Method and apparatus for optically tracking subject movements - Patents.com

Abstract

A method and apparatus for motion tracking of a subject located within a scanner is presented. The method comprises generating a baseline 3D surface representation of a surface area of ​​the subject at a first time point (T(0)), generating a subsequent 3D surface representation of the surface area of ​​the subject at a subsequent time point (T(s)), determining a best match registration of the subsequent 3D surface representation with at least one constraint to the baseline 3D surface representation and determining at least one motion tracking parameter. The method may include selecting at least one virtual feature and associating the at least one virtual feature with the baseline 3D surface representation, the constraint including limiting at least one parameter of the at least one virtual feature associated with the best match subsequent 3D surface representation to at least one parameter of the at least one virtual feature associated with the baseline 3D surface representation.

Description

[Technical field]

[0001] The present invention relates to a method and apparatus for tracking the movement of a subject located within a scanner, such as within the bore of a medical scanner, such as a Magnetic Resonance Imaging (MRI) scanner, or a Positron Emission Tomography (PET) scanner, or a combined MRI / PET scanner. [Background technology]

[0002] Over the past decade, many methods have been developed to track subject movements during medical scans, especially for medical brain scans. However, head movement during scanning is a significant problem that causes artifacts and significantly reduces image quality. Also, for scans of other body parts of a subject, such as cardiac or pulmonary scans, unwanted subject movements during the scanning procedure can significantly reduce scan quality.

[0003] Known methods include external tracking systems as well as image-based motion tracking. Many external tracking systems use markers on the subject's head, which can introduce errors and complicate the process of preparing the subject for the scan, making them less user-friendly in a clinical setting. In general, known tracking systems track surface portions of the object during the scanning procedure and apply the detected motion to a correction of the medical scan image or identify whether there is motion above a threshold that indicates that the results of the medical scan may be unacceptable. The operator can then resume the scanning procedure, thereby saving time and reducing costs. However, tracking systems often detect "false motions" that can result in poor or degraded "corrected" medical scan images and / or an unnecessary termination of the scanning procedure.

[0004] US 2020 / 146554 describes a scan monitoring device for medical imaging comprising a controller and a display, where the controller during a scanning session is configured to obtain tracking data of a subject in a medical scanner, obtain scanner data indicative of operating parameters of the medical scanner, determine an output of a verification function based on the tracking data and the scanner data, and control the scan monitoring device in response to the output of the verification function. If the output indicates an erroneous scan, a notification signal may be provided.

[0005] EP 3896441 A1 describes a method for correcting object motion occurring during MR image acquisition for detecting MR signals of the object. The described method is aimed at helping to distinguish between falsely detected motions and real motions of the object under examination, since it is possible to determine whether a detected motion of a marker placed on the object is a real motion that can be performed by the object. The method comprises determining a motion model describing possible motions of the object based on a model function with a defined number of degrees of freedom that the object can use for its motion. The motion of the markers on the object is detected with a motion sensor. The description of the motion model and the motion of the markers are determined in a common coordinate system, and a first motion of the object is determined in the common coordinate system using the description of the motion model, this first motion being the motion that best matches the motion of the markers determined in the common coordinate system using the defined number of degrees of freedom.

[0006] Although this system can reduce the detection of some "false motions", it still requires the generation of a motion model for each object. Furthermore, there is a high risk that some motions, such as unexpected or uncommon motions, will be falsely determined to be false. In addition, the described method is not suitable for detecting false motions that do not exceed the motion model. Summary of the Invention

[0007] It is an object of the present invention to provide a method and apparatus for tracking the movement of a subject located within a scanner, such as a scanner bore, which alleviates at least some of the above-mentioned problems.

[0008] In one embodiment, it is an object to provide a method for ensuring improved quality of motion tracking.

[0009] In one embodiment, it is an object to provide an apparatus capable of performing high quality real-time motion tracking.

[0010] These and other objects are solved by the present invention or its embodiments as defined in the claims and / or described hereinafter.

[0011] The present invention or embodiments thereof have been found to have many additional advantages which will become apparent to those skilled in the art from the following description.

[0012] According to the present invention, it has been found that by providing suppression of false motion detections, as provided by methods and apparatus including the features as defined in the claims and described herein, surprisingly high quality tracking can be obtained.

[0013] In particular, it has been found that the motion of a surface region is not always completely representative of the motion one wishes to track. By suppressing certain detected motions of the surface region, in particular certain local detected motions of the surface region, more accurate motion tracking can be obtained. The present invention presents a method and a system that are systematically applied to suppress such detected motions of surface regions that are not completely representative of the motion one wishes to track. Such motions are also called "false motions".

[0014] Thus, in one embodiment where a particular location of a tracked surface is prone to local movement and does not reflect the global, i.e. tracked, movement of the body part in question, the method and system of the present invention can be effectively applied to suppress such spurious movement in the motion tracking of a subject located within a scanner, which may be, for example, movement due to muscle tension, blinking, movement of the mouth area, etc., depending on the body part being motion tracked.

[0015] In addition to detecting spurious motion, it has been found that detection errors can occur due to, for example, motion (whether spurious or not) that produces shadow areas, changes in the angles of incidence and / or reflection or other optical phenomena, apparatus errors and / or calibration errors, and it has been found that the method and apparatus of the present invention are able to desirably and effectively suppress such detection errors.

[0016] Further advantages and applications will be apparent to those skilled in the art from the following description and examples.

[0017] The method of tracking the movement of a subject positioned within a bore of a scanner of the present invention comprises: generating a baseline 3D surface representation of a surface area of ​​the subject at a first time point (T(0)); generating a subsequent 3D surface representation of the subject's surface region at a subsequent time point (T(s)); determining a best-fit registration of a subsequent 3D surface representation having at least one constraint to a baseline 3D surface representation; Determining at least one motion tracking parameter; and Includes.

[0018] By applying at least one constraint in determining the best matching registration of the subsequent 3D surface representation to the baseline 3D surface representation, false motion detection and / or detection errors can be suppressed, thereby improving the accuracy and quality of motion tracking.

[0019] The method may comprise a computer-implemented method.

[0020] As used herein, the term "subject" is used to mean any subject capable of movement, preferably a living subject, preferably a mammal such as a human being.

[0021] Often, only a portion of the subject is subject to motion tracking, and such a portion is referred to herein as a body part. Advantageously, the body part includes a body part that undergoes a medical scan in the scanner. Advantageously, the body part is an internal part of a body part, such as the brain part of the head, the knee joint part, the heart part of the chest or the lung part.

[0022] The surface area of ​​the subject may advantageously be the surface area of ​​the body part in question, which is advantageously selected to be a characteristic surface area, for example a surface area having a change in curvature and / or color and / or structure, for example a surface area comprising the bridge of the nose if the body part comprises the head, or a surface area comprising the nipples if the body part comprises a part of the chest.

[0023] Motion tracking can advantageously be performed using markerless scanning, i.e., without the addition of markings on the subject or on parts of the subject's body.

[0024] However, in some circumstances it may be beneficial to apply markings, such as markings to the skin of the body part, for example to mark the location of the body part to be scanned. The use of markings is beneficial, for example, when the surface portion of the body part located above the body part to be scanned is relatively uncurved and uniform in structure and / or color.

[0025] The scanner is advantageously a medical scanner, such as an X-ray scanner, an MRI scanner, a CT scanner, a PET scanner, an ultrasound scanner, a bone densitometry (DXA) scanner, and / or any combination thereof.

[0026] The subject, or at least a body part of the subject, is located in the scanner by being located in the scanner target space, i.e. the area in which the scanner can perform a scan. The subject, or at least a body part of the subject, is advantageously located in the scanner bore of the scanner.

[0027] The motion tracking may advantageously be performed in real time.

[0028] In this specification, the term "real-time" is used to mean that the time between generating a subsequent 3D surface representation of the surface area and determining at least one motion tracking parameter is less than 1 second, e.g., less than 0.1 seconds, e.g., less than 0.01 seconds.

[0029] If the motion tracking is performed in real time, one or more scan parameters may be adjusted in response to the real-time motion tracking, such as adjusting the magnetic field, radio waves and / or sequences in the MRI scan, and optionally re-acquiring parts of the scan procedure, preferably in real time. Due to the accuracy of the tracking and the suppression of false motion and / or scan errors, it is often sufficient to re-acquire a relatively short part of the scan procedure, for example up to 5 minutes or less, for example up to 1 minute, for example 1-30 seconds, for example 2-10 seconds, to obtain a high-quality scan. This avoids the need to repeat the entire image acquisition time for acquiring the scan and / or image data, allowing the scan procedure to be performed relatively quickly while ensuring very high scan quality.

[0030] If the motion tracking indicates motion beyond the correctable range, in one embodiment the scan may be stopped, for example to restart the entire scanning procedure, or stopped by the operator, saving both time and cost, as any scanning errors are observed immediately, eliminating the need for the patient to undergo a rescan at a later time.

[0031] In one embodiment, the motion tracking may advantageously be performed with a delay, meaning that the time between generating the subsequent 3D surface representation of the surface region and determining at least one motion tracking parameter is longer than 1 second, e.g., longer than 5 seconds, or even longer. Motion tracking with a delay may be applied, for example, for motion examination, such as motion patterns for selecting appropriate constraints, or for determining the accuracy of the scan results, and / or for post-correction of the scan results, as further described below. For example, in bone densitometry, several scans, e.g., two, three or more scans, are often performed and the results averaged. The method of the present invention may be used to determine whether undesired motion has occurred during the scanning of any of the scans, and if so, such scans may be considered to have a scan error and discarded. For example, in the embodiment, the motion tracking may be real-time or not.

[0032] As used herein, the term "computer system" is used to mean a single computer or multiple computers with data connections via wireless, wired, and / or internet, and the term "computer" means a machine or device that can process data based, at least in part, on instructions provided by a software and / or hardware program. It has the ability to accept data (input), process it, and generate output.

[0033] The term "configured for" is used in the sense of "programmed for" and / or taught by machine learning, which can be, for example, supervised or unsupervised machine learning.

