DE102007028906B4 - Use of an image acquisition, image archiving and image visualization system and methods to support staging or restaging performed under nuclear medicine or radiological imaging to register and assess the development of spatially and / or structurally alterable pathological structures - Google Patents

Abstract

Method for supporting a staging or restaging carried out under nuclear medical or radiological imaging for registering the stage and assessing the development of spatially and / or structurally variable pathological structures inside the body of a patient to be examined, with the steps: - Registering (S3a + b or S2 ' + S3 ') of two digital images taken at different times, representing the same tissue area, in which image data (14, 17) of corresponding partial areas of the same tissue structures depicted in both images sink together and with the help of mutually independent rigid coordinate transformations with regionally different positional offset (Δx optj, Δy optj or Δz optj) and / or angular displacement parameters (Δφ optxj, Δφ optyj or Δφ optzj) are co-registered, and - converting (S4) the image data of these co-registered partial areas into composable parts of a D to be displayed Ifference image of the two images characterized in that the regionally different position offset (Δx optj, Δy optj or Δz optj) and / or angular offset parameters (Δφ optxj, Δφ optyj or Δφ optzj) of each individual area using the method of the smallest Error squares based optimization criteria are determined, according to which ...

Description

The present invention relates to a use of an image acquisition, image archiving and image visualization system for generating, storing, postprocessing, retrieving and graphically visualizing medical image data, e.g. B. can be successfully used in the clinical field in the context of nuclear medicine or radiological sectional image diagnostics. Furthermore, the present invention relates to a method that can be carried out by this system for supporting a staging or restaging performed under magnetic resonance or computer tomographic imaging for registering the stage and evaluating the development of spatially and / or structurally variable pathological structures in the interior of a patient to be examined (follow-up ), in which structural changes of a pathological tissue region of interest are clearly highlighted by subtraction of two MRI or CT images recorded during different nuclear medicine or radiological examinations and thus at different times (reference and follow-up recording). In the pathological structures to be examined, it may in particular z. Malignant (malignant) tumors, such as metastatic carcinomas or sarcomas, osteolytic (bone disintegrating), osteoblastic (bone-forming) or mixed osteolytic / osteoblastic metastases, or osteoporotic altered bone structures.

The skeletal tissue of the human skeleton is one of the fourth most common areas of cancer affected by metastases, after lymph nodes, lung and liver. The most common and well-known cancers responsible for the formation of metastases in bone tissue include prostate cancer and breast cancer. In about 50% of all female breast cancer patients, bone metastases develop during the course of the disease. Other examples of cancers that can lead to metastases in the bone tissue are lung cancer and kidney cancer. Often, several areas of the skeleton are affected (although in different degrees). Most commonly, the vertebral bodies of the spine become infested, and then, with decreasing frequency, femurs, pelvis, ribs, sternum, skull, and humerus.

Bone metastases usually develop as a result of hematogenous or lymphogenic (that is, via the blood or lymph circulation) sowing malignant tumor cells that colonize in other parts of the body and multiply there. On average, bone metastases are detected in approximately 30% of cancer patients with malignancies at the time of initial diagnosis. They are often symptomatic and severely affect a patient's quality of life. The clinical picture is characterized by chronic pain, pathological fractures, spinal compression syndromes and hypercalcemia (blood calcium level disorders). The changes in the affected bone tissue are a consequence of the ability of tumor cells to activate local osteoclasts (osteolytic cell tissue) or osteoblasts (osteoblastic cell tissue), ie cellular tissue which either leads to increased bone breakdown or promotes increased bone formation. While a healthy bone has a stable balance between osteoblasts and osteoclasts and thus a stable balance between bone formation and degradation, this balance is severely disturbed in a malignant cancer characterized by formation of osteoblastic or osteolytic metastases in the area of the bone tissue. While tumor cells, with a few exceptions, are unable to destroy or rebuild bone tissue, the metastatic cell tissue can either dissolve or re-form bone tissue. However, the bone material resulting from osteoblastic metastasis is unusable because it leads to the formation of an unstable bone. In this context, it should also be mentioned that bone metastases are usually not "purely osteolytic" or "purely osteoblastic", but rather get their name according to which of the two processes predominates. If the bone-dissolving process and the bone-forming process are equally strong, this is called "mixed osteolytic / osteoblastic metastases".

Changes in the bone structure of a cancer patient are usually difficult to detect on a CT image and require a high degree of attention and experience of the respective treating radiologist in the diagnosis.

A non-invasive imaging procedure that has become established in recent years as the standard method of nuclear-medical imaging diagnostics for generating bone scans and is therefore often used for early diagnosis and determination of the extent of bone metastases is skeletal scintigraphy. With the aid of this imaging method, image data combined into a volume data set are obtained from sectional images of different layer planes of a bone tissue to be examined, which are then analyzed using an image rendering method (eg by means of multiplanar Reformatting or by volume rendering technique) to 2D projections or 3D views of different viewing angles can be reconstructed. By contrast, the 2D information of the sectional images generated with the aid of such a nuclear medicine imaging method only allows an assessment of the position and the size stage of metastases of interest in the respective layer plane of the relevant slice image, ie a highly limited spatial assessment.

