.

[0107] Now turn to Fig.12 , showing an example of multiple segmentation contours. In some examples, the multiple segmentation contours can be displayed as overlays within corresponding images displayed within the GUI. In other examples, the multiple segmentation contours can be displayed as overlays within each of the multiple images displayed within the GUI.

[0108] As an example, the GUI may display a first image 1200, a second image 1202, a third image 1204, and a fourth image 1206. Each of the first image 1200, the second image 1202, the third image 1204, and the fourth image 1206 may belong to a different image volume of the multi-volume imaging dataset. In some examples, each of the first image 1200, the second image 1202, the third image 1204, and the fourth image 1206 may include a region of interest 1250. Each of the plurality of segmentation contours may approximate the shape of the region of interest 1250. The plurality of segmentation contours may include a first segmentation contour 1210 corresponding to the first image 1200, a second segmentation contour 1212 corresponding to the second image 1202, a third segmentation contour 1214 corresponding to the third image 1204, and a fourth segmentation contour 1216 corresponding to the fourth image 1206.

[0109] like Fig.12 As shown, multiple segmentation contours can be displayed as overlays within the corresponding image in various stages of completion. For example, in the first row 1230, multiple segmentation contours are displayed in an unfinished state (such as when the user is actively annotating). For example, when user input is added to repeatedly add points to the segmentation contours, the multiple segmentation contours can be dynamically updated, such as with respect to Figure 4 When displayed in an unfinished state, each of the plurality of segmented contours may include a current contour point, such as the current contour point 1240 of the first segmented contour 1210. As shown in the second row 1232, once the annotation is complete, the plurality of segmented contours may be displayed. In some examples, completion of the annotation may be reached when the current contour point is equal to or approximates the initial contour point.

[0110] In some examples, such as regarding Figures 3 to 5 As described, each of the plurality of segmentation contours may be independently adjustable. For example, the selection of an area or point within the segmentation contour may be moved. In this way, a given segmentation contour may be adjusted to better approximate a region of interest.

[0111] As shown in the third row 1234, multiple segmentation contours can be displayed in each of the multiple images. Fig.12 As shown, each of the plurality of segmentation contours may be colored to distinguish between them. For example, the first segmentation contour 1210 may be orange, the second segmentation contour 1212 may be gray, the third segmentation contour 1214 may be blue, and the fourth segmentation contour 1216 may be yellow. In this way, when all are displayed as an overlay in a single image, the different segmentation contours may be distinguished from each other.

[0112] Displaying all segmentation contours within each of the multiple displayed images allows the user to view the segmentation contours compared to the region of interest. In this way, the user can determine which of the multiple segmentation contours best approximates the region of interest as visualized in each of the multiple images. As previously described, each of the multiple segmentation contours can be selected via one or more types of user input (e.g., mouse click, key touch, etc.) so that the user can select one of the multiple images. With annotations of multi-volume imaging (such as multi-parameter and / or multi-phase imaging), the region of interest can be viewed as imaged from various sequences or phases, and segmentation contours can be generated for each of the volumes (e.g., at a specific slice). Multiple segmentation contours can allow for increased accuracy of annotation and segmentation because a more robust approximation of the region of interest can be determined.

[0113] Now go to Fig.13 , shows a diagram of the geodesic propagation around the user cursor and the corresponding defined area. Figure 4 As described, when the contour point positions are determined from the user cursor position, a geodesic propagation 1302 is generated around the user cursor 1308. The geodesic propagation 1302 can be circular. The geodesic propagation 1302 is constrained by a radius 1320, which depends on the tonal distribution, as previously described. When the pixels (or voxels) within the geodesic propagation 1302 have the same intensity, the area of ​​the geodesic propagation can be considered to be a flat area 1306.