[0034] It is emphasized that the term "comprises / comprising" as used in this specification should be interpreted as an open term, i.e., should be taken as specifying the presence of specifically stated features, such as elements, units, integers, steps, components, and combinations thereof, but does not exclude the presence or addition of one or more other stated features.

[0035] Throughout this specification or claims, unless otherwise indicated or required by context, the singular encompasses the plural and the plural encompasses the singular.

[0036] "Embodiments" should be construed to include examples of the invention that include features of the referred embodiment.

[0037] The term "about" is generally used to include things that are within the uncertainty of measurement. When used in reference to ranges, the term "about" should be interpreted herein to mean that things that are within the uncertainty of measurement are included in the range.

[0038] As used herein, the phrase "region of interests" (ROI) is used to mean a sub-region or set of sub-regions of a baseline 3D surface representation that corresponds to an actual sub-region of the surface region or corresponds to an actual sub-region of a portion of a subject that at least partially correlates with the actual sub-region of the surface region. As used herein, the term "substantially" should be interpreted to mean that normal manufacturing variations and tolerances are included. All features of the invention and embodiments of the invention as described herein, including ranges and preferred ranges, may be combined in various ways within the scope of the invention, unless there is a specific reason not to combine such features.

[0039] The generation of the baseline 3D surface representation and the subsequent 3D surface representations of the surface region is advantageously performed optically by detecting light reflections from the surface region. The baseline 3D surface representation is acquired at a time T(0), which may be any time prior to the time T(s) of acquisition of the subsequent 3D surface representation of the surface region. Advantageously, the method comprises the repeated generation of subsequent 3D surface representations of the surface region and the determination of at least one motion tracking parameter, the applied baseline 3D surface representation being the same or different for each subsequent 3D surface representation of the surface region. In one embodiment, the first baseline 3D surface representation is maintained as the same for multiple, e.g. all subsequent 3D surface representations of the surface region. In one embodiment, the first baseline 3D surface representation is replaced by a second baseline 3D surface representation, e.g. after a selected scanning time, at a selected stage of the scan and / or when the first baseline 3D surface representation is suspected to contain a scanning error.

[0040] In one embodiment, generating the baseline 3D surface representation includes generating a single 3D surface representation of the surface region and applying the single 3D surface representation of the surface region as the baseline 3D surface representation.

[0041] Generation of the baseline 3D surface representation may include one or more filtered images, such as two or more images acquired over a period of up to 30 seconds, e.g. up to 10 seconds, at a frame rate of 1 Hz or more, such as a frame rate of 2 to 100 Hz, e.g. up to 1000 Hz.

[0042] In one embodiment, generating a baseline 3D surface representation comprises generating a plurality of 3D surface representations of the surface area, providing an average or median of the plurality of 3D surface representations of the surface area, and applying this average / median 3D surface representation of the surface area as the baseline 3D surface representation. The average or median of the plurality of 3D surface representations of the surface area may be determined as a point-by-point average / median, where each point may be a respective pixel, or a group of pixels, or a beam of light, depending on the required / desired resolution.

[0043] In one embodiment, generating a baseline 3D surface representation includes generating multiple 3D surface representations of the surface region until a number of generated surface representations of the surface region (e.g., 3 to 10, such as 2 or more) are identical within a selected threshold, and applying the identical surface representations as the baseline 3D surface representation.

[0044] At least one constraint advantageously includes a relative restriction. It is therefore found that spurious movements and / or detection errors are rarely decoupled from real movements, i.e. may be related to and / or partially caused by real movements. Thus, in one embodiment, the constraint includes a relative restriction, which may be a function of the difference between each position of the baseline 3D surface representation and the corresponding respective positions of the subsequent 3D surface representation. The constriction may be applied as a special distance restriction between each position of the baseline 3D surface representation and the corresponding respective positions of the subsequent 3D surface representation, which are homographically coincident positions, i.e. positions representing coincident points of the surface area.

[0045] In one embodiment, the method includes selecting at least one virtual feature and associating the at least one virtual feature with a ROI (region of interest) of the baseline 3D surface representation, wherein the constraint includes limiting a change of at least one parameter of the at least one virtual feature associated with a corresponding ROI of the best-matching subsequent 3D surface representation relative to at least one parameter of the at least one virtual feature associated with the ROI of the baseline 3D surface representation.

[0046] At least one feature may advantageously have a spatial location (i.e. located in the image) relative to the ROI and / or the corresponding ROI. For example, the distance between the virtual feature and each of the ROIs of the baseline 3D surface representation and the corresponding ROI of the subsequent 3D surface representation may be a parameter of the constraint, i.e. the constraint may include a limit on the change in distance from the distance between the virtual feature and the ROI of the baseline 3D surface representation to the distance between the virtual feature and the ROI of the subsequent 3D surface representation, so as to reach a constrained change in distance from the distance between the virtual feature and the ROI of the baseline 3D surface representation to the distance between the virtual feature and the ROI of the best match subsequent 3D surface representation.

[0047] The parameter to be restricted is also called a constraint parameter, for example a constraint parameter may be a distance between the ROI / corresponding ROI.

[0048] The virtual features may advantageously be virtual features having a spatial location at a spatial distance relative to the 3D surface representation of the surface region. The at least one virtual feature may for example include one or more of a virtual point, a (2D or 3D) virtual point cloud, a virtual volume, a virtual region, a virtual line, a virtual bone structure, and / or any combination including one or more of the aforementioned virtual features.

[0049] The association to the baseline 3D surface representation may include an association to a spatial location, orientation, extent, or a combination of one or more, for a point, region, or line of the baseline 3D surface representation.

[0050] Advantageously, the association to the baseline 3D surface representation may include an association to the ROI, such as the spatial position of the ROI, the orientation of the ROI, the extent of the ROI to a point or one or more combinations, a portion or line of the ROI in the baseline 3D surface representation, etc.

[0051] Advantageously, the at least one parameter comprises a position parameter, an orientation parameter, a range parameter, and / or a combination comprising at least one of the aforementioned parameters of the at least one virtual feature.

[0052] In one embodiment, the at least one parameter includes a position parameter, a distance parameter, an orientation parameter, and / or two or more virtual features relative to another or any combination including at least one of the aforementioned parameters of a virtual feature associated with the ROI of the baseline 3D surface representation.

[0053] The constraints on at least one parameter may advantageously include a linear limit, a logarithmic limit, an exponential limit, a maximum limit, and / or a conditional limit including one or more of the foregoing conditions based on a difference in a parameter (e.g., a baseline parameter) from a baseline 3D surface representation to an unconstrained parameter of the subsequent 3D surface representation.

[0054] This suppresses spurious motion and / or detection errors, the degree of which depends on the choice of one or more selected constraints.

[0055] In one embodiment, the constraint of the at least one parameter includes a conditional restriction including one or more of the above, and the condition of the conditional restriction is based on an unconstrained difference of at least one parameter of at least one virtual feature associated with a ROI of the baseline 3D surface representation relative to at least one virtual feature associated with a corresponding ROI of the subsequent 3D surface representation.

[0056] For example, if a subject blinks or closes his / her eyes during a brain scan in which the surface regions include the nose and eye regions, one or more virtual features may be spatially located at a corresponding position inside the head and associated with a portion (e.g., a ROI) of the baseline 3D surface representation that represents the portion of the surface region provided by soft tissues such as the eye region. The constriction may include, for example, a spatial distance constraint indicating a relative constraint on the distance between the virtual feature and the ROI of the baseline 3D surface representation relative to the distance between the virtual feature and the corresponding ROI of the subsequent 3D surface representation. This allows spurious movements, i.e., movements of the eye region that are not related to the movement of the body part during the scan, to be suppressed, while other movements may be recorded with a best match registration of the subsequent 3D surface representation with at least one constraint to the baseline 3D surface representation.

[0057] In one embodiment, the ROI is a subregion or set of subregions of the baseline 3D surface representation that corresponds to an actual subregion of the surface region or that corresponds to an actual subregion of a portion of the subject that at least partially correlates with the actual subregion of the surface region.

[0058] The set of sub-regions may, for example, include two or more points, lines, curves, or areas of the baseline 3D surface representation that correspond to two or more points, lines, curves, or areas of the actual surface region.

[0059] In one embodiment, the ROI includes at least one subregion of the baseline 3D surface representation, and the corresponding ROI is a subregion of the subsequent 3D surface representation and / or a subregion of the subsequent 3D surface representation that is a best match that corresponds to the actual subregion that corresponds to the ROI of the baseline 3D surface representation.

[0060] The ROIs may conveniently be located in the baseline 3D surface representation, with the location of the ROI corresponding to the location of the corresponding ROI in the actual sub-region of the surface area.

[0061] In one embodiment, the ROI may be characterized by at least one of spatial location, orientation, shape, extent, area, or a combination including one or more of the aforementioned characteristics, and each virtual feature may be associated with the baseline 3D surface representation by an association including one or more of the aforementioned characteristics.

[0062] An ROI may include, for example, a 0D subregion such as a point, a 1D subregion such as a line, a 2D subregion such as a surface portion, a 3D subregion such as a volume, or any combination thereof.

[0063] It is desirable to illuminate the surface area to ensure sufficient brightness of the reflected light, so that the angle of incidence of the light, the type and structure of the light can also be selected to provide the desired quality of motion tracking.

[0064] In one embodiment, the generation of the baseline 3D surface representation and each of the subsequent 3D surface representations does not require special illumination and does not require the projection of light onto the surface area; incident light is sufficient.

[0065] In one embodiment, generating each of the baseline 3D surface representation and the subsequent 3D surface representation includes projecting light from a projector device towards a surface area of ​​the subject and detecting a reflection at a detection position of the projected light.

[0066] In one embodiment, the constraint comprises providing that at least one virtual feature associated with the best match registered subsequent 3D surface representation is transformed compared to a case without the at least one constraint, whereby at least one spurious movement and / or at least one detection error is suppressed in the best match registration of the subsequent 3D surface representation. Advantageously, the transformed virtual feature comprises a transformation of at least one parameter (also referred to as a "transformed parameter"), the transformation comprising at least one restriction of the at least one parameter.