Computed tomography (CT) is today regarded as a standard imaging method for monitoring the development of metastatic malignant tumors in various body regions. In the course of various cancer therapy procedures (eg in the context of chemotherapy or radiotherapy, etc.) CT-based follow-up examinations are usually performed at intervals of a few weeks in order to monitor changes in the tumor burden (ie the percentage of tumor) Proportion of tumor tissue to the total mass of a patient). In the event that the cancer breaks out again in a tumor patient or a new malignant tumor develops from metastases, they can be detected in good time and treated as quickly as possible via a minimally invasive procedure or surgical treatment under local anesthesia. Compared to scintigraphy, computed tomography provides volumetric image data that includes morphological (but not functional) information. In a CT scan, osteolytic bone metastases are presented as tissue areas with relatively low Hounsfield density and thus regions of low X-ray attenuation (ie relatively dark), while osteoblastic metastases are tissue areas with relatively high Hounsfield density and thus regions of higher X-ray attenuation (ie relatively bright ) being represented. However, the changes in the CT image are often marginal and, especially with regard to osteoblastic metastases, difficult to recognize or only to guess.

DE 10 333 563 A1 For example, therefore, a temporal image comparison method using a temporal image processing system which uses a segmentation module to isolate a region of interest between image signals and generate a segmentation signal therefrom generates a registration signal by means of a registration module which registers the region of interest. A comparison module generates a comparison signal of the image signals from the segmentation signal and the registration signal.

DE 24 066 22 C2 discloses an apparatus for generating a difference image from a first and a second image by manipulating the digital gray scale values. In particular, the device serves to process X-ray images of the same person in order to make clear anatomical changes which occur in the time interval between the acquisition of the two X-ray images so as to produce a difference image which can be analyzed better.

However, routine assessment of bone metastases on the bare (unprocessed) CT images is still cumbersome and requires a high level of experience from the radiologist conducting the examination. Conventional diagnosis, routinely performed in a tumor staging study, therefore, is correspondingly time consuming and has a low sensitivity in assessing changes.

OBJECT OF THE PRESENT INVENTION

Based on the above-mentioned prior art, the present invention is directed to improving the workflow of a staging or restaging for the registration and evaluation of changes in spatially and / or structurally variable pathological structures that have been presented by means of nuclear medicine or radiological imaging , Such an improvement in the finding could significantly increase the importance of magnetic resonance and computed tomography in performing such staging or restaging, particularly with regard to an assessment of the development of osteolytic bone metastases, osteoblastic or mixed osteolytic / osteoblastic metastases, and the need for follow-up Reduce upscintigraphies.

This object is achieved by the subject matter of the independent claims. Advantageous embodiments which further develop the idea of the invention are defined in the dependent claims.

SUMMARY OF THE PRESENT INVENTION

• – Registrieren (S3a + b oder S2' + S3') zweier zu unterschiedlichen Zeitpunkten aufgenommener, denselben Gewebebereich darstellender digitaler Bilder, bei dem Bilddaten (14, 17) einander entsprechender Teilbereiche derselben, in beiden Bildern abgebildeten Gewebestrukturen miteinander verlinkt und mit Hilfe voneinander unabhängiger rigider Koordinatentransformationen mit regional unterschiedlichen Lageversatz-(Δx opt / j , Δy opt / j bzw. Δz opt / j ) und/oder Winkelversatzparametern (Δφ opt / xj , Δφ opt / yj bzw. Δφ opt / zj ) koregistriert werden, und
• – Umrechnen (S4) der Bilddaten dieser koregistrierten Teilbereiche zu zusammensetzbaren Teilen eines anzuzeigenden Differenzbildes der beiden Bilder.

The present invention, in a first aspect, relates to a method of assisting a nuclear medicine or radiological imaging. Stagings to Registration of the stage and assessment of the development of spatially and / or structurally changeable pathological structures in the interior of the body of a patient to be examined with the steps:

• Registering (S3a + b or S2 '+ S3') two digital images recorded at different times and representing the same tissue area, in which image data ( 14 . 17 ) of corresponding subregions of the same tissue structures depicted in both images are linked to one another and with the aid of mutually independent rigid coordinate transformations with regionally different positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ), and
• - converting (S4) the image data of these coregistered subregions to composable parts of a difference image of the two images to be displayed.

The inventive method is characterized in that the regionally different Lageversatz- (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) of each individual subarea are determined by means of an optimization criterion based on the method of least squares, according to which the sum of the quadratic distances between the points of a point quantity predetermined on the respective subarea for each subarea of a reference image representing a specific tissue region, and with the aid of the individual, each point of the set of points on the respective subarea of the reference image via a bijective mapping rule exactly one same point of the relevant tissue area indicative point of the set of points on this sub-area calculated locally different rigid coordinate transformations transformed points of a point set assigned subarea of the comparison image is assigned.

The partial regions of the two images (RB and VB) are advantageously non-overlapping image regions which have been created by partitioning these two images (RB and VB) into subregions of a predetermined shape and size and have been delimited from each other, each subregion of the reference image is assigned to a partial area of the comparison image via a bijective mapping rule or equally advantageous to non-overlapping image regions, which were created by segmentation of these two images in coherent image objects and were delimited against each other, each subregion of the reference image over a bijective mapping rule exactly a portion of the comparison image assigned.