[0114] When the user cursor 1308 is within the edge of a region with varying pixel intensities, which in some examples may be a region of interest of a medical image, one or more gradients may be detected. When a gradient is detected within the geodesic propagation 1302, the user cursor 1308 may be within the edge of the region of interest. The gradient may define an edge 1322 that causes the designated region 1304 around the user cursor 1308 to not be circular. The closest point 1310 of the edge 1322 to the user cursor 1308 may be a contour point for the corresponding segmentation contour.

[0115] The generation of geodesic propagation and the detection of gradients may be performed independently for each volume of a 3D multi-volume imaging data set comprising a plurality of registered image volumes. For example, 2D slices may be displayed, each 2D slice belonging to one of the plurality of registered image volumes, and contour points may be defined individually for each 2D slice based on the geodesic propagation and the detected gradients, as previously described. In this way, a single user cursor position input by a user during annotation may define points of a segmentation contour for each volume.

[0116] about Fig.14 and Fig.15 The changing tonal distribution and resulting radius of the designated area around the user cursor are shown. Fig.14 1400. In the first position, the user cursor 1404 is superimposed on the medical image 1402 displayed within the GUI 1400. In the first position, the user cursor 1404 is positioned to be superimposed on a region 1406 of the brain. The region 1406 may have a defined hue distribution and / or a range of hue distributions thereof. The user cursor 1404 may be within a distance of the region of interest 1408, wherein the distance is less than a radius of geodesic propagation around the user cursor 1404. The radius of the geodesic propagation region may depend on the defined hue distribution of the region 1406. The designated area 1410 around the user cursor 1404 displayed as an overlay within the GUI 1400 may take into account detected gradients, such as those associated with the edges of the region of interest 1408, and therefore may not be circular, as with respect to Fig.13 as described.

[0117] exist Fig.15 14. In the example of FIG. 14A , a user cursor 1504 is superimposed on a medical image 1502 displayed within the GUI 1500 in a second position. In some examples, the user cursor 1504 may be the user cursor 1404, the medical image 1502 may be the medical image 1402, and the GUI 1500 may be the GUI 1400. In the second position, the user cursor 1504 may be positioned superimposed on a background 1506, as opposed to the user cursor 1404 being positioned superimposed on a region 1406 of a brain in the medical image. The background 1506 may include pixels of all the same pixel intensity. The radius 1512 of the geodesic propagation around the user cursor 1404 may be larger than the radius of the geodesic propagation region around the user cursor 1504 because the defined hue distribution of the background 1506 is smaller than the hue distribution of the region 1406. A smaller hue distribution may result in a larger radius of the potentially designated region. Similar to Fig.14 , edge 1508 (e.g., the edge of the brain) may be within the geodesic propagation region because the distance between edge 1508 and user cursor 1504 is less than the radius of the geodesic propagation region. Designated area 1510 displayed as an overlay within GUI 1500 may not be circular due to detected edge 1508, as described with respect to FIG. Figure 4 as described.

[0118] The technical effect of the methods and systems provided herein is that multi-volume imaging data can be annotated to more accurately segment regions of interest of the multi-volume imaging data. The annotation can include user input to one of the multiple volumes (e.g., to define the position of a user cursor) and generation of contour points and segmentation contours for each of the multiple volumes based on the position of the user cursor. The determination of the contour points can be based on flat area characterization and determining the closest point of the gradient / detected edge to the user cursor position. Multiple options for the selected segmentation contour can allow for more accurate segmentation.

[0119] In addition, by generating a segmentation contour for each of the volumes of multi-volume imaging data via user input to one volume (e.g., an image of the volume), the processing requirements for the computing device can be reduced because the amount of user input can be reduced. For example, when generating a segmentation contour for a set of image slices, the input to a single image slice is used to determine the corresponding contour points for each of the two or more image slices. The algorithm that defines the corresponding contour points can be applied to each of the two or more image slices simultaneously based on the input to the single image slice. Therefore, applying the input to each individual image slice and separate completion of the algorithm (which would result in increased processing power and time spent) is avoided, and the computing system as a whole requires a smaller amount of processing.

