MODEL-BASED IMAGE SEGMENTATION

Abstract

Presented are concepts for initialising a model for model-based segmentation of an image which use specific landmarks (e.g. detected using other techniques) to initialize the segmentation mesh. Using such an approach, embodiments need not be limited to predefined model transformations, but can initialise a segmentation mesh with arbitrary shape. In this way, embodiments may provide for an image segmentation algorithm that not only delivers a robust surface-based segmentation result but also does so for strongly varying target structure variations (in terms of shape).

Claims

1. A computer-implemented method of initialising a model of a model space for model-based segmentation of an image, the method comprising: defining a set of landmarks in a model space, wherein defining a set of landmarks comprises positioning the set of landmarks based on a predefined structure, wherein positioning the set of landmarks comprises distributing the set of landmarks within, in or around a boundary of the predefined structure; for each of the landmarks, determining a respective connection to the model in the model space, the connection being representative of the position of the landmark relative to the model; detecting the set of landmarks in the image; and placing the model within the detected set of landmarks based on the determined connections for the landmarks.

2. The method of claim 1, wherein placing the model within the detected set of landmarks comprises: interpolating the model shape based on the determined connections for the landmarks.

3. The method of claim 1, wherein placing the model within the detected set of landmarks comprises: deforming the model while, for each landmark, maintaining the position of the landmark relative to the model represented by its respective connection.

4. The method of claim 3, wherein deforming the model is based on the internal energy of the model.

5. The method of claim 1, wherein distributing the set of landmarks comprises: uniformly distributing the set of landmarks within, in or around the boundary of the predefined structure.

6. The method of claim 1, wherein distributing the set of landmarks comprises: arbitrarily distributing the set of landmarks with a predefined spatial organisation within, in or around the boundary of the predefined structure.

7. The method of claim 1, wherein a landmark's respective connection comprises a linear connection between the landmark and the model.

8. The method of claim 7, wherein the linear connection defines the distance between the landmark and the model.

9. A computer-implemented method of model-based segmentation of a medical image comprising: initialising a model for model-based segmentation of the medical image according to claim 1.

10. A computer program comprising computer program code means adapted, when said computer program is run on a computer, to implement the method of claim 1.

11. A system for initialising a model of a model space for model-based segmentation of an image, the system comprising: a definition component configured to define a set of landmarks in a model space and configured to position the set of landmarks based on a predefined structure, wherein positioning the set of landmarks comprises distributing the set of landmarks within, in or around a boundary of the predefined structure; an analysis component configured to determine, for each of the landmarks, a respective connection to the model of the model space, the connection being representative of the position of the landmark relative to the model; a detection component configured to detect the set of landmarks in the image; and a placement component configured to place the model within the detected set of landmarks based on the determined connections for the landmarks.

12. The system of claim 11, wherein the placement component is configured to interpolate the model shape based on the determined connections for the landmarks.

13. The system of claim 11, wherein the placement component is configured to deform the model while, for each landmark, maintaining the position of the landmark relative to the model represented by its respective connection.

14. The system of claim 11, wherein the definition component is configured to: uniformly distribute the set of landmarks within, in or around the boundary of the predefined structure; and/or arbitrarily distribute the set of landmarks with a predefined spatial organisation within, in or around the boundary of the predefined structure.

15. The system of claim 11, wherein a landmark's respective connection comprises a linear connection between the landmark and the model.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0024] For a better understanding of the invention, and to show more clearly how it may be carried into effect, reference will now be made, by way of example only, to the accompanying drawings, in which:

[0025] FIG. 1 is a flow diagram of a computer-implemented method for initialising a model of a model space for model-based segmentation of an image according to an embodiment of the invention;

[0026] FIGS. 2A and 2B illustrate an example of a proposed embodiment of initialising a model for model-based segmentation of an image; and

[0027] FIG. 3 is a simplified block diagram of a system for initialising a model of a model space for model-based segmentation of an image according to an embodiment of the invention.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0028] The invention will be described with reference to the FIGS.

[0029] It should be understood that the detailed description and specific examples, while indicating exemplary embodiments of the apparatus, systems and methods, are intended for purposes of illustration only and are not intended to limit the scope of the invention. These and other features, aspects, and advantages of the apparatus, systems and methods of the present invention will become better understood from the following description, appended claims, and accompanying drawings. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0030] Variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure and the appended claims. In the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude a plurality.

