JP2008036395A - View preparing method for reproducing tomographic image data - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To enable to prepare a view according to various image parameters, especially various image definitions/image noises and various image contrasts with requirement of only small memory demand. <P>SOLUTION: A CT image data is prepared with a three-dimensional modulation transfer function of the CT image data and coordinate information which can derive the coordinates of voxel surrounded by the CT image data in a fixed reference system of an object. From the CT image data, at least one view of an object region is prepared by image filter processing using the modulation transfer function and the coordinate information. A filter algorithm used in the image filter processing and another view parameter required for preparing a view in some case are stored in a view file 7 to which the CT image data is allocated, the view file 7 is called for preparing a new view, and the filter algorithm and another view parameter in some case are newly applied to the CT image data. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a reproducible view creation method for computer tomography (CT) image data of a target region including at least one X-ray attenuation value in each voxel of the target region of interest.

  For example, tomographic image data of a measurement object to be examined, such as a patient, can be obtained by a modern medical imaging method such as X-ray computed tomography. X-ray computed tomography is a special X-ray imaging method that can obtain transverse tomographic images, ie, body tomographic images that are oriented substantially perpendicular to the body axis. The tissue-specific physical quantity displayed in the image is the X-ray attenuation value distribution μ (x, y) in the cross section. In addition to these transverse tomographic images, a volume image of the detected target region can also be obtained, and the volume image means a three-dimensional distribution μ (x, y, z) of attenuation values. Each volume element (also called voxel) in the detected target area is assigned an X-ray attenuation value.

  Computed tomography (CT) images can be created by a C-arm device as well as by a conventional computed tomography device having an endless rotating scanning system. In the following, the term “CT” is used for both types of imaging equipment. The CT image data is calculated from the raw data supplied from the CT apparatus by a reconstruction algorithm including a convolution process using a special convolution kernel (Kernel). The mathematical form of the convolution kernel can affect the image quality exactly when reconstructing the CT image from the raw data. For example, high frequencies can be emphasized to increase the spatial resolution in the image with a suitable convolution kernel, or damping can be applied to reduce image noise with a convolution kernel of other nature. Thereby, during image reconstruction in computed tomography, the image characteristics characterized by image sharpness / image noise and image contrast can be influenced through the selection of convolution kernels.

  CT images must now be calculated from the manufacturer and product specific raw data, especially on high power image reconstruction computers developed for that purpose. In this case, the raw data is stored in a manufacturer specific format and does not include pixel data or voxel data. Thus, the raw data cannot be displayed directly and is only useful for the reconstruction of images or image series.

  However, users often require different views of the region of interest detected by different image parameters. Conventionally, for this purpose, an arbitrarily large number of tomographic image series can be generated from raw data in different settings by reconstruction of the raw data, for example different nuclei, different slice thicknesses, different increments, different FoV (= Field). of view, imaging field of view) or in various orientations. However, these parameters and the accompanying reconstruction algorithm are closely related to the CT apparatus that detected the target region, and therefore differ from manufacturer to manufacturer and from device type to device type. Therefore, the desired tomographic image can be reconstructed from the raw data without any problem only in the CT system in which they are acquired.

  These series of tomographic images form the basis for post-processing, record keeping, transfer to other image processing or imaging equipment, and ultimately diagnostic findings. The so-called DICOM (= Digital Imaging and Communications in Medicine) standard is used for the transfer. With this DICOM standard, images and data of various imaging devices and image processing devices can be exchanged with each other. However, the supply of various image series requires not only raw data memory demand, but also memory demand for image series, which can unconditionally be several gigabytes. Thereby, the transfer of the image series to the CT application for post-processing is also very time consuming and has a detrimental effect on the hospital workflow. Reconstruction of various image series from raw data is likewise time consuming. The accumulation of raw data in a CT system for later reconstruction purposes requires a large amount of memory and is therefore very limited in time for cost reasons. In general, raw data is only accumulated for a few days today. In order to extend the accumulation time, the raw data may be stored, for example, on an optical storage medium. However, the record keeping process and the import after the raw data are similarly time consuming and therefore very little in practice.

