EP2755556A1

EP2755556A1 - Nuclear imaging system and method for updating an original nuclear image

Info

Publication number: EP2755556A1
Application number: EP12791108.9A
Authority: EP
Inventors: Thomas Wendler
Original assignee: SURGICEYE GmbH
Current assignee: SURGICEYE GmbH
Priority date: 2011-09-16
Filing date: 2012-09-17
Publication date: 2014-07-23
Also published as: WO2013038011A1; US9286732B2; US20140369560A1; DE102011053708A1; JP2014530348A

Abstract

A nuclear imaging system for updating an original nuclear image (10), the nuclear imaging system comprising: a data memory for storing the original three-dimensional nuclear image (10); a nuclear radiation detector (50) which is movable along a freely variable path and is intended to measure nuclear radiation in order to obtain nuclear radiation values; a tracking system (60) for tracking the nuclear radiation detector (50) while measuring the nuclear radiation, thus obtaining detector coordinates which indicate a posture of the tracked nuclear radiation detector in relation to an image coordinate system of the nuclear image; a nuclear data input (70) configured to receive the nuclear radiation values from the nuclear radiation detector (50) and the detector coordinates from the tracking system (60) and to associate the nuclear radiation values with the respective detector coordinates; and an image update module (20) with an update rule for changing the original nuclear image (10) on the basis of the nuclear radiation values and the detector coordinates, wherein the image update module is configured to produce an updated three-dimensional nuclear image (18) by applying the update rule to the original nuclear image.

Description

description

NUCLEAR IMAGE SYSTEM AND METHOD FOR UPDATING AN ORIGINAL NUCLEAR IMAGE

The invention is in the field of nuclear imaging such as PET or SPECT, and aspects of the invention relate to an apparatus for updating an original nuclear image in response to measurement data of a freely movable tracked nuclear radiation detector. Further aspects of the invention relate to a corresponding method for updating the original nuclear image.

Various methods of nuclear imaging such as PET and SPECT imaging methods are known. These methods allow radiation emitted by a body radioactive sources spatially resolved as a three-dimensional

Nuclear image (radiation distribution image) play. This makes it possible, for example, to determine the distribution of a radiopharmaceutical in the body, in this way

To draw conclusions about the function and / or metabolic processes of various organs or regions in the body. For example, these methods allow a reliable assessment of various cancers and a localization of the affected tissue. This information has a high utility value for surgical procedures, e.g. to remove diseased tissue.

Many of these imaging techniques require a fixedly mounted or fixed trajectory mobile nuclear detector such as in conventional PET or SPECT methods. However, if such a detector should provide sufficient information for a good quality nuclear image, it should at least in principle be able to detect radiation which leaves the person to be imaged in as different a direction as possible. For example, US 2010/74498 AI describes such a nuclear detector in which a patient bed is moved into a PET gantry to generate a nuclear image. However, such a nuclear detector takes up considerable space and makes it difficult to access the person to be imaged. For this reason, such a detector can be used with difficulty, if at all, in an operating environment. However, when used outside, the image is no longer necessarily current, such as due to interim movements of the person being imaged.

Based on a fundamentally different approach so-called Freehand (freehand) nuclear detectors such as Freehand SPECT detectors: Here is a nuclear probe (typically a detector for nuclear radiation with a collimator connected in front of it) freely movable, ie at least locally - in a limited spatial area - movable in all three spatial directions and, as a rule, also rotatable with respect to all three solid angles. These freehand nuclear detectors make it possible to detect radiation from different directions with respect to the person being imaged without being in the way, and can therefore also be used without problem in an operating environment. Such Freehand nuclear detectors are described for example in DE 10 2009 042 712 AI.

Tracking techniques are of particular importance to these freehand nuclear detectors, and these tracking techniques are in part a basis for completely new imaging techniques. Thus, using tracking techniques, various medical imaging techniques such as nuclear spin resonance, SPECT, optical or PET imaging with hand-held probes can also be performed in situations where this would otherwise not be possible. Here, a tracking system is used which allows objects used in the imaging process, such as

To track radiation detectors as targets, that is, to detect their pose (three spatial coordination and two or three angle coordination, which indicate the orientation), in particular to record their time dependence. A medical application is generally referred to herein as an operation, i. Surgical operations in the strict sense are only a part of what is more commonly referred to herein as surgery. In addition, this also means an imaging procedure, or preparatory or supporting actions of a surgical operation, as an operation. In this case, the tracking takes place in an operating environment or a part of it. An operational environment is generally understood to be an environment in its domain

medical applications such as a surgical operation, an imaging procedure, parts of a diagnostic procedure, and the like. For example, the operating environment may be part of an operating room, in particular an area for a patient to be operated on. Even if in the following the application of a

Special attention is paid to surgical operation by a doctor, so others can

Embodiments may also be directed to other medical applications, such as applications for instrument guidance methods or for guided imaging.

The tracked freehand imaging methods, however, take place in part under less controlled conditions compared to the conventional methods. Whether a picture can be obtained with good quality, largely depends on the experience of the user and may also take more time. During this time, the operating environment is occupied in the case of intraoperative imaging.

Against this background, therefore, a nuclear image system for updating an original nuclear image according to claims 1 and 2, and a corresponding method according to claims 10 and 11 is proposed. Further preferred aspects of the invention will become apparent from the dependent claims, from the figures and from the description.

The nuclear image system makes it possible to generate an original nuclear image, without necessarily taking into account boundary conditions, which are to be observed in a directly to be included in an operating environment nuclear image, in particular the accessibility of the operating environment. This original nuclear image may not describe actual changes, such as interim movements of the person to be imaged or within the person being imaged (such as organs in the body), changes in the radiation source (the radiopharmaceutical) in the body, and the like. However, this disadvantage is compensated for by enabling up-to-date measurements of the nuclear radiation with the freely movable nuclear radiation detector and / or enabling current position measurements of the patient. The

Image Refresh module allows the original nuclear image to be modified to reflect this current information and to feed into the updated image.

Herein, as a nuclear image, a radiation distribution image is understood, i. E. a three-dimensional intensity distribution obtained from a nuclear measurement (e.g.