The method according to the invention is further characterized by a step of evaluating regional position offset Δx calculated in advance of the coregistration (S3a + b or S2 '+ S3') of the individual, mutually associated subregions of the reference and comparison image opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) with respect to their deviations in sign and magnitude, wherein non-contiguous, mutually corresponding tissue structures, which are shown in subregions of both images, are recognized as not belonging together when a threshold value violation of one of these parameter deviations is detected.

Furthermore, the method according to the invention is characterized by the step of automatic extraction (S6a) of features from the image data of the calculated difference image for the quantitative assessment of spatial and / or structural changes of the tissue structures depicted in the comparison image with respect to the tissue structures depicted in the reference image.

It is advantageous if the features extracted from the image data of the calculated difference image are the mean regional change of the Hounsfield density values and / or the size change of changing tissue structures.

It is likewise advantageous if, in the graphic visualization of the difference image, pixels with positive difference values are displayed in a different color coding than pixels with negative difference values and / or in the graphic visualization of the difference image, pixels with different difference amounts of the same sign are represented with different color saturation values.

An equally advantageous variant of the method according to the invention if the method has a calculation procedure in which the gray values of the pixels of the reference and comparison images are matched to one another such that the mean gray value and the standard deviation of the gray values of the comparison image data set ( 17 ) to the mean gray value or the standard deviation of the gray values of the reference image data set ( 14 ) be adjusted.

The present invention further relates to a use of an image acquisition, image archiving and image visualization system for carrying out the method according to claims 1 to 9, which is characterized by an image visualization tool ( 9 ), which has funds ( 15 respectively. 18 ) for performing registration of two digital images recorded at different times and representing the same tissue region, in which image data ( 14 . 17 ) of corresponding subregions of the same tissue structures depicted in both images are linked to one another and with the aid of mutually independent rigid coordinate transformations with regionally different positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) are coregistered, as well as a computing unit ( 11 ), with the aid of which the image data of the coregistered subregions are converted into composable parts of a differential image of the two images to be displayed.

Furthermore, it is advantageous if the subareas of the two images are non-overlapping image regions which have been created by partitioning these two images into subregions of a predetermined shape and size and have been delimited from one another, each subarea of the reference image being precisely defined via a bijective mapping rule is assigned to a subarea of the comparison image.

It is furthermore advantageous if the subareas of the two images are non-overlapping image regions which have been formed by segmentation of these two images into coherent image objects and have been delimited from one another, each subarea of the reference image via a bijective mapping specification being exactly one subregion of the comparison image assigned.

The most essential aspects and advantages of the system and method are summarized below:
The system provides an image acquisition, image archiving and image visualization system for generating, storing, postprocessing, retrieving and graphically visualizing medical image data with an image visualization tool having means for performing registration of two digital images taken at different times and representing the same tissue region in which image data of mutually corresponding subregions of the same tissue structures depicted in both images are linked to one another and coregistered with the aid of mutually independent rigid coordinate transformations with regionally different positional and / or angular offset parameters. In addition, the image visualization tool includes an integrated arithmetic unit, with the aid of which the image data of the coregistered subareas are converted into composable parts of a differential image of the two images to be displayed.

In the sub-areas of the two images may be z. B. are non-overlapping image regions that have been created by partitioning these two images in sub-areas of a given shape and size and were delimited from each other, each sub-area of the reference image is assigned via a bijective mapping rule exactly a portion of the comparison image.

Alternatively, the partial regions of the two images may be non-overlapping image regions which have been formed by segmentation of these two images into contiguous image objects and demarcated from each other, each subarea of the reference image being assigned to a partial region of the comparison image via a bijective mapping rule ,

In a further aspect, the present invention relates to a method of supporting a staging or restaging performed under nuclear medical or radiological imaging to register the stage and to assess the development of spatially and / or structurally variable pathological structures in the interior of the body of a patient to be examined. For this purpose, z. B. image data of a MRI- or CT-based carried out reference examination with image data that were acquired in the context of at least one magnetic resonance or computed tomography follow-up examination. In the pathological structures to be examined, it may be z. B. osteoporotic altered bone structures to malignant tumors, such as metastatic carcinomas or sarcomas, or in particular to osteolytic, osteoblastic or mixed osteolytic / osteoblastic metastases, the z. B. were detected by skeletal scintigraphy or as part of a computed tomography imaging process. The basic idea of the method according to the invention consists of both structural and spatial changes of the pathological structures to be investigated, ie. H. Changes in their composition, mass and density as well as changes in their size and spatial location, by differentiation of two recorded during different radiological examinations and thus at different times recorded MRI and CT images (reference and comparative image) clearly.

In this case, the method according to the invention is characterized by the step of registering two digital images taken at different times in the same tissue region, linking the image data of mutually corresponding partial regions of the same tissue structures depicted in both images and using independent rigid coordinate transformations with regionally different positional characteristics. and / or angular offset parameters are coregistered, and by a step in which the image data of the coregistered partial areas are converted into composable parts of a differential image of the two images to be displayed.