  In order to create a reproducible view of CT image data, a so-called 2D presentation state (2D display state) is known from the DICOM standard, which includes a range of values to be displayed, so-called windows, pixel data. Image crops and digital zooms to pixel data are stored. The so-called grayscale, color and blending presentation state (mixed display state) is a linear registration with storage of a color / gray value coding table and conversion for two image series (PET-CT, SPECT-CT). Allows storage of a rotation / transform matrix for A variable impression, ie a data view related to the organ of the volume or volume part (3D object) with different image sharpness / image noise or different image contrast is not possible.

  The object of the present invention is to recreate tomographic image data that requires little memory demand and can create views with different image parameters, especially different image sharpness / image noise and different image contrast Is to provide a simple view creation method.

According to the present invention, according to the present invention, computer tomography (CT) image data of a target region of interest including at least one X-ray attenuation value for each voxel of the target region is converted into three-dimensional CT image data. A modulation transfer function or a function capable of deriving a three-dimensional modulation transfer function and coordinates of a voxel surrounded by CT image data are generated together with coordinate information that can be derived in a fixed reference system of interest, and the modulation transfer is generated from the CT image data. At least one view of the region of interest is created by image filtering using the function and the coordinate information, and the filter algorithm used for image filtering and possibly other view parameters required for view creation are CT images. The data is stored in the allocated view file and can be used to create a new view. The view file is called, in some cases as well as filter algorithm is solved by other views parameters are applied to the new CT image data.
The excellent embodiments of the present invention are listed as follows.
(1) By applying different filter algorithms for different applications, different views of the target area in which at least one of image sharpness, image noise and slice thickness of the underlying tomographic area are different are created, A dedicated view file is stored for each different view.
(2) As another view parameter, the spatial orientation of the object selected for the view is stored in the view file.
(3) Information regarding the image cropping selected for the view is stored in the view file as another view parameter.
(4) Rendering parameters selected for volume rendering are stored in the view file as other view parameters.
(5) As other view parameters, the position and extent of one or more volume segments selected for the target view are stored in the view file.
(6) As other view parameters, the flat and / or curved 2D tomographic planes for the key image are stored in the view file.
(7) CT image data is prepared as a volume data set.
(8) A reconstruction algorithm for high spatial resolution is used when creating CT image data.
(9) Three-dimensional modulation transfer function of CT image data or CT image data having a function capable of deriving a three-dimensional modulation transfer function and coordinate information is prepared in a common data set.
(10) A common data set is used as a standard DICOM (Medical Digital Image and Communication) object.
(11) A view file is used as a standard DICOM object.

  In the method according to the invention, the computed tomography (CT) image data of the target area of interest includes at least one X-ray attenuation value for each voxel of the target area, and at least one X for each voxel of the target area. Computed tomography (CT) image data of a target region of interest containing a line attenuation value is surrounded by a CT image data and a 3D modulation transfer function of CT image data or a function that can derive a 3D modulation transfer function and CT image data Are generated together with coordinate information that can be derived in a fixed reference frame of interest. From the CT image data, at least one view of the region of interest is created by image filtering using the modulation transfer function and coordinate information, and the filter algorithm used for image filtering and possibly required for view creation Other view parameters are stored in the view file assigned the CT image data. The view file is then called to create a new view, and the filter algorithm and possibly other view parameters are newly applied to the CT image data.

  The method according to the invention makes use of the preparation of CT image data and the preparation of additional information including the three-dimensional modulation transfer function of CT image data and coordinate information. CT image data is generated and stored by X-ray imaging, that is, using a CT apparatus, as is well known. The raw data generated upon detection of the region of interest is first reconstructed by an image reconstruction computer, preferably with a suitable convolution kernel that creates a high resolution image. This should be done directly at the data creation site, i.e. by the image reconstruction computer of the CT device recording the data. The raw data can then be discarded. The CT image data is preferably stored in a common data set together with a corresponding coordinate information and a function capable of deriving a three-dimensional modulation transfer function of the image data or a three-dimensional modulation transfer function. This data set forms the basis for later post-processing, record keeping, transfer to other devices or stations, and finally for diagnosis. The data set contains all the information to create an arbitrary view of the region of interest from the data (hereinafter also referred to as the data view) at any other workstation, along with the 3D modulation transfer function and coordinate information. .