Radiation distribution image, i. reconstructed radiation intensity as a function of the three space coordinates Χ, Υ, Ζ). Also other intensities than a reconstructed one

Radiation intensity can be imaged in a nuclear image, e.g. a

Absorption intensity against nuclear radiation. Even data that yields a three-dimensional intensity distribution after decoding or other real-time processing can be understood as the nuclear image.

The invention also relates to an apparatus for carrying out the disclosed methods and also includes apparatus for carrying out individual process steps. These process steps may be performed by hardware components, through a computer programmed by appropriate software, by a combination of both, or in any other way. The invention is further directed to methods according to which the devices described in each case operate. It includes steps to perform each function of the devices. One possible aspect of the invention also relates to that described herein

Nuclear imaging system for use in intraoperative nuclear imaging.

In addition, the invention with reference to shown in FIGS

Embodiments will be explained, from which further advantageous parts and

Variations result. To show:

FIG. 1 shows a device for 3D acquisition and 3D visualization;

Figure 2 shows components of a nuclear imaging system according to an embodiment of the invention;

Figures 3 to 6 show components of nuclear imaging systems according to further embodiments of the invention;

Figure 7 shows a diagrammatic representation of the operation of a

A nuclear imaging system according to an embodiment of the invention;

Figure 8 shows a diagrammatic representation of the operation of a

Image update module of the nuclear image system of Fig. 7;

FIG. 9 shows a schematic view of a projection image for the

Nuclear image system of Fig. 7;

FIG. 10 schematically shows a method for eliminating

Nuclear image information for the nuclear image system of Fig. 7;

Figure 11 is a flowchart showing steps of a method of updating an original nuclear image according to an embodiment of the invention; and

FIG. 12 shows a diagram which in the method of FIG

Quality value Q as a function of an iteration value x represents.

In the following, various embodiments of the invention will be described, some of which are also exemplified in the figures. In the following description of the figures, like reference numerals refer to the same or similar components. In general, only differences between different Embodiments described. Herein, features described as part of one embodiment may also be readily combined with other embodiments to produce still further embodiments.

Fig. 1 shows parts of a nuclear image system according to a first aspect of the invention, which is capable of updating a stored original nuclear image. The original nuclear image has already been taken in the situation illustrated in Figure 1 and is stored on a data storage, e.g. in the data acquisition system 90.

The original nuclear image is a complete three-dimensional nuclear image which has been reconstructed on the basis of an earlier measurement. The measurement can be done with the system shown in Figure 1; Alternatively, the original nuclear image can also by a separate system (such as by means of a not freely movable, so stationary or movable along a fixed trajectory

Nuclear detector) have been generated. The original nuclear image represents a spatial radiation intensity distribution of a radiation source, such as the radiation intensity as a function of respective voxels. An image coordinate system of the nuclear image is displayed in

Hereafter referred to as KN.

Further, in Fig. 1, a nuclear radiation detector 50 for measuring nuclear radiation is shown. The nuclear radiation detector 50 is movable along a freely variable path, that is to say that it can be moved at least locally in all three spatial directions and, in this case, also with respect to all three solid angles. Other examples of a detector movable along a free-floating path are about one on a controllable one

Robotic arm with sufficient degrees of freedom mounted detector, a freely displaceable during the measurement and height-adjustable detector, and the like.

The nuclear radiation detector 50 comprises a collimator which selectively passes only radiation from a certain angular range and in the beam direction behind it a detector element to obtain nuclear radiation values, more specifically a measured radiation intensity or frequency as a function of time. Optionally, the nuclear radiation detector 50 includes an energy filter to energetically filter nuclear radiation prior to detection.

In general, the nuclear radiation detector may be a detector of dimension 0, 1 or 2. In this case, a detector of dimension 0 is such that it becomes a given time allows the measurement of a single radiation value in the detector volume. An example of this is a single gamma probe, optionally with a collimator mounted in front of it. Dimension 1 means that the detector allows the spatially resolved measurement along a one-dimensional line; and dimension 2 means that the detector allows a spatially resolved measurement in both directions of a 2D surface. An example of a 2D sensor is a gamma camera. An analogy for a 2D sensor in the case of an optical sensor would be for example a CCD sensor. In addition, another

Spatial resolution by moving the detector and time-dependent measurement done.

The detector may be, for example, a nuclear probe, a nuclear camera, a coincidence camera, a composite camera, or a combination thereof.

In general, it is a Freehand detector, and the detector is designed for handling with one hand and has a grip area for gripping and holding the detector by a hand. Optionally, a detector holder may be provided which holds the detector in a defined position even when released.

The nuclear radiation detector 50 may further include a data interface for transmitting the measurement data to a suitable data acquisition system 90, such as a suitable computer. Parts of the nuclear radiation detector 50, for example parts of the detector element, can also be accommodated spatially separated from the remaining nuclear radiation detector 50.

Further, Fig. 1 shows a tracking system 60, which serves to track the nuclear radiation detector 50 during the measurement of nuclear radiation and will be described below. Tracking herein means the time dependent detection of the pose of the nuclear radiation detector 50, i. its spatial location and its orientation. The situation is described by means of three detector coordinates for describing the position in three-dimensional space; and the orientation is done with the help of

usually three detector solid angles are described. In embodiments, however, two solid angles are sufficient, namely when a rotation of the detector about its axis anyway has no effect on the measurement, as is the case with some 0-dimensional detectors.

The tracking system shown in Fig. 1 is an optical tracking system. For this system, the nuclear radiation detector 50 is an optical Tracking target object 62 fixed with optical markings. Based on the relative

Positions of the markings by means of a stereoscopic camera 64a, 64b

The position and orientation (pose) of the nuclear radiation detector 50 can be estimated with high precision in real time. This pose can be expressed in a detector coordinate system, hereinafter referred to as KD.

The tracking system is equipped to detect and indicate the pose of the nuclear radiation s detector 50 (coordinate system KD) in relation to the original nuclear image (coordinate system KN). For this purpose, the tracking system is adapted to establish a spatial relationship between the systems KD and KN. In the

Embodiment of Fig. 1, this is accomplished by attaching another tracking target 82 to the person 84 being imaged.