As has already been described with reference to the image acquisition, image archiving and image visualization system according to the first exemplary embodiment of the present invention, the subareas of the two images may be described, for example, as shown in FIG. B. are non-overlapping image regions that have been created by partitioning these two images in sub-areas of a given shape and size and were delimited from each other, each sub-area of the reference image is assigned via a bijective mapping rule exactly a portion of the comparison image.

Alternatively, as already mentioned above in connection with the image acquisition, image archiving and image visualization system according to the first exemplary embodiment of the present invention, the subareas of the two images may be non-overlapping image regions which have been formed by segmenting these two images into coherent image objects are and were delimited from one another, wherein in turn each subregion of the reference image is assigned to a partial region of the comparison image via a bijective mapping rule.

As regards the determination of the regionally different positional and / or angular offset parameters, which are required in order to bring the tissue structures illustrated in the individual subregions at least approximately (that is to say with a tolerable registration error) into registry when the co-registration is carried out, it is possible according to the invention, for example. For example, an optimization criterion working according to the method of least squares is used, according to which the sum of the quadratic distances between the points of a given point range and the point transformed by means of the individual, regionally different rigid coordinate transformations for each subregion of a particular tissue area Points of a set of points is calculated on the portion of the comparison image associated with this subarea. In this case, each point of the point set predetermined on the respective subregion of the reference image is assigned, via a bijective mapping rule, to exactly one point of the point set on the subregion of the comparison image assigned to the same subarea of the relevant tissue region.

Moreover, in the method according to the invention, an additional step is provided in which regional positional and / or angular offset parameters calculated with respect to their deviations in sign and magnitude are evaluated in advance of the coregistration of the individual, mutually associated subregions of the reference and comparative image. In this case, non-contiguous, mutually corresponding tissue structures, which are shown in partial areas of both images detected upon detection of a threshold exceeding one of these parameter deviations as not belonging together and then, as after performing a segmentation, be treated separately as independent, individual image objects.

In addition, according to the invention, a step may also be provided according to which features which are required for the quantitative assessment of spatial and / or structural changes of the tissue structures depicted in the comparison image with respect to the tissue structures depicted in the reference image are automatically extracted from the image data of the calculated difference image. The extracted features may be z. These may be, for example, the average regional change in Hounsfield density values and / or the resizing of changing tissue structures.

In the graphical visualization of the difference image pixels with positive difference values z. B. are displayed in a different color coding than pixels with negative difference values. Pixels with different difference amounts of the same sign can according to the invention z. B. are displayed with different color saturation values.

Moreover, the method according to the invention comprises a calculation procedure in which the gray values of the reference and comparison image pixels are matched to one another in such a way that the mean gray value and the standard deviation of the gray values of the comparison image data set are adapted to the mean gray value or the standard deviation of the gray values of the reference image data set ,

According to a further aspect, the present invention relates to a computer program with program code for carrying out all method steps according to one of claims 1 to 9, when the program is executed in an image archiving and image visualization system as described above.

BRIEF DESCRIPTION OF THE DRAWINGS

Further features of the present invention will become apparent from the dependent patent claims and from the following description of exemplary embodiments of the invention, which are illustrated by the following drawings:

1 10 is a block diagram illustrating the system architecture of the image acquisition, image archiving and image visualization system according to a first example;

2 10 is a block diagram illustrating the system architecture of the image acquisition, image archiving and image visualization system according to a second example;

3 1 shows a flow chart of a first variant of the method according to the invention for supporting a staging or restaging of spatially and / or structurally changeable tissue regions using the example of bone structures of a human or animal skeleton, which are affected by osteolytic, osteoblastic or mixed osteolytic / osteoblastic metastases are and

4 shows a flowchart of a second variant of this method according to the invention.

DETAILED DESCRIPTION OF THE INVENTION

In the following sections, the system components of the image acquisition, image archiving and image visualization system and the steps of the method according to the invention will be described in detail with reference to the accompanying drawings.

In 1 Fig. 3 is a schematic block diagram of an image acquisition, image archiving and image visualization system according to a first example which enables an imaging system 1 to capture generated image data of internal organs, tissue regions of interest, pathological structures, introduced interventional tools, medical instruments, implants, etc. in the body of a patient to be examined in the form of two-dimensional radiographs or in the form of three-dimensionally reconstructed image data sets to archive and either separately or in the form of registered graphics on the display screen 2 visualize a screen terminal. As an imaging system 1 can be z. As a conventional CT or MRI device, a C-arm X-ray recorder or a two-level Durchlichtuchtungsanalage (Biplansystem) serve.

As in 1 Outlined are those of the imaging system 1 generated image data via an input / output interface 3 an image processing system 4 fed. The image processing system 4 can in addition to a central control device 5 communicating with the imaging system 1 as well as the data exchange between the individual system components of the image processing system 4 controls, including a preprocessing module 6 with a digital filter for noise reduction, contrast enhancement and edge detection. One in the image processing system 4 integrated 2D / 3D image rendering application 7 is used to generate multiplanar reformation or reconstructed 3D views of tissue areas to be displayed using an image visualization tool 9 on the display screen 2 a screen terminal can be graphically visualized.