[0120] In addition, the annotation and generation of the segmentation contours as described herein can be performed in real time, whereby when an annotation input (e.g., a mouse click or drag) is added to an image, contour points are added to the corresponding image in real time. In this way, the methods and systems herein can reduce the time spent by the user in annotation and segmentation and increase processing efficiency.

[0121] The present disclosure also provides support for a method, the method comprising: obtaining three-dimensional (3D) multi-volume imaging data, the 3D multi-volume imaging data including a plurality of registered image volumes; determining one or more positions of a user cursor within a first image of a first image volume in the plurality of registered image volumes; determining a contour point position corresponding to each of the one or more positions of the user cursor for two or more of the plurality of registered image volumes; generating two or more segmentation contours corresponding to two or more of the plurality of registered image volumes; determining a selected segmentation contour of the two or more segmentation contours via a user input to a display device; generating a three-dimensional (3D) segmentation mask based at least in part on the selected segmentation contour; and saving the 3D segmentation mask to a memory. In a first example of the method, the plurality of registered image volumes are each divided into a plurality of image slices, and the first image is a first image slice of the first image volume. In a second example of the method, optionally including the first example, the contour point is one of a plurality of points corresponding to the segmentation contour. In a third example of the method, optionally including one or both of the first example and the second example, the 3D multi-volume imaging data is acquired by an imaging system configured as one of or multi-parameter imaging data and multi-phase imaging data. In a fourth example of the method, optionally including one or more or each of the first to third examples, a representation of the segmentation mask is displayed to a user via a display device. In a fifth example of the method, optionally including one or more or each of the first to fourth examples, each of the plurality of registered image volumes includes data of a region of interest, and the segmentation mask corresponds to the region of interest. In a sixth example of the method, optionally including one or more or each of the first to fifth examples, determining each of two or more segmentation contours includes, for each of one or more positions of a user cursor: defining a geodesic propagation around the position of the user cursor, wherein the position of the user cursor is within a certain distance of the region of interest of the first image, the distance being less than a radius of the geodesic propagation; detecting one or more gradients within the geodesic propagation, wherein the one or more gradients are associated with an edge of the region of interest; defining an edge of the region of interest based on the one or more gradients; and determining a point of the edge closest to the position of the user cursor, wherein the point is one of the two or more points of the corresponding segmentation contour. In a seventh example of the method, optionally including one or more or each of the first to sixth examples, when the geodesic propagation is 2D, the geodesic propagation is circular and the radius of the geodesic propagation depends on the distribution of tones within the geodesic propagation. In an eighth example of the method, optionally including one or more or each of the first to seventh examples, generating the segmentation mask includes interpolating between two or more selected segmentation contours.

[0122] The present disclosure also provides support for a system, the system comprising: an imaging system, the imaging system being configured to acquire three-dimensional (3D) multi-volume imaging data; and a computing device, the computing device being communicatively coupled to the imaging system, the computing device being configured with instructions stored in a non-transitory memory and executable by a processor, the instructions, when executed, causing the processor to: obtain a 3D multi-volume imaging dataset from an imaging archive, wherein the 3D multi-volume imaging dataset comprises a plurality of registered image volumes, the plurality of registered image volumes each comprising a plurality of image slices; display a first image of two or more of the plurality of registered image volumes in a graphical user interface (GUI) of a display device communicatively coupled to the computing device; An image slice in a set of image slices; determining two or more positions of a user cursor within a first image slice of a first image volume in response to a user input; determining contour points within two or more image slices corresponding to two or more of a plurality of registered image volumes based on the two or more positions of the user cursor within the first image slice; generating two or more segmentation contours, wherein each of the two or more segmentation contours corresponds to one of the two or more image slices and includes a corresponding plurality of contour points; selecting a selected segmentation contour of the two or more segmentation contours in response to the user input; generating a 3D segmentation mask based on the selected contour; and saving the 3D segmentation mask to a non-transitory memory. In a first example of the system, each of the plurality of registered image volumes corresponds to one of a different imaging sequence and a different imaging phase. In a second example of the system, optionally including the first example, each of the two or more segmentation contours is displayed on a display device in a different color, wherein the first segmentation color is displayed in a first color and the second segmentation contour is displayed in a different second color. In a third example of the system, optionally including one or both of the first example and the second example, two or more positions of the user cursor within each of the two or more image slices are known. In a fourth example of the system, optionally including one or more or each of the first to third examples, the computing device is further configured with instructions stored in a non-transitory memory, which when executed causes the processor to determine two or more second segmentation contours corresponding to the second set of image slices. In a fifth example of the system, optionally including one or more or each of the first to fourth examples, the computing device is further configured with instructions stored in a non-transitory memory, which when executed causes the processor to select a second selected segmentation contour among the two or more second segmentation contours in response to user input.