The position of the tracking target 82 in the nuclear image coordinate system KN is known (e.g., by calibration) or calculated by registration at the beginning of the procedure. For example, already during the

Recording of the original nuclear image, the tracking target object 82 to be attached to the person 84, and the nuclear image detector used with another

Tracking target object (analogous to the Nachführzielobjekt 62) be equipped to the

Tracking target object 82 in the nuclear image coordinate system KN, that is to establish a relation between the position of the tracking target object 82 and the nuclear image coordinate system KN. More generally, registration may be done as point-based registration, surface-based registration, image-based registration (such as with the aid of an image recognition that matches recognizable landmarks in the image), or a combination thereof.

By the pose of Nachführzielobjekts 82 is now detected by the tracking system 60 in the detector coordinate system KD, the context

(Coordinate transformation) between KD and KN are determined in a manner known to those skilled in the art. Because of this relationship, the detector coordinates also reflect the pose of the tracked nuclear radiation detector 50 in relation to that

Image coordinate system KN of the nuclear image. For this purpose, it is not necessary that the detector coordinates are actually converted into coordinates of the nuclear image coordinate system KN; it suffices the presence of one - in this case by the

Follow-up target object 82 produced - comprehensible relation or transformation between KD and KN. In this embodiment, the stereoscopic camera 64a, 64b may optionally be moved, also time-dependent, and need not be tracked as long as both tracking targets 62 and 82 are in view, knowing the pose of these two tracking targets 62 and 82 in FIG an arbitrary common coordinate system is sufficient to calculate the pose of the detector 50 in relation to the nuclear image coordinate system KN (definable by means of tracking target object 82), eg as relative pose or

Difference pose. Alternatively, the stereoscopic camera can fix on one of the

Tracking object be bound.

Like the measurements of the nuclear probe 50, the tracking information (time-dependent pose) is also sent to the data acquisition system 90 with a data interface (not shown). The data acquisition system 90 has a nuclear data input for receiving the nuclear radiation values from the nuclear radiation detector 50 and the detector coordinates (pose) from the tracking system 60. The nuclear data input also includes a synchronization module which transmits the nuclear radiation values to the can assign respective detector coordinates. For this purpose, for example, time stamps with which the nuclear radiation values and the detector coordinates are provided can be compared and variables which have been carried out in a common time window can be assigned to one another. Alternatively, the pose associated with a particular nuclear radiation value may also be determined by interpolation of temporally adjacent measured poses. Generally this assignment is also called

Synchronization of the nuclear radiation values with the detector coordinates (the detector pose).

As a general aspect, the measured nuclear radiation levels per se do not necessarily allow a complete 3D reconstruction of a nuclear image. Nevertheless, they are useful for updating the stored original nuclear image, as described below.

With reference to Fig. 7, the update of the stored

original nuclear image described in more detail. The update is performed by an image update module 20, which may, for example, be implemented as a program running on the computer 90 (see FIG. 1). Further illustrated in Figure 7 are the following elements already described above: nuclear radiation detector 50, tracking system 60, and nuclear data input 70 with the synchronization module for associating the nuclear radiation values with the respective detector coordinates. The image update module 20 includes an update policy to change the original nuclear image 10 stored in the data store in response to the nuclear radiation values and the detector coordinates to obtain an updated three-dimensional nuclear image 18 the information obtained from the measured data is taken into account. The image update module 20 thus generates an updated three-dimensional nuclear image 18 by applying the update policy to the original nuclear image 10.

For this purpose, the image update module 20 includes an image variation function 22 and an image quality functional 24. The image quality functional 24 accesses as

Inputs to the nuclear radiation values (more precisely, the data set comprising the nuclear radiation values and the associated detector poses) from the nuclear data input 70 and to the nuclear image. This may be the original nuclear image or an already varied nuclear image. From these inputs, and if necessary from other quantities, the image quality functional 24 calculates an image quality value (in real number) that indicates a quality of the nuclear image with respect to the newly measured nuclear radiation values. Described below is an example implementation of an image quality functional 24 in which a low image quality value expresses a high quality of the nuclear image. In this case, as will be assumed hereinafter, the image update module 20 is implemented to vary the nuclear image 10 to approximate a minimum of the image quality functional 24. However, too - with

obvious changes - an equivalent implementation possible in which a high image quality value expresses a high quality of the nuclear image and approximates a maximum.

The image variation function 22 serves to vary the nuclear image 10 (or an already varied nuclear image) in dependence on a number of arguments or parameters, which may also be random parameters. Below, an example implementation of an image variation function 22 will be described.

The image update module 20 accesses the original nuclear image 10 and varies it using the image variation function 22 such that the result approximates a minimum of the image quality function 24 (within the scope of FIG

Image Variation Function 22 Produceable Nuclear Images). This can - depending on the

Image Quality Functional 24 - in some cases done by exact computation, but can usually be achieved most efficiently by an iterative method. It will the image variation function 22 is applied multiple times to the original nuclear image 10, or recursively applied to the previously obtained and stored varied nuclear image. Alternatively, the image variation function 22 is repeatedly applied to the original nuclear image 10, with parameters of the image variation function 22 being iteratively adjusted so that the minimum is approximated stepwise. The

Nuclear image 10 can also be varied only indirectly in this case by merely varying the parameters of the image variation function 22 and expressing the image quality functional 24 in response to these parameters. Also such

Variation only of these parameters is to be regarded as (indirect) variation of the nuclear image, since the varied nuclear image at any time by the image variation function 22 of the

Parameters can be obtained.

When the number of arguments of the image variation function 22 is large, it is a high-dimensional minimization problem. Iterative algorithms for efficiently solving such problems as the conjugate gradient method, simplex method, simulated annealing method are known and disclosed, for example, in the publication "Numerical Recipes" (available at http://www.nr.com). Further methods are Monte Carlo-based or Random-Walk methods, etc. With the aid of these or other methods, the image update module 20 iteratively varies the (original or already varied) nuclear image 10 by means of the image variation function 22 such that the resulting nuclear image gradually approaches a minimum of the image quality functional 24. When the iterations are completed, the final resulting nuclear image is output as the updated nuclear image 18.