[0067] A constraint can in principle have any number of degrees of freedom, the term "degrees of freedom" being referred to below as DOF.

[0068] In one embodiment, the constraint has at least one DOF, preferably selected from a translational axis and a rotational axis. Optionally, the constraint is limited to only one translational direction along the translational axis. In one embodiment, the at least one DOF includes at least one parametric DOF, such as, for example, a DOF of a transformation of the at least one parameter.

[0069] The constraint may advantageously be an X DOF constraint, where X is an integer between 1 and 6. Preferably, X is an integer between 1 and 3, such as a 1 DOF constraint, a 2 DOF constraint or a 3 DOF constraint, and the constraint is or includes 1, 2 or 3 translational axes.

[0070] Thus, a constraint may have one or more weight attributes, i.e., one for each DOF, with at least one of the weight attributes providing a restriction. If a constraint has only one DOF, then the constraint may have only one weight attribute.

[0071] In certain motion tracking, it may be desirable for the constraints to include more than one DOF, especially if a significant number of spurious motions and / or detection errors are expected. However, the more DOFs a constraint has, the more complex the decision may become. Alternatively, having too many DOFs in a constraint may increase the number of constraints.

[0072] Advantageously, a constraint is associated with a weight attribute representing a weight value of the constraint, and preferably a constraint is associated with a set of weight attributes comprising at least two weight attributes. Advantageously, a constraint is associated with a weight attribute for each DOF.

[0073] As used herein, a weight attribute refers to the weighting of at least one parameter of at least one virtual feature associated with the best-matching subsequent 3D surface representation relative to at least one parameter of at least one virtual feature associated with the baseline 3D surface representation.

[0074] The virtual features may be associated with the best-matched subsequent 3D surface representation by being associated with the corresponding ROI of the best-matched subsequent 3D surface representation.

[0075] The virtual features may be associated with the baseline 3D surface by being associated with an ROI in the baseline 3D surface representation.

[0076] For example, if the parameters include a distance parameter providing a constraint selected, for example, from a linear or logarithmic constraint, the weight attribute represents the weight of the constraint, i.e., how much the constraint should be applied.

[0077] Advantageously, the set of weight attributes includes, for two or more particular DOFs, a weight attribute that represents a weight of the constraint for the particular DOF, hi one embodiment, the constraint is a six DOF constraint and the set of weight attributes includes a weight attribute for each DOF of the six DOF constraint.

[0078] The set of weight attributes may include weight attributes whose values ​​are equal or different from one another, e.g., the values ​​of each weight attribute in the set of weight attributes may be selected independently of one another.

[0079] Advantageously, at least one virtual feature associated with the baseline 3D surface representation has baseline parameters, and a value of each weight attribute of the set of weight attributes may advantageously be selected depending on the at least one virtual feature and its baseline parameters, such as a baseline position parameter, a baseline orientation parameter, a baseline range parameter, and / or a baseline distance parameter.

[0080] For example, in one embodiment, the virtual feature is a virtual point located on the subject's head, e.g., at the center of the head, and the parameter of the virtual feature associated with the baseline 3D surface representation is a distance parameter, i.e., the baseline parameter is the distance between the virtual feature and the ROI of the baseline 3D surface representation with which the virtual feature is associated, e.g., a point associated with the baseline 3D surface representation. And the constraint may include a constraint on the distance from the virtual feature of the subsequent 3D surface representation of the surface region to the ROI that includes the relevant point of the subsequent 3D surface representation of the associated corresponding surface region, and the weight of the DOF of the pinch may be selected depending on the virtual feature and the baseline parameter.

[0081] Advantageously, the baseline parameters are at least one parameter of said at least one virtual feature associated with the ROI of the baseline 3D surface representation.

[0082] The weights of each weight attribute may be selected in a number of ways. For example, the weight attributes may be selected arbitrarily by an operator based on experience or by trial and error. However, for some types of scans, such an arbitrary selection may not be desirable or appropriate. In order to optimize the weight attributes, it is desirable for such weight attributes to be derived from modeling.

[0083] In one embodiment, the weights of each weight attribute of the set of weight attributes are preferably derived from modeling the expected movement of at least one virtual feature due to a movement of an anatomical model of the subject's body part including the surface region. Deriving the weight attributes by such a modeling method allows a very accurate correlation between the movement of the virtual points and the movement of the surface region with respect to the movement of the body part being scanned, thereby allowing an optimized weight attribute to be conveniently obtained. Advantageously, the weights of each weight attribute of the set of weight attributes are derived from modeling the expected movement of at least one virtual feature due to a movement of an anatomical model of the subject's body part including the surface region, relative to a baseline parameter of the virtual feature. In one embodiment, the subject's body part includes a surface region including a human head.

[0084] In one embodiment, the value of each weight attribute of the set of weight attributes may be dynamically adjusted. For example, the value of each weight attribute of the set of weight attributes may be dynamically adjusted depending on a subsequent 3D surface representation of the surface region, such as depending on noise in the subsequent 3D surface representation of the surface region. For example, in one embodiment, the value of at least one weight attribute of the plurality of weight attributes may increase as noise increases, thereby increasing suppression of potential detection errors that may be contributed to by the noise.

[0085] In one embodiment, it may be desirable to determine unconstrained parameters of at least one virtual feature associated with a corresponding ROI of a subsequent 3D surface representation of the subject's surface region at a subsequent time point, where a movement of the subject's body part including the surface region between the first time point and the subsequent time point provides that the unconstrained parameters are different from the transformed parameters, i.e. the parameters after being subjected to the constraints. The unconstrained parameters in the parameters, i.e. the distance, orientation, etc. of the virtual feature associated with the corresponding ROI of the subsequent 3D surface representation of the unconstrained surface region, can then be compared to the transformed parameters to observe the level of constraints. This can be applied, for example, as a control of the operation of the method and system.

[0086] If the parameter is a distance parameter, the transformed parameter may advantageously be closer to the baseline parameter than the unconstrained parameter.

[0087] The 3D surface representation may advantageously be acquired using an acquisition device. The acquisition device may for example be one described in EP 10,912,461 or EP 2,547,255. Thus, in one embodiment, the acquisition device comprises a fiber bundle having a first end for collecting reflected light from the surface area and a second end for delivering light propagated in the fibers of the fiber bundle to a detector / camera. In one embodiment, the acquisition device comprises a borescope. In one embodiment, the acquisition device comprises one or more 2D and / or 3D readers, such as one or more detectors / cameras, including at least one mono or stereo camera comprising an array of pixel sensors. Each pixel sensor preferably comprises a photodetector, such as an avalanche photodiode (APD), a photomultiplier tube, or a metal-semiconductor-metal photodetector (MSM photodetector). The pixel sensor may advantageously comprise one or more active pixel sensors (APS). Each pixel sensor may comprise an amplifier. The associated detector may advantageously include at least about 1 kilopixel, for example at least about 1 megapixel. In one embodiment, the acquisition device is a charge-coupled device (CDD) image sensor, or a complementary metal-oxide semiconductor (CMOS) image sensor.

[0088] Advantageously, at least one virtual feature associated with the baseline 3D surface representation is located in the surface region or at a greater distance from the surface region than an acquisition configuration that acquires reflected light from the surface region to generate the respective surface representation.

[0089] Advantageously, at least one virtual feature associated with the baseline 3D surface representation is located between the surface region and a support supporting the subject, such as a scanner bed. Preferably, at least one virtual feature associated with the baseline 3D surface representation is located in an intermediate region between the surface region and the support, which may extend from 40% of the distance from the support to the surface region to 75% of the distance from the support to the surface region, for example from 50% to 65% of the distance from the support to the surface region. The distance from the support to the surface may be the distance between the surface region and the support determined from the acquisition device, for example from a point of the acquisition device closest to the surface region and / or from a focal point of the reflected light of the acquisition device.

[0090] In one embodiment, the surface region comprises a surface region of a body part of the subject, the virtual features comprise virtual points, virtual lines or virtual bone structures located within the volume of the body part at least at a first time point (i.e. when acquiring reflected light for generation of the baseline 3D surface representation at T(0)), and the parameters comprise position parameters, orientation parameters and / or distance parameters. Preferably, the body part comprises a human head and the surface region comprises at least a portion of the curvature of the bridge of the subject's nose.

[0091] In one embodiment, the virtual feature, such as a virtual point, is located within the body part, e.g., within the body part being scanned. The virtual feature may be located, for example, at a distance from the surface area, determined as the minimum distance at time T(0), of at least 1 cm, such as at least 2 cm, for example at least 3 cm, such as at least 4 cm, for example at least 5 cm, determined from the position of the acquisition device.

[0092] In one embodiment, the body part includes a bone structure, the bone structure including a surface area that contributes to shaping a surface portion of the subject within the scanner, and the method comprises selecting a plurality of virtual features (e.g., virtual points) located on the bone structure surface area, each virtual feature associated with an ROI (e.g., a position of the baseline 3D surface representation at a respective position corresponding to the shortest distance between the virtual feature and a (ROI) position of the baseline 3D surface representation) and / or a distance line between the virtual feature and the associated ROI / position of the baseline 3D surface representation is perpendicular to the bone structure surface portion.

[0093] In this embodiment, the constraints may advantageously include limitations on the length of each of the distance lines (hence the limited parameters). The limitations on the length of each of the distance lines may advantageously be different in at least two limits. The distance lines may be considered as tension / compression springs that may have an initial length determined with respect to a baseline 3D surface representation, and the change in spring length is limited by a value k, which may be a constant or a function, and k may be different from one tension / compression spring to another. For example, if k is 0.1 and the change in the distance line with respect to a subsequent 3D surface representation of an unconstrained surface area is a millimeters, then the resulting length of the distance line (i.e. the length of the resulting constant / compression spring with k value 0.1) will be 0.1 x a.