  With the knowledge of the modulation transfer function of CT image data, pure image filtering with an appropriate filter can achieve the same results as obtained from conventional reconstruction from raw data. Based on the knowledge of the 3D modulation transfer function, images having various sharpnesses and various noises can be created by using various filters of CT image data. This type of image filtering is known from German Offenlegungsschrift DE 10238322, the disclosure of which can be incorporated into this patent application in this respect. Of course, the aforementioned advantages are also obtained when the prepared data or information is not stored in a single data set, but is divided into multiple data sets, but from these data sets, The association must be recognizable.

  In the method according to the invention, at least one desired view of the region of interest is created from the CT image data prepared with the additional information by image filtering using the modulation transfer function and the coordinate information. This view may be a data view, in particular a tomographic or slice image, having different sharpness, different noise, different orientation, different image clippings and different slice thicknesses of the CT image data. The filter or filter algorithm used in this case for filtering the CT image data and other view parameters such as image clipping or image orientation are stored in the view file assigned to the CT image data. This view file is hereinafter referred to as an Advanced Presentation State (APS) as well as a 2D presentation state known from the DICOM standard. Thus, APS represents any view based on CT image data that may exist as a volume data set or tomographic image stack. APS replaces today's reconstruction. Because APS no longer needs raw data, it simultaneously stores all results of the application in a standard format. With the storage of this parameter, in particular the filter algorithm used, the corresponding view can be recreated reproducibly on any device at any time. Similar to the tomographic image stack or volume data set, the APS is also stored as a DICOM object in a DICOM archive (eg PACS = medical image management system). Thereby, the user can create all image views important for diagnosis on any DICOM workstation with very little memory requirement. All interactions with CT image data that determine diagnostically important image data may be stored in the APS.

  Today's applications typically manipulate raw data sets, ie raw data, with image clipping, slice thickness, filters, color coding, spatial orientation, calculate measurements, and segment to limit organs and blood vessels Render volumes to generate 3D effects in 2D views, quantify image results, or define paths for virtual camera guidance through the body (eg, colon, bronchi). The CAD algorithm (CAD = Computer Aided Diagnosis) marks the volume segment and classifies it as ill or healthy. All these results are stored in the APS to obtain exactly the same view as once created for other users or at other data stations at a later time. For this reason, a new reconstruction from the raw data is no longer necessary. This is because the method according to the invention uses a CT image data set that has been reconstructed once and stored together with the 3D modulation transfer function and coordinate information.

  This CT image data set with additional information forms a database for all parameters stored in the APS. Preferably, this CT image data set is preferably a volume data set, which is hereinafter also referred to as IVB (= Image Volume Database, image volume database) together with additional information. IVB replaces CT raw data specific to today's manufacturers and products during post-processing, record keeping and transfer to other image processing or imaging equipment. The IVB is stored in the new DICOM standard format. A possibility for this lies in the extension of today's DICOM-CT multiframe. In this case, since the IVB replaces the CT raw data, the CT raw data is not required long after a one-time reconstruction of the IVB. In this case, the IVB includes volume data having a prescribed coordinate system, not a tomographic stack from various slice images. In the IVB, in addition to CT image data, when a CT examination topogram is detected, this type of topogram is also stored.