[0123] The present disclosure also provides support for a method, the method comprising: determining a position of a user cursor within a first image of a first image volume in a plurality of image volumes of a 3D multi-volume imaging data set; generating a geodesic propagation around the position of the user cursor within two or more images of two or more corresponding image volumes in the plurality of image volumes; determining one or more gradients within the geodesic propagation of each of the two or more images; defining an edge of an area of ​​interest within the geodesic propagation of each of the two or more images based on the one or more gradients; determining a point of the edge closest to the position of the user cursor within each of the two or more images; and saving the point as one of a plurality of points of a segmentation contour for each of the two or more images. In a first example of the method, the method further comprises: generating a segmentation mask based at least in part on a selected segmentation contour of one of the two or more images, and displaying a representation of the segmentation mask. In a second example of the method, optionally including the first example, a previous point of the corresponding segmentation contour stabilizes the point. In a third example of the method, optionally including one or both of the first and second examples, the geodesic propagation is constrained to a radius according to a hue distribution within the geodesic propagation. In a fourth example of the method, optionally including one or more or each of the first to third examples, the edge of the region of interest is located at a distance from the position of the user cursor that is less than the radius of geodesic propagation.

[0124] As used herein, the elements or steps listed in the singular and beginning with the word "one" or "an" should be understood as not excluding a plurality of the elements or steps, unless such exclusion is explicitly stated. In addition, the reference to "one embodiment" of the present invention is not intended to be interpreted as excluding the existence of additional embodiments that also include the cited features. In addition, unless explicitly stated to the contrary, "comprising", "including" or "having" an embodiment of an element or multiple elements with a specific characteristic may include additional such elements that do not have the characteristic. The terms "including" and "in..." are used as the concise language equivalents of the corresponding terms "including" and "wherein". In addition, the terms "first", "second" and "third" etc. are only used as marks, and are not intended to impose numerical requirements or specific positional order on their objects.

[0125] This written description uses examples to disclose the invention, including the best mode, and also to enable a person skilled in the relevant art to practice the invention, including making and using any devices or systems and performing any included methods. The patentable scope of the invention is defined by the claims, and may include other examples that occur to a person skilled in the art. Such other examples are intended to be within the scope of the claims if they have structural elements that do not differ from the literal language of the claims, or if they include equivalent structural elements with insignificant differences from the literal language of the claims.

Claims

1. A method, comprising: obtaining three-dimensional (3D) multi-volume imaging data, the 3D multi-volume imaging data comprising a plurality of registered image volumes (302); determining one or more positions of a user cursor within a first image of a first image volume of the plurality of registered image volumes (306); determining, for two or more of the plurality of registered image volumes, a contour point location corresponding to each of the one or more positions of the user cursor (308); generating two or more segmentation contours corresponding to the two or more of the plurality of registered image volumes (308); determining, via user input to a display device, a selected segmentation contour of the two or more segmentation contours (314); generating a three-dimensional (3D) segmentation mask (516) based at least in part on the selected segmentation contour; as well as The 3D segmentation mask is saved to a memory.