Herein also the following nomenclature is used:

Size Icon Description

Nuclear image 10, 18 V X * Y * Z voxels in the coordinate system KN

(reconstructed intensity distribution)

Image variation function T, T (V) Figure T: V -> T (V); T (V) is the varied one

22 Nuclear image (X * Y * Z voxel, coordinate system KN).

T depends on further transformation parameters.

Nuclear data b, b (i) Vector of length N (= number of measurements), i = l..N, where the values b depend on the dimension d of the Detector numerical values (d = 0) or arrays of the

Dimension d (d = 1, 2).

Pose E »E (i) For each measurement p (i) the corresponding pose information

(Location and orientation of the nuclear detector), in a coordinate system KD whose relation to KN is known.

Projection P (V) Returns simulated simulations associated with a nuclear image V.

or nuclear data b = P (V), which should be measured by a detector with P (E, V) poses rj. Since P is of interest herein only for a given pose (the measured pose rj), the dependency of the projection on the pose p_ is not explicitly mentioned hereafter.

Quality Functional Q (V, £, b) Returns a real number, see e.g. Eq. (1) below

This nomenclature calculates the one described above and shown in FIG

Algorithm, the resulting nuclear image Vr as a minimum, Vr = argmin _T Q (T (V), p_, b).

The image quality functional 24, Q (T (V), p_, b), can be implemented as follows:

Q (T (V), _E , b) = II P (E, T (V)) - b II _L2 = Σί (P (E, T (V)) (i) - b (i)) ² (1 )

This functional is calculated from the nuclear radiation values and the varied nuclear image - and additionally from the pose - and gives the deviation of the measured

Nuclear data b from the expected or simulated for the nuclear image V measurement data P (p, V). Thus, a low image quality value expresses a high quality of the nuclear image. Instead of the L2 standard used in (1), any other functions for

Calculating a deviation, e.g. any L-norm, any correlation function, or the like.

Further extensions to equation (1) are possible. For example, a regularization term R (T) may be added to the above image quality functional (1). Such a regularization term R (T) can be a plausibility of a certain

Express image transformation T. For example, the regularization term R (T) can output an increased value for the nuclear image to highly distorting image transformations T (in the expression R (T), the argument T stands for a set of parameters that represent the Determine image transformation function T). Such a regularization term can be interpreted physically as the elasticity of the image transformation, since strong

Distortions are suppressed. In such or similar ways, various physical model assumptions can be implemented via possible image distortions, such as mass conservation, activity conservation, elasticity, possible changes in a tracer in the human body. In particular, these models may include physical models of deformations that are possible in the human body, such as discrete models (e.g., particle-and-spring models) or continuous models (e.g., hyperelasticity models). The regularization terms can also provide additional external information

consider, for example, information about intermediate tracer or

Radioactivity changes within the body (e.g., amount of additionally injected tracer, amount of tissue already removed, etc.). Such regularization terms allow me to ascertain a higher degree of security in a physically meaningful image transformation and to find a resulting image of good quality. Further regularization terms and methods are known from mathematics and can also be used here. Instead of using a regularization term, certain models of possible deformations can also be implemented as part of the image variation function 22, for example by allowing this function only for certain image changes.

The regularization term can depend not only on the image transformation function T or its parameters, but additionally on further parameters such as the transformed image T (V) itself or on the history of the previous transformations. The regularization term may also include models of the image. Such models can be implemented as so-called Finite Element Models (FEM), Finite Volume Models (FVM), Statistical Shape Models, Statistical Deformation Models. Examples of one

Regularization terms are referred to below together with possible image transformations T.

An abort condition may also be implemented in addition. For example, if the image quality value calculated by the image quality functional is too low and / or an acceptable value requires unrealistic image deformation, then the issue of a warning signal may be initiated.

The image variation function 22, T (V), may be as follows, for example

By means of a segmentation function, the original nuclear image 10 is first segmented (subdivided) into sub-volumes. This subdivision can be after a fixed rule (approximately in s equal subvolumes of the size X / s * Y / s * Z / s) or in dependence on an image recognition and / or an anatomical body model carried out, which recognizes and segmented certain body parts as a whole. Possible

Image segmentation algorithms are so-called region growing, level sets, snakes, histogram-based, atlas-based, graphene-based or similar

Segmentation algorithms. The image variation function 22 is then configured to apply at least one or more operations to each, or at least some, of these sub-volumes depending on respective parameters. Soche operations can be selected from the following: rotate, move, stretch, distort, shear, or a mixture thereof.

In a simple but already useful example, the

Image variation function 22, for example, only an independent shift of the respective sub-volumes. In this case, the transformation can be described by a set of displacement vectors ti... T _m (one vector for each subvolume displaceable by m). In a more complicated example, in addition, the boundary regions of the subvolumes are deformed (expanded or compressed) in dependence on the displacement such that the boundary voxels of adjacent subvolumes are not shifted too much relative to one another.

A similar effect can also be achieved by means of a regularization term which includes a summand for each pair of adjacent sub-volumes k, 1

f (i _k> yes) = (£ _k - yes), which acts like an elastic spring and suppresses too much different displacements:

More generally, and in accordance with a general aspect, the image quality functional may thus include an error term (consistency term) and a regularization term, the error term being derived from the varied nuclear image, the

Nuclear radiation values and the detector coordinates depend (at least indirectly) and a consistency (ie, a goodness of agreement or an inconsistency) expressed between the varied nuclear image and the nuclear radiation values; and wherein the regularization term depends at least on the varied nuclear image (at least indirectly, eg via variation parameters) and expresses a plausibility of the varied image.

Even a simple shift or rotation of the nuclear image as a whole is not yet regarded as an image variation function, since in this case the nuclear image is not varied in the true sense. For a picture variation is a more complex

Alteration of the nuclear image, at least one compression, elongation, or segmentation in at least two sub-volumes and their at least partially independent shift required.