The coordinate information from which the coordinates of the voxel surrounded by the CT image data of the target area can be derived in the fixed reference system of the target is composed only of the coordinate value for one of the voxels and the direction information and distance information for the remaining voxels. It is preferable. Since voxel data is generally reconstructed on an equidistant raster, the voxel coordinates are expressed as direction vectors (eg, unit vectors e _x , e _y , e _z ) and increments or deltas (eg, d _x, d _y, can be defined by a d _z). To describe the voxel aperture, for example, three one-dimensional modulation transfer functions in the x, y, and z directions are used, and these constitute a 3D modulation transfer function. As an alternative to the modulation transfer function, of course, a point image function or sensitivity profile may be used that can derive a 3D modulation transfer function. This transfer function knowledge allows for later modification of image definition / slice thickness for each CT image. The 3D modulation transfer function describes the mapping quality of the region of interest to the CT image data depending on the imaging system itself, ie the CT apparatus and possibly the reconstruction algorithm used. The modulation transfer function can be determined for any CT system, for example, stored as a value table after digitization.

  According to the method according to the invention, for different applications, different views of the target area differing in at least one of image sharpness, image noise and slice thickness of the underlying slice of the target area by using different filter algorithms. Can be created. For each different view, a dedicated view file, i.e. a dedicated APS, is preferably stored. The APS contains information about the relevant basic data set, ie the volume data set (IVB) or the fault stack, as well as the applied filter algorithm.

  As other view parameters, for example, image clipping (FoV = Field of View, imaging field of view), orientation in space and / or VRT presets (rendering parameters, color table) may be stored in the APS. In addition, any number of 3D volume segments in the coordinate system of the volume data set, guided views, eg passages for the colon / bronchial fly-through (Fly-Through), or so-called key images, It can be stored in the APS as a 2D tomographic plane (or tomographic curved surface) directed to a defined rotation in the coordinate system. Additionally, markers, annotations or distance / volume measurements performed in the view may be stored in the APS. The marker may be stored as 2D or 3D information, for example as a point, as a ROI (= Region of Interest) or as a VOI (Volume of Interest).

  According to the method according to the invention, image series with different quality characteristics no longer have to be individually reconstructed from raw data. Each desired image impression may be displayed directly to the user. As a result, there is no waiting for the result that has been necessary for the reconstruction in the past, and there is no quality check first performed to confirm the desired image quality. In addition to significant time savings when creating new views, the data volume can be drastically reduced. This is because, for example, a thin slice series, which conventionally forms the input for the post-processing step with the highest priority, no longer needs to be reconstructed from raw data, but rather only stored as an APS. APS allows the user to use the stated possibilities of real-time filtering in the axial direction (equivalent to today's reconstruction kernel) and the z-direction (slice thickness and reconstruction increment) at any time after data acquisition. All interactions to the volume that determine diagnostically important image data may be stored in the APS. In this case, in addition to the definition of the image series, the marking of individual images as well as the input of annotations or findings may be made possible.

  As an additional advantage over previously available 2D presentation states, the proposed APS can also store 3D segments for separating tumors, vascular trees, and the like. In addition, any number of straight and curved sections with arbitrary orientations in the coordinate system of the volume data set can be stored in the APS, and which one can be used to address a particular 2D section as a key image It can be read from the application accordingly.

  An apparatus for performing the method includes a computing unit having an interface to a data bank in which CT image data is preferably stored as IVB. The image creation module accesses the desired APS in the data bang, reads it, and processes the CT image data according to the parameters specified in the APS. In addition, an APS creation module is provided, which stores the filter algorithm and possibly additional view parameters applied to the image data by the user as APS in the data bank. In this way, a new APS can be created from the original CT image data set through user interaction, or a view can be displayed based on a known APS.

In the following, the method according to the invention will be described in detail again with reference to the drawings, without limiting the scope of protection set by the claims.
Figure 1 shows an overview of the preparation of tomographic image data,
FIG. 2 shows an example of an application server for creating or reading an APS,
FIG. 3 shows an example of a central memory with IVB and APS,
Figure 4 shows an example of APS creation and storage,
FIG. 5 shows a first reference example of a 3D segment,
FIG. 6 shows a second reference example of a 3D segment,
FIG. 7 shows a third reference example of a 3D segment,
FIG. 8 shows a fourth reference example of a 3D segment,
FIG. 9 shows examples of modulation transfer functions as well as filter functions that can produce the desired image impression.