[0094] In one embodiment, the k value is a function, such as a function that depends on the magnitude of the change in the distance line for a subsequent 3D surface representation of the unconstrained surface region, or a function that depends on another distance line for a subsequent 3D surface representation of the unconstrained surface region.

[0095] In one embodiment, the body part is a portion of the thorax including the heart, including the body part that is the subject of the scan, and the virtual feature, such as a virtual line, is located partially or completely within the body part.

[0096] In one embodiment, the body part includes a joint such as a knee, elbow, ankle, wrist or shoulder. The virtual features, such as two or more points, may be conveniently located partially or completely within the knee, elbow, ankle, wrist or shoulder.

[0097] In one embodiment, the surface area comprises a surface area of ​​a body part of the subject, the virtual features comprise the surface area of ​​the body part and / or a virtual volume and / or a virtual area located at least partially within the volume at a first time point, and the parameters comprise position parameters, orientation parameters, and / or extension parameters. Preferably, the body part comprises a human head and the surface area comprises at least a portion of the curvature of the bridge of the subject's nose.

[0098] In one embodiment, the method comprises selecting at least one virtual feature having a feature position at T(0) by providing an estimated motion of the subject during the scanning procedure, and selecting a virtual feature having a feature position at T(0) that is subject to less constrained parameter variations than other virtual features of the subject and / or is subject to parameter variations below a preselected level. Providing the estimated motion of the subject during the scanning procedure may preferably be done from determined motions of the subject during previous scanning procedures and / or from modeling the subject's motion within the scanner.

[0099] Advantageously, the method comprises selecting at least one virtual feature having a feature location at T(0) by providing an estimated motion of the subject during the scan section, and selecting the virtual feature having the feature location at T(0) such that the virtual feature is positioned to undergo less parameter change than other virtual features of the subject. In one embodiment, the method comprises selecting at least one virtual feature having a feature location at T(0) by providing an estimated motion of the subject during the scan section, and selecting the virtual feature having the feature location at T(0) such that the virtual feature is positioned to undergo less parameter change than an associated baseline 3D surface representation.

[0100] In one embodiment, the surface region comprises a surface region of the subject's body part, and the virtual feature having a feature position at T(0) is selected to be a virtual feature that is at least partially located in a central region of the body part at T(0), the central region being defined as a region within the body part that is at a distance to any surface of the body part of at least 1 cm, such as at least 2 cm, for example at least 3 cm, such as at least 4 cm, for example at least 5 cm, as determined from the acquisition device.

[0101] In one embodiment, the surface region includes a surface region of a body part of the subject, and the virtual feature having a feature position at T(0) is selected to be a virtual feature that is at least partially located on the body part that is the subject of the scan.

[0102] In one embodiment, the virtual feature at the position at T(0) is selected to be a virtual point located in the central region of the head at T(0). In this embodiment, the constraint parameters may be, for example, position parameters and / or distance parameters, such as, for example, parameters of the distance of the virtual point to another virtual feature and / or to a real feature.

[0103] In one embodiment, the virtual feature having a feature position at T(0) is selected to be a virtual line located across the central region of the head at T(0). In this embodiment, the constraint parameters may be, for example, position parameters, orientation parameters, and / or distance parameters, such as a parameter of the average distance of the virtual line to another virtual feature and / or to a real feature.

[0104] In one embodiment, the virtual feature having a feature position at T(0) is selected to be a virtual volume that is located within and includes a central region of the head at T(0), e.g., a volume of the head at T(0). In this embodiment, the parameters may be range parameters, e.g., position parameters, distance parameters, size of the volume, and / or orientation parameters, where the volume is not perfectly rotationally symmetric.

[0105] In one embodiment, the virtual feature having a feature position at T(0) is selected to be a virtual region located in a hard portion of the surface region at T(0). In this embodiment, the parameters may be, for example, position parameters, distance parameters, range parameters such as size of the region, and / or orientation parameters, where the region is not perfectly rotationally symmetric.

[0106] In one embodiment, the virtual feature having a feature position at T(0) is preferably selected to be a virtual bone structure, also called a bone rig, which includes at least two virtual bone sections that may be advantageously interconnected. In this embodiment, the parameters may be, for example, position parameters, such as position parameters for the relative positions between the bone sections of the bone structure, orientation parameters, such as orientation parameters for the relative orientation of the bone sections, distance parameters, such as the relative positions between the bone sections of the bone structure, and / or any combination thereof.

[0107] The method may preferably include dynamically selecting a virtual feature having a feature position and at least one parameter at T(0), preferably in response to previously detected movement of the subject.

[0108] In one embodiment, the at least one constraint comprises a constraint on two or more parameters, which may be of the same or different parameter types, e.g., a parameter type comprising one or more position parameters, a parameter type comprising one or more orientation parameters, a parameter type comprising one or more range parameters, and / or a parameter type comprising one or more distance parameters, etc. The limitations on the two or more parameters may advantageously be parameters in two or more respective DOFs.

[0109] In one embodiment, the method includes determining a best matching registration of a subsequent 3D surface representation having at least two constraints to a baseline 3D surface representation. For example, the method may include determining a best matching registration of a subsequent 3D surface representation having N constraints to the baseline 3D surface representation, where N is advantageously up to 10, such as between 2 and 8, for example between 3 and 6.

[0110] Each of the constraints and / or each of the constraint parameters may, independently of one another, have at least one, preferably between 1 and 6 DOFs. Advantageously, each of the constraints or each of the constraint parameters is, independently of one another, associated with a set of weight attributes including at least one weight attribute, preferably including at least two weight attributes, such as up to 6 weight attributes, each weight attribute being associated with a respective DOF and representing the weight of the constraint for a particular DOF.

[0111] A constraint may advantageously have at least one parameter for each DOF of the constraint. The parameters for each DOF of the constraint may be equal or different from each other.

[0112] In one embodiment, the method includes selecting a number of constraints for one or more virtual features, and preferably selecting a set of weight attributes for each constraint, where the weights of the constraints are selected depending on the parameters of the constraints, preferably the baseline parameters (i.e., the parameters at T(0)).

[0113] In one embodiment, the surface area is a surface area of ​​a body part that includes the body part being scanned, and the at least one virtual feature comprises a cloud of points (3D or 2D) located within the body part.

[0114] The point cloud may include points at positions determined by a Gaussian function around a central position (center of rotation) of a body part, such as the head, for example.

[0115] The ROI of the baseline 3D surface representation includes a set of positions in the surface region, and at least a set of points of the points of the cloud are associated with the ROI of the baseline 3D surface representation. The method may include, for example, inserting a plurality of points to form the points of the cloud. The points of the cloud may be associated with the baseline 3D surface representation by being associated with the ROI of the baseline 3D surface representation individually with each other, and the ROI may advantageously be or include a point or a set of points. Thereby, the points of the cloud associated with the baseline 3D surface representation may be individually constrained, i.e. form a group of constraints. Each constraint may have one or more weights as described above, and the weight of each constraint may be fixed or dynamically adjusted.

[0116] The center of mass is considered the center of rotation for realistic head movements.

[0117] To compensate for the fact that there is no single fixed point due to the type of head movement, a Gaussian distribution is used to strategically place points around the center.

[0118] The density and size of the swarm may be determined by the number of points and the standard deviation of the Gaussian, respectively.

[0119] The density and size of the point cloud may be dynamically adjusted.

[0120] The set of points of the point cloud may advantageously include two or more points, such as at least five points of the point cloud, e.g. at least ten points of the point cloud, e.g. at least fifteen points of the point cloud, etc., and the points of the point cloud may be individually associated with the ROI and / or group-wise associated with the ROI, which group-wise association with the ROI may include associating a line between two or more points of the point cloud with the ROI. As mentioned above, it may be beneficial to illuminate and / or project light onto the surface region.

[0121] In one embodiment, the method includes projecting light onto a surface region, where the light may be visible to the human eye (i.e., within the range of 380-700 nanometers) and / or invisible (i.e., outside the range of 380-700 nanometers).

[0122] The projected light, or at least a portion thereof, is conveniently detectable by a capture device.

[0123] In one embodiment, the projected light is a simple illumination, including, for example, polychromatic light.

[0124] In one embodiment, the projection light comprises structured projection light, preferably structured light having an area in cross section in the propagation direction, optionally having optically identifiable regions, optionally in the form of a pattern.

[0125] Examples of optically distinguished regions include patterns of regions with light and regions without light, and / or regions of light of a first quality of characteristic and regions of light of a second quality of characteristic, the characteristic being advantageously selected from light intensity, wavelength and / or range of wavelengths.

[0126] In one embodiment, the method includes projecting light onto the surface region, where the projected light includes monochromatic light (same frequency), preferably coherent light (same frequency and same phase), and in this embodiment, the reflected light from the surface region may include a 3D pattern, such as a 3D point cloud, a speckle pattern, a hologram, and / or any combination thereof, as well as variations thereof.

[0127] In one embodiment, the method includes projecting light onto a surface area, the projected light including a 3D pattern, such as a speckle pattern.

[0128] The use of speckle techniques for motion tracking can include, for example, the use of speckle techniques described in the conference paper by Zizka et al., "SpeckleSense: UIST'11, October 16-19, 2011, Santa Barbara, CA, USA. DOI:10.1145 / 2047196.2047261," and / or by Relly et al., "Adaptive Noncontact Gesture-Based System for Augmentative Communication," IEEE TRANSACTIONS ON REHABILITATION ENGINEERING, VOL. 7, NO. 2, JUNE 1999.

[0129] The method may be advantageously performed using machine learning for selecting virtual features, constraints, weight attributes, parameters, etc.

[0130] In one embodiment, the method includes providing a trained computer system by training the computer system to select at least one virtual feature associated with a baseline 3D surface representation, preferably using a set of reference data to select an ROI and at least one constraint of the baseline 3D surface representation, each reference data set including reference data representing previously determined or modeled movement of a reference subject correlated with reference data representing determined or modeled movement of a reference surface region, the reference surface region being a surface of a reference body part of the reference subject.