2. The method of claim 1, wherein the plurality of registered image volumes are each divided into a plurality of image slices, and the first image is a first image slice of the first image volume (504).

3. The method of claim 1, wherein the contour point is one of a plurality of points corresponding to a segmentation contour (422).

4. The method of claim 1, wherein the 3D multi-volume imaging data is acquired by an imaging system (100) configured to acquire one of multi-parameter imaging data and multi-phase imaging data.

5. The method of claim 1, wherein a representation of the segmentation mask is displayed to the user via the display device (310).

6. The method of claim 1, wherein each of the plurality of registered image volumes comprises data of a region of interest (610), and the segmentation mask corresponds to the region of interest.

7. The method of claim 1 , wherein determining each of the two or more segmentation contours comprises, for each of the one or more positions of the user cursor: defining a geodesic propagation around a position of the user cursor (408), wherein the position of the user cursor is within a distance of a region of interest of the first image, the distance being less than a radius of the geodesic propagation; detecting one or more gradients within the geodesic propagation, wherein the one or more gradients are associated with edges of the region of interest (412); defining the edge of the region of interest based on the one or more gradients (414); and A point of the edge that is closest to the location of the user cursor is determined (418), wherein the point is one of two or more points of a corresponding segmentation contour.

8. The method of claim 7, wherein when the geodesic propagation is 2D, the geodesic propagation is circular (408), and the radius of the geodesic propagation depends on a hue distribution (410) within the geodesic propagation.

9. The method of claim 1, wherein generating the segmentation mask comprises interpolating between two or more selected segmentation contours (514).

10. A system, comprising: An imaging system (100) configured to acquire three-dimensional (3D) multi-volume imaging data; and A computing device communicatively coupled to the imaging system, the computing device being configured with instructions stored in a non-transitory memory executable by a processor, the instructions when executed causing the processor to: Obtaining a 3D multi-volume imaging dataset (302) from an imaging archive, wherein the 3D multi-volume imaging dataset comprises a plurality of registered image volumes, each of the plurality of registered image volumes comprising a plurality of image slices (504); displaying image slices from a first set of image slices of two or more of the plurality of registered image volumes in a graphical user interface (GUI) of a display device communicatively coupled to the computing device (404); determining two or more positions of a user cursor within a first image slice of a first image volume in response to the user input (406); determining contour points within two or more image slices corresponding to two or more of the plurality of registered image volumes based on the two or more positions of the user cursor within the first image slice (308); generating two or more segmentation contours, wherein each of the two or more segmentation contours corresponds to one of the two or more image slices and includes a corresponding plurality of contour points (308); selecting a selected segmentation contour of the two or more segmentation contours in response to a user input (314); generating a 3D segmentation mask based on the selected contour (516); as well as The 3D segmentation mask is saved to a non-volatile memory.

11. The system of claim 10, wherein each of the plurality of registered image volumes corresponds to one of a different imaging sequence and a different imaging phase (302).

12. The system of claim 10, wherein each of the two or more segmentation contours (1210, 1212) is displayed on the display device in a different color, wherein a first segmentation contour is displayed in a first color and a second segmentation contour is displayed in a different second color.

13. The system of claim 10, wherein the two or more positions of the user cursor are known within each of the two or more image slices (306).

14. The system of claim 10, wherein the computing device is further configured with instructions stored in a non-volatile memory that, when executed, cause the processor to determine two or more second segmentation contours corresponding to a second set of image slices (510).

15. A system according to claim 14, wherein the computing device is further configured with instructions stored in a non-volatile memory, which instructions, when executed, cause the processor to select a second selected segmentation contour (512) among the two or more second segmentation contours in response to user input.

Images