  FIG. 1 shows an overview of a workflow when creating a view of tomographic image data according to the method of the present invention. First, computer tomography (CT) image data of a target region is recorded by the computer tomography (CT) apparatus 1. The CT apparatus 1 may be a computed tomography apparatus or a so-called C-arm apparatus. In order to reconstruct an image from a CT scan, the measured raw data is transferred to the image computer 2 and a 3D image data set is created by the image computer 2 from the raw data using an appropriate reconstruction algorithm. This 3D image data set is a volume data set in this example. In the volume data set, the gray value derived from the X-ray attenuation value μ (x, y, z) or the X-ray attenuation value μ (x, y, z) for each volume element (voxel) in the examined area. Is assigned.

  From this volume data set, a data set having additional information including a 3D modulation transfer function of CT image data and information related to voxel coordinates of the CT image data set is created. This new volume data set, i.e. IVB3, is transmitted to the central data bank 4 and stored there. The central data bank 4 is connected to an application server 5, where a desired data view is created from the IVB 3 and transmitted to the corresponding client 6.

  The central data bank 4 also stores an APS 7 attached to the data view and is assigned to each IVB 3. This can be recognized in FIG. FIG. 3 schematically shows three IVB3 with an exemplary three APS7. Of course, the number of APSs is not limited and only relates to the number of various data views desired. In the central range of central data bank 4, in this example, based on IVB 3a of the patient's chest, one APS 7a contains a representation of the pulmonary nodule, the other APS 7b contains a representation of the extracted heart, Contains data views specially created by the user.

  FIG. 2 shows an example of creating two APSs 7. For this purpose, the application server calls in the first application one of the APSs stored in the data bank 4, reads it and applies the parameters and filters contained therein to the attached IVB3. The view is displayed on the corresponding client 6. The user can change this view by setting other image parameters. For this purpose, the user can use the appropriate tool 11 and function 12. After adjusting the view with the desired image parameters, the image parameters required for this view, in particular the filter functions that are possibly changed, as well as other view parameters, are stored in the new APS 7 in the central data bank 4 by the application server. . The same method is performed by a second application that produces a different APS7. An APS can be automatically created by a corresponding application that requires a predetermined image parameter. This can be replaced, for example, for automated pulmonary nodule assessment with specific requirements for resolution, image sharpness and noise to ensure reproducible results. This type of application automatically applies the appropriate image parameters to the IVB by calling the attached APS, and then segmentation is performed based on the created view.

  FIG. 4 shows yet another example for adjusting an existing APS to create a new APS. In this case, the application in the application server 5 calls an existing APS for the IVB in order to provide data in a form that can be processed by the application. Subsequent processing includes filtering, windowing, and data reduction to a predetermined 2D surface or 3D volume segment. All necessary view parameters, especially orientation, position, zoom factor, filter, segmentation, marker, color / gradation, etc. are stored as a new APS or yet another APS for the resulting view. .

  FIG. 5 shows a reference example of a 3D volume segment in one APS. In this simplest case, a predetermined z direction (system axis direction) is used during CT imaging, and this direction can be derived from coordinate information in IVB. The specified 3D segment in the volume data set is responded by the designation of the start and end coordinates in the z direction. This designation restricts the data created for the view to a defined area, eg surrounding the heart. Subsequently, appropriate filters are applied to these reduced data, possibly to create the desired view for subsequent processing.

  FIG. 6 shows another example, in which the 3D coordinates of the center of the desired 3D volume segment and its extent in the x, y and z directions are specified.

  FIG. 7 shows another example of this type in which the 3D volume segment is formed in a spherical shape. The segment is defined here by specifying the three-dimensional coordinates of the center of the volume segment and the radius.

  Finally, FIG. 8 shows a reference to a segmented organ or segmented region within a region of interest surrounded by CT image data. Again, the coordinates of the center of the segmented organ or segmented region are specified for reference.