According to a further embodiment of the image update module 20, this may also include an image reconstruction algorithm 36, as shown in Fig. 8. The image reconstruction algorithm 36 receives the original nuclear image 10 as a seed vector (input 32), and then applies at least some steps of an iterative image reconstruction process to reconstruct a reconstructed nuclear image from the original nuclear image as the starting image. On the one hand, the image reconstruction algorithm 36 uses at least part of the measurement data 12 (data 34) used to reconstruct the original nuclear image, but also the nuclear radiation values 35 from the nuclear radiation detector 50 and the detector coordinates from the tracking system 60 Iterations step the reconstructed nuclear image is then output as the updated nuclear image 18. Such an iterative image reconstruction method of the image reconstruction algorithm 36 may also be based on iteratively varying the nuclear image 32 by means of a

Image variation function and an image quality function.

Fig. 9 illustrates a possible operation of the described

Image reconstruction algorithm in more detail. According to this operation, the original nuclear image (radiation distribution Vo with X ₀ * Yo * Z ₀ voxels) was obtained from No original measurement data. This nuclear image is now using new

additional nuclear data (N-No Nuclear Radiation values with associated pose of the Nuclear Radiation Detector) are updated. Optionally, the number of voxels can also be increased, for example because the additional measurements allow a higher spatial resolution or an increased VOI (volume of interest).

In Fig. 9, the already mentioned projection image P is shown as a matrix. As described above, the map P for a nuclear image V outputs associated simulated nuclear data b = P (V) which should be measured by a detector with poses p. P is a linear map, so P can be represented as a matrix. The values of the matrix P are obtained from the poses p of the detector with the aid of a geometric model of the measurement environment. The matrix P for all nuclear radiation values (nuclear radiation values of the original nuclear image and newly added nuclear radiation values) can be divided into partial areas Pa, Pb, Pc and Pd. Pa is the sub-matrix that is the

is assigned to the original nuclear image: This matrix forms for a given

Intensity distribution (original nuclear image) Vo (Xo * Yo * Z ₀ voxel in

Coordinate system of the original nuclear image) associated simulated nuclear data bo = Pa (Vo), which would be expected at the original measurement (No measurements). The entries of the matrix Pa depend on the pose p_ of the detector in the original measurement. The original nuclear image can thus be obtained by approximating the equation bo = Pa (Vo), with bo as the (approximately) actually obtained nuclear data, or by minimizing the corresponding function of Eq. (1) above.

The newly added nuclear radiation values can be combined in one

Embodiment together with the original Nuclear Radiation values, and in the same way as they are used to reconstruct the radiation distribution. If the coordinate system for the radiation distribution remains unchanged (X ₀ * Yo * Z ₀ voxels), this only leads to an extension of the matrix around the sub-matrix Pb, since the

Reconstruction is now done on the basis of N instead of No measurements. Optionally, portions of the original measurements may also be removed, as described below with respect to Pd

described.

For better spatial resolution or enlargement of the volume, the additional measurements can optionally be used to increase the number of voxels

Contribute to radiation distribution. In this case, the number of rows of the matrix P increases to X * Y * Z rows (= dimension of the output vector V). This increase can also be used so that the equation to be solved is less overdetermined and thus used to reduce artifacts. In this case, further sub-matrices Pc are added to the matrix P.

The updated nuclear image (updated intensity distribution) can then be obtained by reconstruction analogous to the original nuclear image, eg by (approximate) extremalization of the one shown in Eq. (1) defined functions for the matrix P. In this case, as described above, an interactive method for minimization can be used. In the iterative method, an arbitrarily chosen start vector can be used. For example, a seed vector may be used that can be generated using the original nuclear image was generated, for example, as the entries associated with the sub-matrix Pa (first Xo * Yo * Zo entries) containing the original nuclear image. The further entries of the start vector may be, for example, random values.

Optionally, the size of the matrix can be reduced by deleting selected rows and / or columns of the matrix P (matrix areas Pd). In other words, image areas of the radiation distribution (rows of the matrix P) are omitted or

summarized, and / or nuclear radiation values (columns of the matrix P) are omitted or summarized. This can speed up the calculation.

According to a possible general aspect, according to the updating rule, for example, one or more of the following criteria for omitting the image areas of the radiation distribution (rows of the matrix P) may be applied:

The significance of the measured nuclear radiation values for these image areas is low (standard of the row vector of the matrix P falls below a predetermined limit value);

The image areas are outside a given VOI (volume of interest);

The informative value of the nuclear radiation values is not sufficient for the image area (no pose of the nuclear radiation values fulfills a predetermined reference criterion in relation to the image area, and / or norm of the corresponding row of the matrix P falls below a predetermined limit value).

According to another possible general aspect, according to the updating rule, for example, one or more of the following criteria for omitting the nuclear radiation values (columns of the matrix P) may be applied:

The validity of these nuclear radiation values for all image areas to be imaged is low (norm of the column vector of the matrix P falls below a predetermined limit value);

The expressiveness of one of the nuclear radiation values or a combination of nuclear radiation values for all image areas to be imaged is similar to the significance of another of the nuclear radiation values or another Combination of nuclear radiation values (norm of a linear combination of column vectors of the matrix P falls below a given limit value);

The pose of these nuclear radiation values meets a predetermined omission criterion;

The time at which the nuclear radiation values were measured is longer than a predetermined time;

A (equal or different) combination of these criteria.

Examples of a given omission criterion are shown in FIG. 10. FIG. 10 schematically illustrates the trajectory 7 of FIG

Nuclear radiation s detector and the detector coordinates (pose) 6 at the respective nuclear measurements. The detector coordinates 6 are represented by a respective arrow whose foot point symbolizes the location of the detector and its direction the solid angle of the detector in the respective measurement.

The group 6a groups three very similar detector coordinates (poses), for which very similar column entries of the matrix P are expected. To reduce computation time, the three associated measurements can be combined into one measurement. This is achieved by comparing poses of the measurements and similar poses (norm of difference of poses smaller than a given limit)

be summarized.

A similar criterion leads to the same result: The similar poses lead to similar columns of the P-matrix. Therefore, similar columns of the P-matrix can be summarized. More generally, columns of the P matrix may be grouped together that have a linear combination whose norm is less than a predetermined limit. As a result, an over-determination of the system of equations to be solved can be avoided.