Apart from the above-mentioned filtering procedure, the preprocessing module performs 6 a procedure for "normalizing" the gray levels of all via the input / output interface 3 obtained image data sets, since it can happen that, depending on the type of scanner used in the imaging system 1 and possible gray-scale distortions in the 2D or 3D reconstruction, both the average gray value and the standard deviation of the gray values in different coregistered image data records of one and the same imaged bone structure may differ slightly. In this context, the term "normalization" is understood to mean a computation procedure in which the gray values of the pixels of each image data set of two image data sets to be registered with one another are matched to one another, the mean gray value (expected value) and the standard deviation of the gray values of the one image data set are adapted to the average gray value or the standard deviation of the gray values of the respective other image data set. In this case, this gray-scale alignment can preferably take place on the basis of one or more bone structures which show no pathological changes (eg due to bone metastases or osteoporosis). A research performing radiologist can be z. For example, on the already registered image data sets, define a volume range of interest (VOI) within which scaling or normalization parameters are determined, which are then applied to the mean value and the standard deviation of the gray values of all imaged bone structures (histogram adaptation). Normalization may also be performed automatically without user interaction on large contiguous bone structures where relatively minor pathological changes do not greatly affect the scaling parameters of mean and standard deviation of each gray level.

Whenever the imaging device BGS generates image data and the image processing system BVS via its input interface 3 can be provided, these are caused by the central control device 5 After completion of preprocessing in preparation for a later graphical visualization, depending on the system configuration, temporarily or persistently in an image data memory of an image archive 8th where they are written to a patient-specific examination log of a log file stored in a memory area of that image library 8th is deposited. In addition to the image data acquired in the context of the imaging process, all the acquisition parameters set manually by the radiologist conducting the examination as well as all presentation and reconstruction parameters used to visualize reconstructed 2D projections or 3D views of specific tissue areas in the patient's body can also be used in a standardized data format (eg in DICOM format) via a data output interface 19 of the image processing system 4 be written to the patient-specific examination log of the log file. For graphical visualization, the stored image data, capture and reconstruction parameters can then be accessed via a data entry interface 10 of the image processing system BVS in a not shown, local temporary memory of the image visualization tool 9 getting charged.

As in 1 presented, has the image visualization tool 9 both access to the record 14 an acquired, filtered and, for example, in two-dimensional rendered form in the image archive 8th reproached reference image of a tissue to be examined as well as on the record 17 one also in the picture archive 8th stored comparison image of the same tissue area, which was acquired at a later date and is also present in a filtered, two-dimensionally rendered form. The two datasets are taken from the image visualization tool 9 loaded, coregistered, linked and with the help of an image visualization tool 9 integrated digital subtractor 11 in the record 12 an associated difference image converted, which on the display screen AB of the screen terminal z. B. is displayed with a suitable color coding for display. Instead of a stored comparison image data record 17 can also directly from a generated in a follow-up, filtered and the 2D / 3D image rendering application 7 output image data set 13 of the tissue region to be examined is used as the comparison image data set and together with the stored reference image data set 14 the digital subtractor 11 the image visualization tool 9 are fed to one on the display screen 2 to calculate the differential image to be displayed.

To perform coreblocking of reference and comparison image VB, the image visualization tool is provided 9 via a digital subtractor 11 upstream registration tool 15 that of a segmentation and clustering module 16 Image data of contiguous segmented tissue areas or image data of grouped together adjacent tissue areas are supplied. The registration tool carries out a rigid coordinate transformation with which the image data of mutually corresponding tissue structures, which are shown in both images, are matched to each other in a segment-specific or cluster-wise manner with a certain registration quality.

In 2 FIG. 3 is a schematic block diagram of an image acquisition, image archiving and image visualization system according to a second example, different from that in FIG 1 sketched embodiment only differs in that instead of the segmentation and clustering module 16 with a downstream registration tool 15 a registration tool 18 is used, which makes a partially rigid registration of the records of reference and comparison image VB. More details below with reference to the in 4 illustrated flowchart described.

mit der 3×3-Rotationsmatrix

beschreibbaren dreidimensionalen Koordinatentransformationen, die zur Überführung der einzelnen Punkte jedes dieser N Punktepaare (P_i(x_i, y_i, z_i), P_i'(x_i', y_i', z_i')) erforderlich sind, der mittlere quadratische Fehler

und der 3×3-Rotationsmatrix

(i ∊ {1, 2, ..., N}), der die Güte der Registrierung von Referenz- und Vergleichsbild VB angibt, minimiert werden. Hierin bezeichnen Δφ ^_x, Δφ ^_y, Δφ ^_z, Δx ^, Δŷ und Δz ^ sechs durch das Optimierungskriterium

mit den beiden hinreichenden Bedingungen

zu optimierende Schätzparameter, die vor Beginn der Registrierung geeignet vorgegeben werden müssen,