[0031] It should be understood that the Figs, are merely schematic and are not drawn to scale. It should also be understood that the same reference numerals are used throughout the Figs. to indicate the same or similar parts.

[0032] Proposed are concepts for model initialisation for model-based segmentation (MBS) of an image. Such concepts may employ a set landmarks to initialise a segmentation mesh.

[0033] As a result, embodiments need not be limited to predefined model transformations, but can initialise a segmentation mesh with arbitrary shape. Embodiments may therefore facilitate an image segmentation algorithm that not only delivers a robust surface-based segmentation result but also does so for a segmentation mesh with arbitrary shape.

[0034] In particular, proposed concepts may an MBS algorithm for medical image segmentation that delivers a robust result for strongly varying target organ variations. Accordingly, embodiments may be used in relation to medical images and/or provide improved Clinical Decision Support (CDS).

[0035] According to a proposed concept, there is provided a landmark-based model initialization approach for model-based organ segmentation. Such an approach uses a set of landmarks (e.g. distributed over the image) to initialize the MBS model, wherein the spatial positions of the landmarks can be correlated with respect to the MBS model. After detecting the landmarks in the image to be segmented, embodiments may place the model shape into the landmark set in such a way that relative positions from model to landmarks are maintained. Once the model is placed into the image, the result can be used as initialization to consequently trigger a MBS pipeline.

[0036] Illustrative embodiments may, for example, be employed in model-based image segmentation systems, such as in medical imaging analysis systems.

[0037] Referring to FIG. 1, there is depicted a flow diagram of a computer-implemented method 100 for initialising a model of a model space for model-based segmentation of an image according to an embodiment of the invention.

[0038] Here, the image may, for example, be a volumetric medical image. For example, the volumetric image may be a computed tomography (CT) image, a magnetic resonance (MR) image, a nuclear medicine image, such as a positron emission tomography (PET) image or a single photon emission computed tomography (SPECT) image, or a volumetric ultrasound image.

[0039] The method begins with step 110 of defining a set of landmarks in a model space. In this example, the step 100 defining the set of landmarks comprises the step 115 of positioning the set of landmarks based on a predefined structure (e.g. a target/reference organ structure). Here then set of landmarks are positioned by distributing the set of landmarks uniformly within, in or around a boundary of the predefined structure. In this way, the landmarks are placed into the model space.

[0040] The method then comprises the step 120 of, for each of the landmarks, determining a respective connection to the model in the model space. Here, a landmark's respective connection comprises a linear connection between the landmark and the model, and is thus representative of the position of the landmark relative to the model. In this example, a linear connection defines the shortest distance between the landmark and the model. It will therefore be a linear connection defined by a line connecting the landmark to the closest point of the model surface, the line being perpendicular to the tangent of the closest point of the model surface. However, in other examples, the connection need not be the shorted distance.

[0041] Then, in step 130, the landmarks are detected in the image to be segmented. For this, a landmark detector (using any known algorithm) identifies the set of landmarks in the image to segment. Several approaches for landmark detection/localization in medical images are described in existing literature. Detailed description of landmark detection is therefore omitted from this description for conciseness.

[0042] After detecting the landmarks, the method continues to the step 140 of placing the model within the detected set of landmarks (so as to ‘activate’ or ‘initialise’ the model). Such placement of the model is based on the determined connections for the landmarks. In particular, placing 140 the model relative to the detected set of landmarks comprises the process 145 of deforming the model while maintaining the position of the landmarks relative to the model represented by their respective connections. This may, for example, be done while considering the model's internal energy to only allow realistic deformations. Further, placing 140 the model within the detected set of landmarks may comprise the step 150 of interpolating the model shape based on the determined connections for the landmarks (e.g. so as to keep the connection lengths stable).

[0043] By way of further description, an example of a proposed embodiment of initialising a model for model-based segmentation of an image will now be described with reference to FIGS. 2A and 2B.

[0044] FIG. 2A depicts a model space, wherein a plurality of landmarks 210 are connected to the model 215 via linear connectors 200.

[0045] FIG. 2B depicts the model space wherein the model has been initialised for the image according to the embodiment, wherein the segmentation mesh 225 is interpolated into the landmark set by keeping the connector lengths stable.

[0046] First, a set of landmarks is defined and then their spatial position with respect to the model 215 is correlated. FIG. 2A illustrates this for a schematic vertebra model. Given a segmentation mesh 225, the set of detectable landmarks 210 is placed into the model space, either outside (e.g. lmk.sub.1 and lmk.sub.2), inside (e.g. lmk.sub.3 and lmk.sub.4), or on (e.g. lmk.sub.5 and lmk.sub.6) the mesh surface 225.