Another omission criterion relates to the pose 6b: Here very few measurements with similar pose are available, and for the associated image area (voxels of the radiation distribution) the system of equations is underdetermined and therefore not stably solvable. This criterion can be used to omit voxels or

summarize. For example, the voxels that can not pose with any pose

has sufficient effect context, be omitted. The sufficient Interaction can either be determined geometrically - for example, each voxel to which a pose vector 6 is directed sufficiently and no or insufficiently marked voxels are eliminated. Alternatively or additionally, voxels whose row vectors of the matrix P fall below a predetermined standard can be eliminated.

The criterion illustrated by pose 6b may also be used as the elimination criterion for the nuclear data: for example, measurements where the detector has been moved at a high speed may be omitted, as poor data quality is expected in these measurements. At the same measuring intervals, the

Speed of the detector can be expressed by a (norm of) a difference between the poses of second temporally adjacent measurements. If the difference exceeds a predetermined limit, the corresponding measurement is omitted.

In the following, further embodiments will be described with reference to FIGS. 2 to 7. The embodiments are similar to the embodiment of Figs

among themselves, and like reference numerals refer to similar components. Apart from the deviations described below, the description of the other figures can also be used for the respective figure described.

The nuclear detector 50 of FIG. 2 is not directly, but only indirectly, connected to a tracking target 62 via a mechanical tracking system 62a. This mechanical tracking system 62a connects the detector 50 via a series of arms to a reference body, the bed 86 for the person to be imaged 84

Tracking system 62a are connected to each other with hinges such that the detector is freely movable (ie at least within a certain volume in all three

Spaces moved and can even be freely rotated here). At the same time, the arms hold the detector in a defined position even when released.

The deflection of the joints of the tracking system 62a is detected and thus the pose tracked relative to the reference body 86. The reference body 86 in turn is fixedly connected to the Nachführszielobjekt 62. Finally, in Fig. 2, the same is achieved as in Fig. 1, namely a relative pose between the nuclear detector on the one hand

(Coordinate system KD) and the tracking target object 82 attached to the patient

(Coordinate system ND) determined.

In the nuclear detector of Fig. 3 is thus a mechanical

Tracking system 62a determines the relative pose of the nuclear detector to the bed 86. The The person to be imaged is thus in a predetermined position on the bed 86 or is fixed in such a way that a fixed position of the nuclear image coordinate system KN relative to the bed 86 can be assumed. Thus, the tracking system 62a again allows a relationship between the detector coordinate system KD and the nuclear image coordinate system CN.

The embodiment of FIG. 3 has the disadvantage that position changes of the patient can only be detected indirectly from the nuclear data (via the image quality functional). According to a general aspect, this disadvantage can be reduced as follows: A localization module localizes at least one reference body location of the person 84 to be imaged relative to the image coordinate system KN, preferably during or immediately before / after measuring the nuclear radiation, so that no substantial change in attitude of the person 84 occurs is expected. This localization does not need coordinates of the

To be image coordinate system. This localization can then, for example, the

Register the at least one reference body location with a corresponding point in the nuclear image (radiation distribution image) allow and so even more reliable

Allow assignment.

According to one example, further location data that depends on a location of the person 84 to be imaged is collected, and the original nuclear image

(Radiation distribution image) is pre-transformed depending on these position data. If the person to be imaged 84 is mentioned here, this does not necessarily have to refer to the entire person, but may also refer to only a part of this person. For example, the location data also the location of only a part of the

person to be imaged, but not their entire body.

In a variation of the embodiment of Fig. 3 according to this aspect, for example, a bearing surface of the bed 86 is equipped with weight sensors. The weight sensors provide weight distribution data representing a weight distribution of the person 84 to be imaged. By comparison with approximately corresponding weight distribution data measured in the acquisition of the original nuclear image, an image preprocessor can now detect a change in position and pre-transform the nuclear image accordingly (e.g., the nuclear image, more specifically the radiation distribution image as a whole or parts thereof corresponding to a centroid shift or

Shift printing positions shift). Alternatively or additionally, the image quality functional has a dependency on these location data. This is illustrated in Figure 7 as an optional variant, according to which the image quality functional 24 is operatively connected to the localization module 80 for receiving localization data of the reference body location. Location data of the

Location module 80 flows into the image quality functional 24 and is used to calculate the image quality value.

In accordance with this aspect, for example, another addend in the quality functional may favor those image transformations that provide good agreement with the location data (eg, the weight distribution). For example, such a summand may be proportional to the L2 norm of the difference between the original data (obtained during the acquisition of the original nuclear image) and the current one, transformed with the nuclear image

Location data, where both location data are represented by a vector.

In an analogous manner, instead of, or in addition to, the weight distribution, the location of a body surface of the person 84 in the room (e.g., a prominent and textured body surface such as the neck or underarm area, or an easily accessible one

Body surface - it does not have to be the entire body surface of the person 84) are used as positional data. This location of the body surface can be determined by a surface determination system and then used for pre-transformation or image regularization. The body surface may be detected, for example, in a scanning mode by mechanical scanning by a tracked sensing object, which may be identical to the nuclear detector 50, or by another instrument for determining the position of a surface. Such an instrument may also be a time-of-flight camera, a stereoscopic camera, a laser scanner, or a combination thereof. For example, to determine the location of the surface, the tracking system may be used directly, such as by attaching one or more tracking targets or fiducials to it. Also commercially available systems such as e.g. The Kinect system can be used to determine the surface or position data.

Also, an image evaluation system for evaluating at least one distinctive feature (so-called landmark) in the nuclear image (more precisely, the

Radiation distribution image) or in an additional image, such as an optical image or an additionally generated sectional image such as US, CT, MR, used in an analogous manner are suitable for the determination of location data and can be used as described above.

The image quality functional 24 thus includes one according to one aspect

A localization term indicating a match of the reference body location (s) located by the location module 80 with the nuclear image (more precisely, the

Radiation distribution image). This localization term can be used as part of the

Consistency term or regularization term, or as an additional term.

In the further embodiment of FIG. 4, a high-energy gamma camera for PET applications is arranged on one side. The original nuclear image is obtained by means of PET imaging (by the two-dimensional detector 50) and a further detector, not shown in the image, for the detection of coincidences. This further detector can be designed analogously to the detector 60 of FIG. 1 and, in particular, be freely movable. In addition, patient location data is determined by the following location sub-modules, as described in greater detail above with respect to FIG. 3: a bearing surface 86 with weight sensors; a stereoscopic camera 60 as a surface determination system; and a tracking target 82 attached to the person 84.