[0131] In one embodiment, the training is unsupervised training, although it may be desirable for the training to include supervised training, such as partially or fully supervised training.

[0132] In one embodiment, a method of training a computer system includes training the computer system to select at least one of a number N of constraints (including constraint parameters) and associated virtual features to be applied, basis parameters for each of the virtual features, a number X of degrees of freedom for each constraint and / or parameter, and optionally a set of associated weight attributes for each. The method of training the computer system may include training the computer system to dynamically perform the selection.

[0133] The reference data representative of previously determined or modelled movements of a reference subject advantageously comprises reference data representative of movements of a reference body part including a reference surface area.

[0134] The reference data representative of the previously determined or modelled movement of the reference subject may comprise reference data representative of a change in a parameter of at least one reference virtual feature associated with the reference surface region caused by the movement. Preferably, the reference virtual feature is located on the reference surface region or at a distance from the reference surface, such as within the volume of the reference body part.

[0135] Advantageously, the reference data representative of the previously determined or modelled movement of the reference subject comprises reference data representative of the movement and / or changes of a plurality of reference virtual features located within the reference subject, preferably at least N reference virtual features, where N is an integer selected from 1 to 10, such as up to 8, such as up to 6, such as 2, 3 or 4.

[0136] The present invention also preferably includes a motion tracking apparatus for tracking the movement of a subject positioned within the scanner bore using the methods as described and claimed herein.

[0137] The motion tracking device an acquisition device configured to acquire a 3D surface representation of a surface area of ​​a subject; a computer system in data communication with the acquisition device for receiving data representative of the 3D surface representation; The computer system is i. receiving data representing a 3D surface representation acquired by a camera arrangement at a first time point and generating a baseline 3D surface representation of a surface area of ​​the subject; ii. associating at least a portion of the baseline 3D surface representation with at least one virtual feature located at the baseline position; iii. receiving data representing the 3D surface representations acquired by the acquisition device at subsequent times to generate subsequent 3D surface representations of the subject's surface region; iv. determining at least one motion tracking parameter, including determining a best match registration of the subsequent 3D surface representation with the constraint to the baseline 3D surface representation; It is structured as follows.

[0138] The acquisition device is configured to acquire a 3D surface representation of the subject's surface area in the form of data representative of the 3D surface representation. The data representative of the 3D surface representation is acquired by acquiring light signals, e.g. reflected light from the subject's surface area, e.g. in the form of an image. The light signals are conveniently converted into digital signals, e.g. a digital point cloud or a digital 3D image.

[0139] In one embodiment, the computer system is configured to control a projector device to project light onto a surface area of ​​the subject and / or to control an acquisition device to acquire a 3D surface representation of the surface area at selected times, preferably at a frequency of at least 1 Hz, such as at least 5 Hz, for example 10-500 Hz, for example 20-100 Hz.

[0140] The projector device may include a projector as described above. The acquisition device may be as described above.

[0141] The computer system may be configured to receive user instructions and / or retrieve instructions from a database, such as instructions related to at least one of criterion related to the scanning procedure to be performed by the scanner, criteria related to the subject, and / or criteria related to the location and / or body part to be scanned.

[0142] Criteria related to a scanning procedure may include, for example, type of scan, scanner specifications, scan session protocol, required scan quality and / or resolution, acquisition method, acquisition method settings, and image acquisition time.

[0143] Subject-related criteria may include, for example, gender, age, weight, body fat level and / or health / mental status.

[0144] Criteria related to the location and / or body part to be scanned may include, for example, the body part and / or body portion of the subject to be scanned.

[0145] The computer system may advantageously be configured to receive user instructions and / or retrieve instructions from a database relating to at least one of the number N of constraints and associated virtual features to be applied, the respective reference parameters of each virtual feature, the number X of DOFs of each constraint, and optionally a respective associated set of weight attributes. Such a selection may be provided, for example, by modeling and / or from previously performed motion tracking, as generated above.

[0146] In one embodiment, the computer system is configured to select at least one of the number N of constraints and associated virtual features to be applied, a respective parameter on which each virtual feature is based, the number X of DOFs of each constraint, and optionally a respective associated set of weighting attributes. The selection may preferably be performed depending on instructions related to at least one of criteria related to the scanning procedure to be performed by the scanner, criteria related to the subject, and / or criteria related to the location and / or body part to be scanned. Thus, for some motion tracking procedures, one single virtual feature and one constraint is sufficient, while for other tracking procedures several virtual features and / or constraints may be desired, e.g. as described elsewhere herein.

[0147] In one embodiment, the computer system is configured to dynamically select at least one of the number N of constraints and associated virtual features to be applied, a respective baseline parameter for each virtual feature, a number X of DOFs for each constraint, and optionally a respective associated set of weight attributes.

[0148] In one embodiment, the computer system is a programmed computer system comprising one or more computers programmed to carry out steps i to iv, preferably to carry out the method as described above.

[0149] The computer system may advantageously include a trained computer that has been trained to select at least one of a number N of constraints and associated virtual features to be applied, respective reference parameters for each virtual feature, a number X of DOFs for each constraint, and optionally respective associated sets of weight attributes.

[0150] The trained computer may advantageously be trained to dynamically select at least one of the number N of constraints and associated virtual features to be applied, respective reference parameters of each virtual feature, the number X of DOFs of each constraint, and optionally respective associated sets of weight attributes.

[0151] The trained computer, in one embodiment, may be an AI computer that has been trained, for example, by unsupervised training.

[0152] In one embodiment, the trained computer is a computer that has been subject to machine learning, preferably a computer that has been trained by the method described above. The trained computer may, for example, include a neural network.

[0153] The trained computer may advantageously be trained to receive user instructions via the interface, the user instructions including at least one of criteria related to the scanning procedure to be performed by the scanner, criteria related to the subject, and / or criteria related to the location and / or body part to be scanned.

[0154] Training the trained computer may advantageously include training the computer to select at least one of: at least one constraint, the number of DOFs of the at least one constraint, weight attributes and / or parameters of the constraint, at least one virtual feature, its position and / or orientation. The computer may be trained to perform such selections and / or decisions.

[0155] The computer may be trained to select at least one of the at least one constraint, the number of DOFs of the at least one constraint, the weight attributes and / or parameters of the constraint, the at least one virtual feature, its position and / or orientation, based at least in part on: i) modeling expected motion of an anatomical model of at least one body part of the subject, including a surface region of the subject; and ii) correlating the motion of the anatomical model with the motion of the surface region, the motion of the potential virtual features, and the motion of the body part being scanned. The potential virtual features may include preselected potential virtual features. In one embodiment, the motion of the anatomical model is supplemented or replaced by data of the subject's motion observed in a previous scanning procedure.

[0156] Further details of the method for training the computer / computer system are provided in the description of the figures and in the description of the examples and elements thereof.

[0157] All features of the invention as described herein, including ranges and preferred ranges, and embodiments of the invention may be combined in various ways within the scope of the invention, unless there is a specific reason not to combine such features.

[0158] A brief description of preferred embodiments and elements of the present invention.

[0159] The above and / or additional objects, features, and advantages of the present invention will become more apparent from the following illustrative and non-limiting description of embodiments, examples, and elements of the present invention, with reference to the accompanying drawings.

[0160] The drawings are schematic, are not drawn to scale and may be simplified for clarity, and the same reference numbers are used throughout for identical or corresponding parts. [Brief description of the drawings]

[0161] [Figure 1] 1 illustrates a schematic diagram of one embodiment of a motion tracking device configured to track a subject positioned within a scanner bore. [Diagram 2] 1 shows a schematic of the scan bore of the scanner as seen in end view. [Diagram 3] 3 shows a schematic side view of a variation of the scan bore of the scanner of FIG. 2; [Figure 4A] 4 illustrates a head tracking session of an embodiment of the method according to the invention; [Figure 4B] 4 illustrates a head tracking session of an embodiment of the method according to the invention; [Figure 4C] 4 illustrates a head tracking session of another embodiment of the method according to the invention; [Diagram 5] 1 shows the different potentially determined head movements. [Figure 6] 13 shows respective tracking sessions of the knee region applying different virtual features and constraints. [Figure 7] 13 illustrates a body part tracking session with and without constraints. [Figure 8] 13 illustrates constrained and unconstrained body part tracking sessions with multiple virtual features applied. [Figure 9] FIG. 1 is a process diagram of one embodiment of a method according to the present invention. [Figure 10] FIG. 2 is a process diagram of an embodiment of a computer system training method according to the present invention. [Figure 11]1 is an exemplary cross-sectional side view of a body part in the form of a head during a portion of a tracking session. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0162] The motion tracking device shown in FIG. 1 includes a controller 2, a computer 1, a first fiber optic bundle 6 having a first lens assembly 6a, and a second fiber optic bundle 7 having a second lens assembly 7a.

[0163] The control unit may include a control computer, not shown. The computer system of the motion tracking device may therefore include the control computer, the computer 1, and further computing units or digital memories arranged for data communication with the control computer and / or the computer 1. The motion tracking device includes a projector device for projecting light onto a surface area of ​​the subject, not shown. The projector may include a light source, for example included in the control unit 2, and a first optical fiber bundle 6 having a first lens assembly 6a arranged to project light onto the surface area. A frame 9a may be applied to fix the position between the first lens assembly 6a and the second lens assembly 7a and / or between the distal end of the first optical fiber 6 and the distal end of the second optical fiber 7, and optionally to fix to the head coil 9.

[0164] The motion tracking device includes an acquisition device configured to acquire a 3D surface representation of a surface area of ​​the subject, i.e. in the form of data representative of the 3D surface representation as described above. The acquisition device includes a second optical fiber bundle 7 having a second lens assembly 7a. Optical signals are collected via the second lens assembly 7a and relayed via the bundle of optical fibers 7 to an optical sensor device, e.g. a camera, located within the control unit 2.