ist die Hesse-Matrix der Funktion des mittleren quadratischen Fehlers f ²(p) in einem beliebigen Punkt P eines durch die sechs Parameter Δφ_x, Δφ_y, Δφ_z, Δx, Δy und Δz aufgespannten Vektorraums, p ^opt := [Δφ ^_x, Δφ ^_y, Δφ ^_z, Δx ^, Δŷ und Δz ^]^T bezeichnet einen aus den vorgenannten sechs optimierten Schätzparametern gebildeten optimierten Parametervektor, und das Argument der vorgenannten multivariaten Funktion f ²(p) ist der Parametervektor p := [Δφ_x, Δφ_y, Δφ_z, Δx, Δy, Δz]^T. Die Parameter des optimierten Parametervektors

werden anschließend zur Durchführung der Registrierung in die rechte Seite der zugehörigen Matrix-Vektor-Gleichung aus Formel (1a) anstelle der Parameter Δφ_x, Δφ_y, Δφ_z, Δx, Δy und Δz eingesetzt, um die Bildobjekte des Vergleichbildes zumindest näherungsweise mit den entsprechenden Bildobjekten, die in dem Referenzbild RB dargestellt sind, zur Deckung zu bringen. In 3 a first variant of the method according to the invention is shown in the form of a flow chart. The procedure begins with the loading (S1a) of a CT-assisted radiological preliminary examination and into an image archive 8th of the image acquisition, image archiving and image visualization system stored reference image data set 14 a tissue structure to be examined (eg a bone tissue) and the loading (S1b) of a comparison image data set generated at a later time 17 ("Follow Up" dataset), which is acquired as part of a follow-up examination also carried out under computed tomography imaging and as the reference image dataset 14 after performing optional filtering for noise reduction, contrast enhancement and edge detection in the respective image archive 8th was saved. Subsequently, the image data of both CT image data sets are subjected to a segmentation and clustering algorithm, with the help of which image data of connected image objects segmented automatically or semiautomatically, ie detected as such (S2a), and (if possible) to groups of closely adjacent more similar or similar image objects (clustered) (S2b). Then, the image data of both CT image data sets are coregistered (S3a) and linked (S3b) using a two-dimensional coordinate transformation by means of which corresponding bone structures in the data sets of the reference and comparison images are made coincident with each other. Since the bones of the human skeleton may be assumed to be undeformable in this context, a rigid coordinate transformation (a special case from the set of affine coordinate transformations) can be used as coordinate transformation, in which the coregistration of the two image data sets by translation and / or rotation the congruent to superimposed, corresponding bone structures is made. This can be done either manually (eg by iterative manipulation of the positional and / or angular offset parameters required for carrying out the coordinate transformation, until the corresponding bone structures, which are shown both in the reference image RB and in the comparison image VB, overlap these two images are congruent to each other) or alternatively automatically. In the latter case, from the three-dimensional Cartesian location coordinates of N different point pairs (P _i (x _i , y _i , z _i ) which can be predetermined by a radiologist, P _i '(x _i ', y _i ', z _i ')) ( for i = 1, 2, ..., N), the positions of N in the comparison image VB registered punctiform position markers P ₁ , P ₂ , ... P _N ("Landmarks"), based on a by the orthogonal coordinate axes x, y and z spanned three-dimensional Cartesian object coordinate system K with suitably positioned coordinate origin O and lie in the layer plane of the comparison image xy plane, denote and the positions of N other, in the reference image RB registered punctiform position markers P ₁ ', P ₂ ', ... P _N ', based on an orthogonal coordinate axes x ', y' and z 'spanned three-dimensional Cartesian coordinate system object K' with suitably positioned coordinate origin O 'and in the layer plane of the refer For example, three angle offset parameters Δφ _x , Δφ _y and Δφ _z in ± φ _x -, ± φ _y - and ± φ _z direction from an angle range between 0 ° and 360 ° and three not -negative, real-valued positional displacement parameters Δ _x , Δ _y and Δ _z are calculated in the ± x, ± y or ± z direction. For this purpose, starting from the N by the N matrix-vector equations

with the 3x3 rotation matrix

writable three-dimensional coordinate transformations necessary for transferring the individual points of each of these N pairs of points (P _i (x _i , y _i , z _i ), P _i '(x _i ', y _i ', z _i ')), the middle one square errors

and the 3x3 rotation matrix

(i ε {1, 2, ..., N}), which indicates the quality of the registration of reference and comparison image VB are minimized. Denote here Δφ ^ _x , Δφ ^ _y , Δφ ^ _z , Δx ^, Δŷ and Δz ^ six by the optimization criterion

with the two sufficient conditions

estimation parameters to be optimized, which must be suitably specified before starting the registration,

is the Hesse matrix of the function of the mean square error f ² ( p ) at any point P of a vector space spanned by the six parameters Δφ _x , Δφ _y , Δφ _z , Δx, Δy and Δz, p ^opt : = [Δφ ^ _x , Δφ ^ _y , Δφ ^ _z , Δx ^, Δŷ and Δz ^] ^T denotes an optimized parameter vector formed from the aforesaid six optimized estimation parameters, and the argument of the aforementioned multivariate function f ² ( p ) is the parameter vector p : = [Δφ _x , Δφ _y , Δφ _z , Δx, Δy, Δz] ^T. The parameters of the optimized parameter vector

are then used to perform the registration in the right side of the associated matrix-vector equation from formula (1a) instead of the parameters Δφ _x , Δφ _y , Δφ _z , Δx, Δy and Δz used to at least approximately match the image objects of the comparison image corresponding image objects, which are shown in the reference image RB to bring.