In addition, the nuclear detector 50 is fixedly attached thereto

Tracking target 52 also tracked. Therefore, the nuclear detector can be moved freely and still maintain a relation between the tracking coordinate system and the detector coordinate system. If necessary, then the nuclear detector is also completely removed after the original nuclear image has been created.

Again, the nuclear image or radiation distribution image may be iteratively varied by the image update module as described above to approach a minimum of an image quality functional. The image quality functional may here include, for example, the deviation between the original image data (ie, the one obtained when the original nuclear image was generated) and the current location data of the localization submodules, which has been transformed with the nuclear image. One with || P (£ _> T (V)) - b II L2 comparable term can also be omitted. An additional regularization term may optionally be included. The tracking system 60 in Fig. 4 is an electromagnetic tracking system. However, any other tracking system may be used.

Fig. 5 corresponds in principle to Fig. 4; however, here the nuclear detector 50 is integrated into the bed 86 used for nuclear image generation. The nuclear detector 50 may be used for the original nuclear image and may additionally provide updated nuclear measurements which are used for updating similar to that described in FIG. Similar to the detector of Fig. 4, this detector 50 is also tracked by a tracking target 62 fixedly connected to the bed 86, in this case by means of an optical tracking system 60. In addition, another freely movable detector may be used to allow different viewing angles. either analogously to the description of FIG. 4 with the aid of a PET method using both detectors, or as an additional detector in analogy to FIG. 1.

For the generation of position data that can be used by means of the localization module for updating the nuclear image, only the tracking target object 82 attached to the patient 84 is shown here, but the other position data described herein can additionally be used.

In all Figures 1 to 5, a screen for outputting the updated nuclear image 92 and possibly additional information is additionally shown. The system 90 reproduces the updated nuclear image (more precisely, the radiation distribution image) in real time (more precisely: fast real time) on the screen 92. Real time means that the image is continuously updated based on the measurements just taken. This is possible because there is no need for complete image reconstruction, and fewer measurements are required than required for complete image reconstruction.

This screen is shown in more detail in FIG. Since the updated nuclear image (more precisely, the radiation distribution image) - as well as the original nuclear image - is a complete three-dimensional image, various perspectives or sections can be displayed to convey the three-dimensional image information.

In addition, further information can be displayed, such as a

Advanced reality information and / or information for guiding the nuclear probe to obtain improved measurement data. Getting such information is about in the Patent applications DE 10 2010 017 543 and DE 10 2009 042 712, the contents of which are incorporated herein by reference.

Instead of the tracking system 60, 62a shown in FIGS

In general terms, various tracking systems are used to determine the detector pose, for example an electromagnetic, optical, mechanical (passive or active, eg fixed to a controllable robot arm) tracking system can be used. The tracking system can also be realized by a fixedly mounted on the gamma probe camera, which aims at several stationary Nachführzielobjekte. Also, a tracking system may be realized by an acceleration sensor and gyroscope integrated with the nuclear detector 50.

Referring now to Fig. 11, a method of updating an original nuclear image will be explained. The method comprises the following steps:

Sl: start of the procedure;

S2: performing an initial measurement of nuclear radiation;

S3: Reconstruction of an original nuclear image from that measured in S2

Nuclear radiation: Thus, an original nuclear image is obtained in steps S2, S3.

This original nuclear image is a complete three-dimensional nuclear image, i. it is finished and can be displayed directly to visualize the radiation distribution.

S4: After the final creation of the original nuclear image, measure nuclear radiation using a moving along a freely movable path

Nuclear radiation s detector to obtain nuclear radiation values, and tracking the nuclear radiation s detector while measuring the nuclear radiation, so that

Obtain detector coordinates indicative of a pose of the tracked nuclear radiation detector in a reference coordinate system;

S5: assigning the nuclear radiation values to respective ones of the detector coordinates;

S6: generating an updated three-dimensional nuclear image by applying an update policy to the original nuclear image, wherein the update policy alters the original nuclear image in response to the nuclear radiation values and the detector coordinates. The original nuclear image contains at least a partial image of a

person to be imaged. The method may then further in step S4 localizing at least one reference body location of the person to be imaged relative to the

Reference coordinate system include.

FIG. 12 shows a diagram which, in the method of FIG. 11, represents a quality value Q as a function of an iteration value x. Here, in step S2 (see FIG. 11), the original nuclear image is represented by the matrix Pa shown in FIG. 9 and the original nuclear data b ₀ (vector of length N ₀ ), and FIG

original radiation distribution image Vo (Xo * Yo * Zo voxel, original nuclear image in the strict sense) is obtained by minimizing the quality value Q of Eq. (1) using the matrix Pa and the original nuclear data bo.

The reconstruction of the original nuclear image (step S3) is accomplished by minimizing the quality value Q in an iterative process, e.g. the maximum likelihood expectation maximization (MLEM) procedure. Here, an arbitrary (e.g., randomly selected) seed value is used for the original radiation distribution image Vo, and iteratively, using the MLEM method, a minimum of Eq. (1) approximated. These

Minimization is shown in the diagram of FIG. 12 during the interval x ₀ ... Xi-1. To prevent artifact formation, the iterative process is aborted once Q has sufficiently converged, even if further minimization would be possible.

Subsequently, steps S4 and S5 are performed. Thus, the user gets to see a picture only when it has sufficiently converged. Thereafter, the nuclear data b is updated (now length N) as described above, and the matrix P is updated using the detector coordinates, i. brought into the shape shown in Fig. 9.

At step S6, the radiation distribution image V (X * Y * Z voxel, updated nuclear image in the strict sense) is obtained by minimizing the quality value Q of Eq. (1) using the updated matrix P and the updated nuclear data b. This is again done by the above-described iterative minimization of the quality value Q starting from a start vector for V. Here, the original result (nuclear image Vo) can be used as the start vector for the entries up to Vo, and the remaining entries can be arbitrary, eg random values. Alternatively, arbitrary values can be used for all entries of the start vector for V as well as for the entries up to Vo. The Minimization of Q is shown in the graph of Fig. 12 during the interval x ₁ ... x ₂ -l.