[0165] The scanner bore 8 can be the scanner bore of any type of scanner. In one embodiment, the scanner bore 8 is an MR scanner bore, which includes a permanent magnet 3 within a scanner housing 3a that forms the scanner bore 8. The scanner includes a head coil 9 for scanning a subject (not shown) located on a support structure (pillar) 4.

[0166] In the embodiment shown, the shielding wall F is positioned to form a Faraday cage that protects the computer 1 from magnetic fields inside the scanner bore 8. Data lines, e.g. optical extenders 5, are arranged to transfer data substantially noiselessly (preferably in the range of 20 Hz to 20 kHz, with less than 50 dB or less than 30 dB noise) between the control unit 2 and the computer 1 outside the scanner bore.

[0167] In the embodiment shown, the control unit is located in a scanner room, which is bounded by a surrounding wall F, here shown to the left of the control unit 2 by wall 52. The device 2 may then be conveniently surrounded by a shielded enclosure, not shown, which serves as a radio frequency shielding box. The enclosure may be made of a frame, e.g. a wooden frame, covered with a 1 mm copper layer. Capacitor filters may be arranged to ensure that electromagnetic noise due to powering the components inside the enclosure does not propagate along the power cable. The power source is optionally a separate, not shown, power source.

[0168] Alternatively, the control unit 2 may be located outside the scanner room defined by the peripheral wall F, if the optical fibers 16, 20 are long enough.

[0169] In a further variation, the scanner is a PET scanner or a combined MR / PET scanner.

[0170] The scanner bore shown in Figure 2 corresponds to the scanner bore in Figure 1. In this embodiment, only a portion of the tracker is shown, i.e., a first fiber optic bundle 6 having a first lens assembly 6a arranged to project light onto a surface area, and a second fiber optic bundle 7 having a second lens assembly 7a for acquiring a light signal reflected from the surface area of ​​the subject.

[0171] In this embodiment, the first lens assembly 6a and the second lens assembly 7a are spaced apart to provide an angle to the projected light and the acquired light signals. The distal ends of each of the first fiber optic bundle 6 and the second fiber optic bundle 7 are fixed to the head coil 9.

[0172] A subject 10 , for example a patient, lies on the support 4 with a body part, here the head 10 a , located under the head coil 9 .

[0173] The scan bore of the scanner in FIG. 3 is seen from the side, with a portion of the scanner coil cut away to view a subject 20 positioned on the post 14.

[0174] The coil 19 is arranged to act as an antenna for receiving radio frequency signals from the relevant body part during scanning. A first optical fiber bundle 16 with a first lens assembly 16a arranged to project light onto the surface area is removably or adjustably fixed to the coil 19 at a desired position. Similarly, a second optical fiber bundle 7 with a second lens assembly 7a for acquiring light signals reflected from the subject's surface area is removably or adjustably fixed to the coil 19 at a desired position. For example, depending on the body part and / or the surface area of ​​the subject to be scanned, e.g. the surface area of ​​the body part constituting the body part to be scanned, the surface area to be subject to motion tracking can be changed or adjusted in a simple manner by changing the position of the first lens assembly 16a and / or the second lens assembly.

[0175] 4A shows a tracking session in which the body part is the subject's head H and the body part being scanned is the brain. The head H is shown in a side cross-sectional view.

[0176] In this example of the method, a baseline 3D surface representation is determined by generating a baseline 3D surface representation 21a of the subject's surface area at a first time point (T(0)). The surface region includes the subject's nasal bridge. The virtual point P is selected by the computer or operator to be located at the center of the head. The virtual point P is associated with the baseline 3D surface representation 21a. In this example, the virtual point P is associated with an ROI in the form of a selected position 22a of the baseline 3D surface representation 21a. The position 22a is selected to provide a vertical distance line d to the virtual point P. The distance line d has a length and an orientation that is vertical in this embodiment. In variations thereof, the distance line may be non-vertical, for example having a selected angle relative to the vertical, and / or the virtual point P may be associated with an ROI of the baseline 3D surface representation 21a that includes multiple positions of the baseline 3D surface representation 21a, each position having an orientation and distance relative to the virtual point P.

[0177] 4A, the vertical orientation is along the y direction of a 3D coordinate system having axes X, Y, and Z. In a variation of the above-described exemplary embodiment, the virtual point P is associated with an ROI in the form of a selected location 22a of the baseline 3D surface representation 21a, the location 22a being selected to provide a distance line d along the Y axis to the virtual point P, a distance not shown along the X axis to the virtual point P, and a distance not shown along the X axis to the virtual point P.

[0178] At a subsequent time point (T(s)), the method includes generating a subsequent 3D surface representation of the surface region 21b. By association between the virtual point P and the position 22a of the baseline 3D surface representation 21a, a corresponding virtual point P' associated with the corresponding ROI can be determined in the form of a corresponding position 22b of the subsequent 3D surface representation 21b. It can be seen that the virtual point P' of the subsequent 3D surface representation 21b is shifted by a shift 23, which can be described as a relative distance along the X, Y and Z axes, with respect to the virtual point P of the baseline 3D surface representation. Here, only the shifts along the Y and X axes are shown. One or more constraints are according to the method applied to suppress spurious movements and / or detection errors, thereby obtaining a more accurate best match registration of the subsequent 3D surface representation 21b to determine a more accurate movement of the head H.

[0179] In this embodiment, the constraints may advantageously include limitations on the shift of the virtual point P to the subsequent virtual point P' along one or more of the X, Y and Z axes (1-3 3DOF). The selected constraints may be selected, as described above, preferably based on knowledge of anatomically plausible and anatomically improbable movements of the body part in question, here the head H. This knowledge may be generated by modeling procedures, observation of the subject's movements, and / or previous tracking procedures. In this embodiment, the constraints may advantageously include limitations on the shift 23 along the X direction, since the movement of the head H along the X direction is generally limited. Thus, the limitations on the shift 23 along the X axis are advantageously selected to be rather tight, such as a 50% or higher limitation, for example a 75-90% limitation. Furthermore, the limit of the shift 23 along the Y axis is advantageously selected to be a function of the limit of the shift 23 along the Y axis "y", e.g. the limit is such that the shift along the Y direction is limited only if y exceeds a selected threshold T(Y), and only the part of y that exceeds the threshold is limited by a selected percentage (PRC). That is, it can be provided that the limited shift R(Y) along the Y axis (i.e. after applying the limit) is: y < T(y) = no limit; y > T(Y), R(Y) = T(Y) + PRC × (y - T(Y)).

[0180] The virtual point Pc associated with the corresponding ROI of the display of the optimal match of the surface area 21c is shown in Figure 4B.

[0181] In this example of the method, the baseline 3D surface display is determined by generating a baseline 3D surface display 31 of the surface area of the subject at the first time point (T(0)). The surface area includes the bridge of the subject's nose. The virtual feature is selected as a volumetric virtual feature, i.e., a volume V that is inscribed in a circle 33a and has a center corresponding to the center of the head H.

[0182] The virtual volume V is associated with the baseline 3D surface display 31 by a circle 33a that intersects the baseline 3D surface display 31 of the surface area of the ROI in the form of two positions 32a, 32b.

[0183] At a subsequent time point (T(s)), the method includes generating a subsequent 3D surface display of the surface area (not shown). By associating the virtual volume V with the position of the circle 33a that crosses the baseline 3D surface display 31, the position of the volume V' and the position of a circle 32b that inscribes the volume V' associated with the corresponding ROI of the subsequent 3D surface display of the surface area can be determined. It can be seen that the virtual volume V' is shifted by a shift dV = V' - V with respect to the virtual volume V. The shift can be seen to include displacements along at least the Y-axis and the X-axis.

[0184] One or more constraints are applied according to a method for suppressing false movements and / or detection errors, thereby obtaining a more accurate optimal match registration of the subsequent 3D surface display and determining a more accurate movement of the head H.

[0185] The constraints may advantageously have three DOFs, such as, for example, constraints by the respective weights of the shifts along each axis (X, Y, Z).

[0186] Figures 5(a)-5(c) show different potentially determined head movements.

[0187] In Figure 5(a), the subject is resting his head H on the support 44 in a highly anatomically sensible position, and therefore any subsequent 3D surface representations of the surface region showing movement to such a position may be deemed sensible and may not require constraints. Reference numeral 41 denotes a surface region that is conveniently tracked to generate the baseline 3D surface representation and subsequent 3D surface representations of the surface region.

[0188] FIG. 5(b) shows an anatomically impossible position of the subject's head H. In particular, note that the head bag region 45 passes under the strut 44. This is only possible if the strut is highly compressible or elastic. A slight depression of the strut is usually appropriate. Any movement to a position such as that shown in FIG. 5(b) may be deemed impossible, and subsequent 3D surface rendering of surface regions showing movement including such positions may contain spurious movements and / or detection errors. By applying one or more constraints such as those described above, such spurious movements and / or detection errors may be suppressed.

[0189] FIG. 5(c) illustrates another anatomically impossible position of the subject's head H. In particular, it should be noted that the neck region 46 of the head is tilted in an orientation that would be impossible for most subjects lying in the scanner bore. Depending on the space in which the head is located during the scan, less movement of the neck region is plausible. Any movement to a position such as that shown in FIG. 5(c) may be determined to be impossible, and subsequent 3D surface displays of surface regions that exhibit movement including such positions may contain spurious movements and / or detection errors. Applying one or more constraints such as those described above may suppress such spurious movements and / or detection errors.

[0190] 6(a) illustrates a tracking session in which the body part is a subject's knee region 50 and the body part being scanned is the internal soft tissue structure of the knee joint. For accuracy, the tracked surface area includes a surface portion of the knee joint, a surface portion of the femur, and a surface portion of the tibia. As can be seen, the joint region 55 is shown as a circular volume, with the femur 54 and tibia extending from the joint region 55.