In this context, it should be noted that instead of by f ² ( p ) given similarity measure based on the Least Mean Square (LMS) criterion, of course, other similarity measures for calculating the optimized estimation parameters Δφ ^ _x , Δφ ^ _y , Δφ ^ _z , Δx ^, Δŷ and Δz ^ can be used. It may be z. B. act similarity measures from the field of systems theory, such. B. by a correlation measure.

Instead of a partial registration of reference and comparison image VB after the division of these two images into subregions of predetermined shape and size, it can also be provided according to the invention that after performing a segmentation and clustering of bone structures of a patient, which is recorded in two successive CT images (reference and comparison image), different length and angle offset parameters of mutually independent, regionally different rigid coordinate transformations, which are required for coregistration of the two CT images, are calculated for already segmented bone structures and thus recognized as not belonging together. These parameters are then used as part of a registration procedure in which the individual regionally different rigid coordinate transformations are carried out separately for the segmented subareas.

According to a second variant of the method according to the invention, the registration, as in the flowchart in 4 optionally also combined with the segmentation and clustering algorithm performed in step S3a + b. In this case, according to the invention, a "partially rigid" registration (S2 '+ S3') can be used. Non-contiguous bone structures (eg the two thigh bones or the individual vertebral bodies of the spinal column of a patient to be examined) are identified as not belonging together in that bone structures shown in the reference image as well as in the comparison image VB are subdivided into M non-overlapping subregions (S2 ') and then M independent, regionally different rigid coordinate transformations (one for each subregion) with M locally different, optimized positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δx opt / xj , Δy opt / yy or Δz opt / zj ) and thus M locally different registration matrices R _j (with j ε {1, 2, ..., M}) are carried out separately after these subareas (S3 '). The Tiefindex j denotes the number of the respective subarea. In the further course of the calculation, the individual subareas are then treated separately as after a segmentation procedure. The simultaneous utilization of the segmentation and registration information means that the quality of the results of these two sub-steps (segmentation and registration) is significantly increased. Thus, boundaries between adjacent bone structures, which otherwise could not be identified as separate anatomical objects in a CT image, can be identified on the basis of the different registration matrices R _{j of} two subareas, provided the boundary of these subareas runs along the boundary between the two bone structures to be separated.

After co-registration of reference and comparison image, d. H. After the corresponding bone structures in these two CT images were superimposed as congruently as possible with the aid of the individual rigid coordinate transformations, the gray values of all pixels of the reference image lying in the overlapping region of both images are subtracted from the gray values of all the pixels of the overlapping comparison image lying in the overlap region (S4 ). Within this range, the resulting difference image then only contains the deviations between the superimposed bone structures imaged in the reference and comparison image VB. All other areas are not shown in the difference image.

Since an increase in bone density (osteoblastic course) leads to positive difference values and a decrease in bone density (osteolytic course) manifests itself in the difference image in the form of negative difference values at the corresponding sites of the bone tissue shown, the two different courses can be visualized using appropriate visualization techniques Display screen of a screen terminal (S5). As a visualization technique offers this z. For example, multipanary reformatting (MPR) or a volume rendering technique (VRT or VRT Thin). By reproducing positive and negative difference values, for example by different colors, different profiles in the MPR or VRT can be clearly distinguished from one another. Depending on the difference, the color in question can be z. B. are displayed in different color saturation.

In addition to the color and color saturation values required for the visualization, other, quantitative features, such as, for example, can be obtained from the difference image. For example, the average regional change in Hounsfield density values and / or the change in size (diameter or volume) of varying tissue structures can be extracted. These features can successively from several CT-based radiological follow-up examinations from the difference images between the comparison image VB of the respective follow-up examination and the z. B. generated in a first examination reference image RB are determined (S6a). This allows a quantitative progression representation of the relevant parameters over time, which can then be represented in the form of suitable time diagrams (S6b).

The method according to the invention allows a direct assessment of bone density changes based on CT images generated in tumor patients in clinical practice for interim assessments of the stage and progress of the respective cancer (staging and restaging) and for generating a prognosis of the further course of the disease. In contrast to the conventional examination of pure CT image data, in a differential image of two coregistered CT images taken at different examination times, direct osteolytic and osteoblastic changes of certain bone structures are shown in the form of changes in the Hounsfield density of these tissue areas. This has the advantage of eliminating the tedious and time-consuming search of such regions in acquired CT images, greatly simplifying the radiologist's workflow in performing staging. In addition, with the aid of the method according to the invention, it is also possible to detect smaller structural changes in the bone tissue which are not recognizable to the naked eye in the original CT. The method according to the invention thus strengthens the role of computed tomography for the assessment of the progress of bone metastases and makes additional skeletal scintigraphy superfluous. In addition, the improved visualization also allows for simplified and more reliable therapy planning, e.g. B. for irradiation or for performing surgical procedures.