The update policy may optionally be applied multiple times or even iteratively. In Fig. 12, the updating rule is also applied to the other times x ₂ , x ₃ , etc. in a manner analogous to that at time x ₁ . As a result, the relative weight of the original nuclear image gradually decreases, and any changes in intensity distribution (soft tissue movement, etc.) can be detected.

[0089] The image reconstruction allows to the times x _0, x, x _2, and so each display an updated image. By using modern hardware and software (GPU-accelerated modern reconstruction algorithms such as MLEM) it is possible to achieve an iteration in fractions of a second even with inexpensive standard PC components. This allows an image update within a few seconds (about 10 seconds per

Update).

In the following, still further variants of a nuclear image system are described: According to a variant, a nuclear image system for updating an original nuclear image (10) comprises: a nuclear detector system (50) for generating a three-dimensional original nuclear image (10) comprising at least a partial image of a person to be imaged (84) receives; a source image localization module configured to generate original location data indicating a location of the person (84) to be imaged relative to the nuclear detector system (50) during creation of the original nuclear image, a location module that is identical to or different from the source image location module configured to generate current location data indicative of a location of the person (84) to be imaged relative to

Nuclear Detector System (50) at a later time after generating the

indicate the original nuclear image; an image update module (20) having a

Update rule to change the original nuclear image (10) depending on the original location data and the current location data, the

An image update module configured to generate an updated three-dimensional nuclear image (18) by applying the update policy to the original nuclear image. According to another variant, a method of updating an original nuclear image comprises: obtaining (S2, S3) the original three-dimensional nuclear image of a person (84) to be imaged; Generating original location data indicating a location of the person (84) to be imaged relative to the nuclear detector system (50) during the generation of the original nuclear image; Thereafter, generating current location data representing a location of the person (84) to be imaged relative to

Nuclear Detector System (50) at a later time after generating the

indicate the original nuclear image; and generating (S6) an updated one

3-dimensional nuclear image by applying an update policy to the original nuclear image, the update policy being the original one

Nuclear image changed depending on the original position data and the current position data. These variants may be combined with any other aspects described herein, particularly with the subject matters of the claims.

Claims

claims

A nuclear image system for updating an original nuclear image (10), the nuclear image system comprising: a data memory for storing the original three-dimensional image

Nuclear image (10); a nuclear radiation detector (50), movable along a free-changing path, for measuring nuclear radiation to obtain nuclear radiation values; a tracking system (60) for tracking the nuclear radiation detector (50) during nuclear radiation measurement to obtain detector coordinates indicative of a pose of the tracked nuclear radiation detector in relation to an image coordinate system of the nuclear image; a nuclear data input (70) configured to receive the

Nuclear radiation values from the nuclear radiation detector (50) and the detector coordinates from the tracking system (60), and to associate the

Nuclear radiation values to the respective detector coordinates; and an update image module (20) having an update policy for changing the original nuclear image (10) in response to the nuclear radiation values and the detector coordinates, wherein the image update module generates an updated three-dimensional nuclear image (18) by applying of the

Update rule is configured to the original nuclear image.

The nuclear image system of claim 1, wherein the original nuclear image

at least one partial image of a person to be imaged, the nuclear image system further comprising: a localization module (80) configured to locate at least one

Reference body location of the person to be imaged (84) relative to the

Image coordinate system.

The nuclear imaging system of claim 2, wherein the locating module preferably comprises at least one of the following: (i) a marking element (82) to be affixed to the person to be imaged (84), the tracking system (60) being configured to obtain marking element coordinates indicative of a pose of the marking element (82);

(ii) a bearing surface (86) having weight sensors for detecting a person (84) to be imaged by weight when the person rests on the bearing surface;

(iii) a surface determination system for locating a

Body surface of the person to be imaged;

(iv) an image evaluation system for evaluating at least one distinctive one

Feature in the nuclear image or in an additional image.

4. The nuclear image system according to one of the preceding claims, wherein the

An image update module (20) comprises: an image variation function (22) for varying the nuclear image; and an image quality functional (24) for computing an image quality value from the nuclear radiation values and the varied nuclear image, wherein the image update module (20) is configured to iteratively vary the nuclear image using the image variation function (22) such that the image update function Module calculates the nuclear image in a plurality of iterations so that the nuclear image gradually approaches an extremal value of the image quality functional, and outputs the nuclear image resulting from the computation as the updated nuclear image (18).

The nuclear image system of claim 4, wherein the image quality functional (24)

is operatively connected to the localization module (80) for receiving localization data of the reference body location and configured to calculate the image quality value in dependence on the location data.

The nuclear image system of any one of claims 4 to 5, wherein the image variation function (22) is configured to segment sub-volumes of the nuclear image and manipulate the sub-volumes in the nuclear image.

The nuclear image system of any one of claims 4 to 6, wherein the image quality functional includes a consistency term and a regularization term, wherein the error term of the varied nuclear image, the

Depends on nuclear radiation values and the detector coordinates and expresses consistency between the varied nuclear image and the nuclear radiation values; and wherein the regularization term depends at least on the varied nuclear image.

The nuclear image system according to one of the preceding claims, wherein the

Nuclear Radiation s detector is a detector of dimension 0, 1 or 2.

9. A method for updating an original nuclear image, the method

full:

Obtaining (S2, S3) the original nuclear image as a three-dimensional nuclear image;

Thereafter, measuring (S4) nuclear radiation by means of a nuclear radiation detector movable along a free-changing path

To get nuclear radiation values,

Tracking the nuclear radiation detector during the measurement of the nuclear radiation so that detector coordinates are obtained which are a pose of the

indicating the tracked nuclear radiation detector in a reference coordinate system;

Assigning (S5) the nuclear radiation values to respective ones of the detector coordinates;

Generating (S6) an updated three-dimensional nuclear image by applying an updating rule to the original nuclear image, wherein the

Update rule the original nuclear image depending on

Nuclear radiation values and the detector coordinates changed.