[0191] In this example of the method, a baseline 3D surface representation is determined by generating a baseline 3D surface representation 51 of a surface area of ​​the knee region at a first time point (T(0)). A virtual point P is selected by the computer or the operator to be located at the center of a circle representing the knee joint 55. The virtual point P is associated with an ROI of the baseline 3D surface representation 51. The ROI advantageously includes at least a point of the surface portion of the knee joint, a point of the surface portion of the femur, and a point of the surface portion of the tibia. The virtual point P can be associated with the ROI of the baseline 3D surface representation by the method as described above.

[0192] At a subsequent time point (T(s)), the method includes generating a subsequent 3D surface representation 52 of the surface region. Association of the virtual point P with the ROI of the baseline 3D surface representation 51 allows the position of a corresponding virtual point P' to be determined that is associated in a corresponding manner with the corresponding ROI of the subsequent 3D surface representation of the surface region.

[0193] It can be seen that the virtual point P′ of the subsequent 3D surface representation 52 is shifted relative to the virtual point P of the baseline 3D surface representation. One or more constraints are according to the method applied to suppress spurious movements and / or detection errors, thereby obtaining a more accurate best match registration of the subsequent 3D surface representation 21 b to determine a more accurate movement of the head H.

[0194] FIG. 6(b) shows another tracking session, where the body part is the subject's knee region 50 as in the example of 6(a).

[0195] The body part being scanned may be the soft tissue structure inside the knee joint. To increase accuracy, the surface area tracked includes a surface portion of the knee joint, a surface area of ​​the femur, and a surface area of ​​the tibia. As can be seen in the figure, the joint area 55 is shown as a circular volume, with the femur 54 and the tibia extending from the joint area 55.

[0196] In this example of the method, a baseline 3D surface representation is determined by generating a baseline 3D surface representation 61 of the surface area of ​​the knee region at a first time point (T(0)). At least three virtual points P0, P1, P2 are selected by a computer system or an operator to be located at the center position (P1), the center position (P2) of the femur, and the center position (P0) of the tibia, respectively, of a circle inscribed in the knee joint 55.

[0197] Each virtual point P0, P1, P2 may be associated with an ROI of the baseline 3D surface representation, advantageously including at least a point ROI-1 on a surface portion of the knee joint, a point ROI-0 on a surface portion of the femur, and a point ROI-2 on a surface portion of the tibia, with P0 associated with ROI-0, P1 associated with ROI-1, and P2 associated with ROI-2.

[0198] At a subsequent time point (T(s)), the method includes generating a subsequent 3D surface representation 62 of the surface region. By association between each virtual point P0, P1, P2 and the baseline 3D surface representation 61, the location of each corresponding virtual point (not shown) is associated in a corresponding manner with a corresponding ROI, including corresponding points ROI-0, ROI-1 and ROI-2, of the subsequent 3D surface representation of the surface region.

[0199] One or more constraints may be applied to suppress spurious motion and / or detection errors, so that a more accurate best match registration of the subsequent 3D surface representation 62 can be obtained to determine more accurate motion of the knee region. The applied constraints may advantageously include keeping the distances from P0 to P1 and from P1 to P2, respectively, constant or limiting the change in the distances to a minimum.

[0200] 7(a)-7(c) show body part tracking sessions with and without constraints.

[0201] 7(a) shows a baseline 3D surface representation of the surface region generated at T(0) and a subsequent 3D surface representation of the surface region generated at T(1) prior to the best match step. The subsequent 3D surface representation includes local moment data for movements that do not correspond to any movement of the body part being scanned, e.g., soft tissue movements.

[0202] In Fig. 7(b), the best match registration of the subsequent 3D surface representation of the surface region without any constraints is shown. Because this tracking involves tracking without suppressing spurious motion, it may result in unnecessary adjustments of scan parameters (resulting in poor quality or wasteful scans) and / or unnecessary stopping and / or restarting of scans.

[0203] The best match registration of the subsequent 3D surface representation of the surface region with at least one constraint is shown in Fig. 7(c). The tracking results show that spurious motion is suppressed, ensuring high tracking quality.

[0204] 8(a)-8(c) show tracking sessions of a body part with and without constraints, the body part including a bone structure B, which is under the skin region and contributes to shaping the subject's surface area in the scanner. The bone structures may include, for example, joint regions or facial regions.

[0205] FIG. 8(a) shows a baseline 3D surface representation of the surface region generated at T(0); and a subsequent 3D surface representation of the surface region generated in T(1) prior to the best match step. The subsequent 3D surface representation includes local motion data, e.g. motions that do not correspond to any motion of the body part being scanned, e.g. soft tissue motions. This motion may be, for example, motions caused by facial muscles, and there may be no or only little motion of the bone structure B.

[0206] The bone structure B may be known or estimated from a bone structure model and / or from one or more previously acquired or simultaneously acquired images, such as ultrasound images, X-ray images, and / or MR images.

[0207] The method includes, in this embodiment, selecting a number of virtual features in the form of virtual points P located on the bone structure surface portion, and associating each virtual feature with a position of the baseline 3D surface representation at a respective position corresponding to the shortest distance between the virtual feature P and a position of the baseline 3D surface representation. The distance between each virtual point and a respective associated point of the ROI of the baseline 3D surface representation (ROI) is called a distance line, and these distance lines are illustrated here as tension / compression springs D1, D2, D3.

[0208] FIG. 8(b) shows the best match registration of the subsequent 3D surface representation of the surface region without constraints. The virtual point p is associated with the respective corresponding point (cROI) in the corresponding ROI of the subsequent 3D surface representation of the surface region. The best match registration without constraints is here a "rigid" registration. Here, distance lines d1, d2, d3 without constraints are shown. Since this tracking involves tracking without suppressing spurious motion, the scanning parameters may be unnecessarily adjusted since the body part being scanned may not be moving or may only be moving slightly, whereas the surface region may be undergoing much larger motion. This may result in poor or wasted scan quality and / or unnecessary stopping and / or restarting of the scan.

[0209] In Fig. 8(c), a (flexible) best match registration of the subsequent 3D surface representation of the surface region with at least one constraint including limiting the length of each of the distance lines (also called tension / compression springs) D1, D2, D3 is shown, where spurious motion is suppressed and high quality tracking is ensured.

[0210] The limits of length of each of the distance lines (spring extension / compression) D1, D2, D3 may advantageously be such that at least two of the limits are different from each other.

[0211] The distance lines are considered here as tension / compression springs D1, D2, D3 having an initial length (L(0)) determined with respect to the baseline 3D surface representation, and the change in each spring length is limited by a value k, which may be a constant or a function. For example, the tension / compression springs D1, D2, D3 may be limited by respective values ​​k1, k2, k3, where at least one of the values ​​k1, k2, k3 may differ from at least one other of the values ​​k1, k2, k3.

[0212] In one embodiment, one or more of the k values ​​k1, k2, k3 are functions that depend on another distance line (d1, d2, or d3) for a subsequent 3D surface representation without surface area constraints. For example, the k values ​​k1 and k3 of tension / compression springs D1 and D3, respectively, may depend on the unconstrained distance line d3, and / or the k2 value of tension / compression spring D2 may depend on one or both of the distance lines d1 and d3.

[0213] The process diagram shown in FIG. 9 illustrates an embodiment of a method according to the invention, in which a procedure of scanning a body part and a procedure of scanning a surface area of ​​the body part comprising the body part are performed.

[0214] In step 71, a patient (subject) is positioned on and supported by a support. The support is located within the bore of a scanner, such as an MR scanner, a CT scanner, or any other scanner mentioned above. Step 72 indicates that an acquisition device for acquiring reflected light data for generating a 3D surface representation of the surface region is or has been positioned at a desired position and orientation relative to the surface region of the body part. In addition, a projector device for imaging the surface region or for projecting optionally structured light (e.g., a light pattern) onto the surface region may be or has been positioned at a desired position and orientation relative to the surface region of the body part. Steps 71 and 72 may be in any order.

[0215] In step 73, the tracker generates a baseline 3D surface representation at T(0).

[0216] In step 74, the tracker generates a subsequent 3D surface representation at T(s1).

[0217] In step 75, the tracker determines a best matching registration of the subsequent 3D surface representation of T(s1) having at least one first constraint including a relative limit.

[0218] In step 76, the tracker generates a further subsequent 3D surface representation at T(s2).

[0219] In step 77, the tracker determines a best matching registration of the subsequent 3D surface representation of T(s2) having at least one first constraint including a relative limit.

[0220] Step 78 indicates that the tracker iterates generating subsequent 3D surface representations in T(s3...n) and determining the best matching registration having at least the first constraint.

[0221] In that variation, the device may generate a "new" baseline 3D surface representation after a selected tracking time and / or in the event of an event such as restarting the scan and / or detection of a scan error and / or spurious motion above a threshold.

[0222] In another variation, the device may adjust the weight of the first constraint depending on a detected shift between a virtual feature associated with a ROI of the baseline 3D surface representation and a corresponding virtual feature associated with a corresponding ROI of the subsequent 3D surface representation that is to be subject to best match registration.

[0223] In a further variant, the device may apply one or more further constraints to the determination of the best match registration of the subsequent 3D surface representation if a detected shift between a virtual feature associated with a ROI of the baseline 3D surface representation and a corresponding virtual feature associated with a corresponding ROI of the subsequent 3D surface representation to be subject to best match registration exceeds a threshold value.

[0224] FIG. 10 illustrates an embodiment of a method for training a computer system of the method according to the invention.

[0225] In step 81, a number of motion data sets are generated, for example using any of the methods described above. Advantageously, the data sets include motion tracking data sets to which constraints are applied, preferably including six DOF constraints. The data sets may further include unconstrained data sets.

[0226] Alternatively, some or all of these motion data sets are generated using any of the other constraints mentioned above, or without applying any constraints.

[0227] Each data set includes a set of attributes representing any constraints (preferably including one or more of weights, virtual features, DOF, etc.), criteria related to the scanning procedure, and / or criteria related to the subject.

[0228] The criteria may include, for example, one or more of the criteria discussed above.

[0229] In step 81a, a number of additional datasets are advantageously provided, including a modelled dataset, a dataset from images, and / or any other dataset.