Claims (13)

Method for supporting a staging or restaging performed under nuclear medical or radiological imaging for the registration of the stage and assessment of the development of spatially and / or structurally variable pathological structures in the interior of a patient to be examined, comprising the steps of: - registering (S3a + b or S2 ' + S3 ') of two digital images recorded at different times and representing the same tissue region, in which image data ( 14 . 17 ) corresponding subregions of the same tissue structures depicted in both images sink together and with the aid of mutually independent rigid coordinate transformations with regionally different positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ), and - converting (S4) the image data of these coregistered partial regions into composable parts of a differential image of the two images to be displayed, characterized in that the regionally different positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) of each individual subarea are determined by means of an optimization criterion based on the method of least squares, according to which the sum of the quadratic distances between the points of a point quantity predetermined on the respective subarea for each subarea of a reference image representing a specific tissue region, and with the aid of the individual, each point of the set of points on the respective subarea of the reference image via a bijective mapping rule exactly one same point of the relevant tissue area indicative point of the set of points on this sub-area calculated locally different rigid coordinate transformations transformed points of a point set assigned subarea of the comparison image is assigned. Method according to Claim 1, characterized in that the subareas of the two images (RB and VB) are non-overlapping image regions which have been created by partitioning these two images (RB and VB) into subregions of a predetermined shape and size and against one another were demarcated, each sub-area of the reference image is assigned via a bijective mapping rule exactly a portion of the comparison image. Method according to Claim 1, characterized in that the subregions of the two images are non-overlapping image regions which have been formed by segmentation of these two images into coherent image objects and have been delimited from each other, each subarea of the reference image being exactly corresponding to a bijective mapping rule Part of the comparison image is assigned. Method according to one of claims 1 to 3, characterized by the step of evaluating in advance of coregistration (S3a + b or S2 '+ S3') of the individual, mutually associated subregions of reference and comparison image calculated regional Lageversatz- (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) with respect to their deviations in sign and magnitude, wherein non-contiguous, mutually corresponding tissue structures, which are shown in subregions of both images, are recognized as not belonging together when a threshold value violation of one of these parameter deviations is detected. Method according to one of Claims 1 to 4, characterized by the step of automatic extraction (S6a) of features from the image data of the calculated difference image for the quantitative assessment of spatial and / or structural changes of the tissue structures depicted in the comparison image with respect to the tissue structures depicted in the reference image. Method according to Claim 5, characterized in that the features extracted from the image data of the calculated difference image are the mean regional change of the Hounsfield density values and / or the change in size of changing tissue structures. Method according to one of claims 1 to 6, characterized in that in the graphic visualization of the difference image pixels with positive difference values are displayed in a different color coding than pixels with negative difference values. Method according to one of claims 1 to 7, characterized in that in the graphic visualization of the difference image pixels with different difference amounts of the same sign are represented with different color saturation values. Method according to one of Claims 1 to 8, characterized by a computation procedure in which the gray values of the pixels of the reference and comparison images are matched to one another such that the mean gray value and the standard deviation of the gray values of the comparison image data set ( 17 ) to the mean gray value or the standard deviation of the gray values of the reference image data set ( 14 ) be adjusted. Use of an image acquisition, image archiving and image visualization system for carrying out the method according to one of claims 1 to 9, wherein the system is characterized by an image visualization tool ( 9 ), which has funds ( 15 respectively. 18 ) for performing registration of two digital images recorded at different times and representing the same tissue region, in which image data ( 14 . 17 ) of corresponding subregions of the same tissue structures depicted in both images are linked to one another and with the aid of mutually independent rigid coordinate transformations with regionally different positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) are coregistered, as well as a computing unit ( 11 ), with the aid of which the image data of the coregistered subregions are converted into composable parts of a differential image of the two images to be displayed. Use of the image acquisition, image archiving and image visualization system according to claim 10, characterized in that the subregions of the two images are non-overlapping image regions which have been created by partitioning these two images into subregions of a predetermined shape and size and have been delimited from one another , Wherein each subarea of the reference image is assigned to a partial area of the comparison image via a bijective mapping rule. Use of the image acquisition, image archiving and image visualization system according to claim 10, characterized in that the subregions of the two images are non-overlapping image regions which have been formed by segmentation of these two images into contiguous image objects and have been delimited from one another, each subregion of the reference image is assigned via a bijective mapping rule exactly a portion of the comparison image. Computer program with program code for carrying out all method steps according to one of Claims 1 to 9, when the program is stored in an image archiving and image visualization system with an image visualization tool ( 9 ), which has funds ( 15 respectively. 18 ) for performing registration of two digital images recorded at different times and representing the same tissue region, in which image data ( 14 . 17 ) of corresponding subregions of the same tissue structures depicted in both images are linked to one another and with the aid of mutually independent rigid coordinate transformations with regionally different positional offset (Δx opt / j , Δy opt / j or Δz opt / j ) and / or angular offset parameters (Δφ opt / xj , Δφ opt / yy or Δφ opt / zj ) and a computing unit ( 11 ), with the aid of which the image data of the coregistered partial regions are converted into composable parts of a differential image of the two images to be displayed.

Images