[0047] The spatial correlation, i.e. the relative position of the landmarks 210 with respect to the model 225 may be represented by connecting the landmarks 210 with the model using linear connectors 200.

[0048] Referring now to FIG. 2B, during application, a landmark detector (using any known landmark detection algorithm) firstly detects the landmarks 210 in the image to be segmented. Several approaches for landmark localization in medical images are known and described in the literature. Description of such landmark detection algorithms is therefore omitted from this description.

[0049] Once the landmarks 210 are detected, the model can be “activated”, i.e. based on the required distances from landmarks 210 to model, the model shape is placed into the landmark set. During placement, the model is deformed (e.g. while considering the model's internal energy to only allow realistic deformations) in such a way that the connector 220 lengths (i.e. the relative positions from model to landmarks) is maintained.

[0050] Once the model is placed into the image so to as obtain a modified model 225, this model 225 is use as an initialization to consequently trigger a MBS pipeline.

[0051] Here, it is noted that, if the detected landmarks are evenly distributed around the target outline (e.g. with arbitrary distance), the initialization should be within a capture range of the model, and a successful segmentation can thus be ensured.

[0052] FIG. 3 illustrates a system 300 for initialising a model of a model space for model-based segmentation of an image 305 according to exemplary embodiment.

[0053] The system 300 comprises a definition component 310 that is configured to define a set of landmarks in a model space. Here, the definition component 310 is configured to position the set of landmarks based on a predefined structure, such as a target/reference structure for example.

[0054] An analysis component 320 of the system is then configured to determine, for each of the landmarks, a respective connection to the model of the model space. A connection is representative of the position of the landmark relative to the model. For instance, in this example, a landmark's respective connection comprises a linear connection between the landmark and the model.

[0055] The system 300 further comprises a detection component 330 that is configured to detect the set of landmarks in an image 305 to be segmented (e.g. received via in input interface of the system).

[0056] Based on the determined connections for the landmarks from the detection component 330, a placement component 340 of the system 300 is configured to place the model within the detected set of landmarks. Specifically, the placement component 340 of this example interpolate the model shape based on the determined connections for the landmarks. The placement component 340 is also configured to output the placed (i.e. initialised) model 350 via an output interface of the system 300. Thus, by way of example, the output model 350 may be provide to a MBS pipeline.

[0057] It will be understood that the disclosed methods are computer-implemented methods. As such, there is also proposed a concept of a computer program comprising code means for implementing any described method when said program is run on a processing system.

[0058] The skilled person would be readily capable of developing a processor for carrying out any herein described method. Thus, each step of a flow chart may represent a different action performed by a processor, and may be performed by a respective module of the processing processor.

[0059] As discussed above, the system makes use of a processor to perform the data processing. The processor can be implemented in numerous ways, with software and/or hardware, to perform the various functions required. The processor typically employs one or more microprocessors that may be programmed using software (e.g. microcode) to perform the required functions. The processor may be implemented as a combination of dedicated hardware to perform some functions and one or more programmed microprocessors and associated circuitry to perform other functions.

[0060] Examples of circuitry that may be employed in various embodiments of the present disclosure include, but are not limited to, conventional microprocessors, application specific integrated circuits (ASICs), and field-programmable gate arrays (FPGAs).

[0061] In various implementations, the processor may be associated with one or more storage media such as volatile and non-volatile computer memory such as RAM, PROM, EPROM, and EEPROM. The storage media may be encoded with one or more programs that, when executed on one or more processors and/or controllers, perform the required functions. Various storage media may be fixed within a processor or controller or may be transportable, such that the one or more programs stored thereon can be loaded into a processor.

[0062] Variations to the disclosed embodiments can be understood and effected by those skilled in the art in practicing the claimed invention, from a study of the drawings, the disclosure and the appended claims. In the claims, the word “comprising” does not exclude other elements or steps, and the indefinite article “a” or “an” does not exclude a plurality. A single processor or other unit may fulfill the functions of several items recited in the claims. The mere fact that certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage. A computer program may be stored/distributed on a suitable medium, such as an optical storage medium or a solid-state medium supplied together with or as part of other hardware, but may also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. If the term “adapted to” is used in the claims or description, it is noted the term “adapted to” is intended to be equivalent to the term “configured to”. Any reference signs in the claims should not be construed as limiting the scope.