[0230] Each data set of step 81a may advantageously include one or more attributes associated with the scanning procedure and / or criteria associated with the subject.

[0231] The data sets of step 81 and / or step 81a may be subjected to a filtering process in step 82. The filtering process may advantageously include removal of outlier data, removal of data representing anatomically unrealistic movements / positions, removal and / or restoration of blurred data, removal or suppression of noise, size reduction (e.g. removal of irrelevant data portions), etc.

[0232] The data set is then sent to a computer system for machine learning, step 83, and at least one machine learning algorithm is generated by the computer system. The computer system is then trained and can be applied for use in the methods as described herein. The trained computer system can be further improved, for example, by being trained on additional data sets generated by use of the computer system.

[0233] In step 84, the actual tracking data set is transmitted to the trained computer system. The actual tracking data may be real-time tracking data. The actual data set includes at least a data set (subsequent data set) representing a baseline 3D surface representation of the subject's surface area at a first time point (T(0)) and a data set (subsequent data set) representing a subsequent 3D surface representation of the subject's surface area at a subsequent time point (T(s)), and the data may be transmitted to the computer system in real-time as it is acquired.

[0234] Step 85 illustrates the processing of a trained computer, which uses machine learning algorithms to perform predictive modeling and determine and apply one or more constraints as described above to limit scanning errors and / or spurious motion. In addition to the actual tracking data set, the computer system may receive and / or obtain (e.g., from a database) data representing criteria related to the scanning procedure performed by the scanner, criteria related to the subject, and / or criteria related to the location and / or body part to be scanned.

[0235] Step 86 shows a determined best match registration with one or more constraints, preferably determined in real time by a trained computer system. The best match registration for each subsequent data set may be transmitted, for example, to a scanner control and / or display, for example, a screen that displaces in the same manner as the actual movement of the subject located within the scanner.

[0236] FIG. 11 is an example cross-sectional side view of a body part in the shape of a head during a portion of a tracking session.

[0237] The at least one virtual feature is selected to include a point cloud that includes points at locations determined by a Gaussian function, for example, near a central location of the head.

[0238] As indicated by the dotted lines, a number of points of the point cloud (point cloud) are associated with the baseline 3D surface representation at respective points (ROI points) of the set of points of the ROI of the baseline 3D surface representation.

[0239] At a subsequent time point, a subsequent 3D surface representation of the surface region is obtained, and each point of the point cloud (point cloud) is associated with a respective corresponding ROI point (cROI point) of a set of points of a corresponding ROI of the subsequent 3D surface representation of the surface region. The change in distance from each ROI point to each point of the point cloud (from each ROI point to each of the point cloud) and the change in distance from each corresponding ROI point of the subsequent 3D surface representation to each point of the point cloud (from each cROI point to each of the point cloud) are subject to the respective constraints, thereby arriving at a best-matching subsequent 3D surface representation as shown, with the corresponding ROI point in best match.

Claims

1. 1. A method of tracking the movement of a subject positioned within a scanner, the method comprising: generating a baseline 3D surface representation of a surface area of ​​the subject at a first time point (T(0)); generating a subsequent 3D surface representation of the surface region of the subject at a subsequent time point (T(s)); determining a best-match registration of the subsequent 3D surface representation with at least one constraint to the baseline 3D surface representation; - determining at least one motion tracking parameter; Equipped with the constraints include relative limits to provide suppression of spurious motion and / or detection errors, the relative limits being a function of a difference between each position of the baseline 3D surface representation and a corresponding each position of the subsequent 3D surface representation. method.

2. The method of determining the best matching registration includes selecting at least one virtual feature and associating the at least one virtual feature with a region of interest (ROI) of the baseline 3D surface representation; the constraints include limiting a change in at least one parameter of the at least one virtual feature associated with a corresponding ROI of the best-matching subsequent 3D surface representation relative to the at least one parameter of the at least one virtual feature associated with the ROI of the baseline 3D surface representation. The method of claim 1.

3. the at least one parameter comprises a position parameter, an orientation parameter, a range parameter, a distance parameter, and / or a combination comprising at least one of the foregoing parameters, of the at least one virtual feature associated with the ROI; The method of claim 2.

4. the at least one virtual feature comprises at least one of a virtual point, a virtual volume, a virtual region, a virtual line, a virtual bone structure, and / or any combination comprising one or more of the aforementioned virtual features; The method of claim 2.

5. the ROI is a subregion or set of subregions of the baseline 3D surface representation that corresponds to an actual subregion of the surface region, or corresponds to an actual subregion of a portion of the subject that at least partially correlates with an actual subregion of the surface region; The method of claim 2.

6. the constraints include providing that the at least one virtual feature associated with the corresponding ROI of the subsequent 3D surface representation for which the best match has been registered is transformed compared to a case in which the at least one constraint is not present; the transformed virtual feature comprises a transformation of the at least one parameter; the transformation includes at least one restriction of the at least one parameter. The method of claim 2.

7. The constraint has at least one degree of freedom (DOF) selected from a translational axis and a rotational axis. The method of claim 1.

8. the constraint is an X DOF constraint, X is an integer from 1 to 6; The method of claim 1.

9. the constraint is associated with at least one weight attribute representing at least one weight value of the constraint; The method of claim 1.

10. The constraint is associated with a set of weight attributes having at least two weight attributes each comprising a weight value, and the weight of each weight attribute of the set of weight attributes is derived from modeling expected movement of the at least one virtual feature associated with at least one ROI of the baseline 3D surface representation due to movement of an anatomical model of the subject's body part including the surface region. The method of claim 2.

11. - the value of each weight attribute of the set of weight attributes is dynamically adjusted, preferably the value of each weight attribute of the set of weight attributes is dynamically adjusted depending on the subsequent 3D surface representation of the surface region. The method of claim 10.

12. the at least one virtual feature associated with the ROI of the baseline 3D surface representation is spatially positioned at the surface region or at a greater distance from the surface region than an acquisition device that acquires reflected light from the surface region to generate each of the baseline 3D surface representations; the at least one virtual feature associated with the ROI of the baseline 3D surface representation is spatially located between the surface region and a support supporting the subject. The method of claim 2.

13. the surface area includes a surface area of ​​the subject's body part; the virtual features include virtual points, virtual lines, or virtual bone structures located inside the volume of the body part at the first time point; The parameters include position parameters, orientation parameters and / or distance parameters. The method of claim 2.

14. the surface area includes a surface area of ​​the subject's body part; the virtual features include a virtual volume and / or a virtual region located at least partially within the surface region and / or volume of the body part at the first time point; the parameters include position parameters, orientation parameters and / or range parameters; The method comprises: i) selecting the at least one virtual feature having a feature space location at the first time point T(0) by providing an estimated movement of the subject during a scan section; and ii) selecting the at least one virtual feature at the first time point T(0) having the feature space location that is subject to less parameter variation than other virtual features of the subject and / or is subject to parameter variation below a preselected level. The method of claim 13.

15. the method includes selecting a plurality of constraints for one or more virtual features, including selecting at least one set of the weight attributes for each constraint; the weight of each constraint is selected depending on the parameters of the constraint, the baseline parameters; The method of claim 2.

16. The method comprises: providing a trained computer including a method for training a computer to select the at least one virtual feature associated with the ROI of the baseline 3D surface representation and the at least one constraint using a set of reference data, each reference data set including reference data representing previously determined or modeled movement of a reference subject correlated to reference data representing determined or modeled movement of a reference surface region; the reference surface area is a surface of a reference body part of the reference subject; The method of claim 12.

17. The reference data representing previously determined or modeled movements of a reference subject includes reference data representing changes in parameters of at least one reference virtual feature associated with the reference surface area caused by the movements, the at least one reference virtual feature being located on the reference surface area or within the volume of a reference body part.

17. The method of claim 16.

18. the surface area is a surface area of ​​a body part that includes the body part being scanned; the at least one virtual feature includes a cloud of points located inside the body part at positions determined by a Gaussian function around a center position of the body part; the ROI of the baseline 3D surface representation corresponds to a set of locations in the surface region; at least a set of points of the point cloud are associated with the ROI of the baseline 3D surface representation; The method of claim 2.

19. 1. A motion tracking device for tracking the movement of a subject located within a scanner, said motion tracking device comprising: an acquisition device configured to acquire a 3D surface representation of a surface area of ​​the subject; a computer system in data communication with said acquisition device to receive data representing said 3D surface representation; Equipped with The computer system includes: i. receiving data representing a 3D surface representation acquired by the camera device at a first time point and generating a baseline 3D surface representation of the surface region of the subject; ii. Associating at least one virtual feature with a ROI of the baseline 3D surface representation; iii. receiving data representing 3D surface representations acquired by the acquisition device at subsequent times to generate subsequent 3D surface representations of the surface region of the subject; iv. determining at least one motion tracking parameter, including determining a best match registration of the subsequent 3D surface representation with the constraints to the baseline 3D surface representation; configured to: the constraints include limiting a change in at least one parameter of the at least one virtual feature associated with a corresponding ROI of the best-matching subsequent 3D surface representation relative to the at least one parameter of the at least one virtual feature associated with the ROI of the baseline 3D surface representation; The computer system is configured to perform the method of claim 1. Motion tracking device.

20. The motion tracking device includes a trained computer; The trained computer is trained to select at least one of: i) a number N of constraints and associated virtual features to be applied; ii) baseline parameters of each of the virtual features; and iii) a number X of DOFs for each of the constraints.

20. The motion tracking device of claim 19.

21. the computer system is configured to receive user instructions and / or retrieve instructions from a database regarding at least one of: i) criteria associated with a scanning procedure to be performed by the scanner; ii) criteria associated with the subject; and / or iii) criteria associated with a location and / or body part to be scanned; and / or configured to receive user instructions and / or retrieve instructions from a database regarding at least one of: i) a number N of constraints and associated virtual features to be applied; ii) a baseline parameter for each of said respective virtual features; iii) a number X of DOFs for each of said constraints; and iv) at least one weight attribute associated with said constraints; 20. The motion tracking device of claim 19.

Images