  The last four examples clearly reduce the amount of data to be processed. This is because only the segmented range or reference range data specified in each APS is displayed. The exact same view is displayed to the user on each new APS call.

  FIG. 9 shows by way of example a one-dimensional modulation transfer function 8 and a filter function 9 that produces the image impression desired by the user, by way of example. The modulation transfer function 8 is included in IVB as a table. The image impression desired by the user corresponds to the function displayed in the diagram by the curve 10. From this desired image impression (noise, sharpness) or function and the modulation transfer function 8 contained in the IVB, it is necessary to reach the image impression or related function 10 from the modulation transfer function 8. A filter function 9 is calculated. The filter function 9 is preferably stored in the APS as a table with a few interpolation points. This APS call allows different users or the same user to always produce the same image impression from the IVB by calling the APS without having to perform time-consuming image reconstruction for this purpose. it can.

  The method according to the invention makes it unnecessary to reconstruct a large number of image series. This saves time and optimizes the hospital workflow. Image series no longer need to be completely reconstructed from raw data. Instead, only a slightly more logical data view is created in the form of APS. A considerable data reduction is thereby achieved. This is because APS requires very little memory. Image data (view in IVB by APS) necessary for diagnosis is created and exchanged regardless of manufacturer and device. The preparation of the image data necessary for diagnosis is no longer required in the CT system itself, but can be performed in any workstation. Loading these data into the CT application for time-consuming transfer and post-processing of the image series is omitted.

Schematic showing an overview of the preparation of tomographic image data according to the present invention Schematic of application server for creating and reading APS Schematic showing an example of a central memory with IVB and APS Schematic showing an example of APS creation and storage Explanatory drawing which shows the 1st reference example of 3D segment Explanatory drawing which shows the 2nd reference example of 3D segment Explanatory drawing which shows the 3rd reference example of 3D segment Explanatory drawing which shows the 4th reference example of 3D segment. Diagram showing examples of modulation transfer functions and filter functions that can produce the desired image impression

Explanation of symbols

1 CT device 2 Image computer 3 IVB
4 Central data bank 5 Application server 6 Client 7 APS
8 Modulation transfer function 9 Filter function 10 Curve 11 Tool 12 Function

Claims (12)

Computed tomography (CT) image data of a target region of interest that includes at least one X-ray attenuation value for each voxel of the target region is converted into a three-dimensional modulation transfer function (8) of CT image data or a three-dimensional A function capable of deriving the modulation transfer function (8) and the coordinates of the voxel surrounded by the CT image data, together with coordinate information that can be derived in a fixed reference frame of interest,
From the CT image data, at least one view of the region of interest is created by image filtering using the modulation transfer function (8) and coordinate information;
The filter algorithm used for image filtering and possibly other view parameters necessary for view creation are stored in the view file (7) to which the CT image data is allocated,
A view file (7) is called to create a new view, and a reproducible view of tomographic image data characterized in that a filter algorithm and possibly other view parameters are newly applied to the CT image data. How to make.   The application of different filter algorithms for different applications creates different views of the target area that differ in at least one of image sharpness, image noise, and slice thickness of the tomogram underlying the target area. 2. Method according to claim 1, characterized in that a dedicated view file (7) is stored for each.   3. The method according to claim 1, wherein the spatial orientation of the object selected for the view is stored in the view file as another view parameter.   4. Method according to one of claims 1 to 3, characterized in that as other view parameters information about the image crop selected for the view is stored in the view file (7).   5. Method according to one of claims 1 to 4, characterized in that as other view parameters, the rendering parameters selected for volume rendering are stored in the view file (7).   6. The position and extent of one or more volume segments selected for the subject view as other view parameters are stored in the view file (7). The method described in 1.   One or more flat and / or curved 2D tomographic planes for the key image are stored in the view file (7) as other view parameters. The method described in 1.   8. The method according to claim 1, wherein the CT image data is prepared as a volume data set.   9. The method according to claim 1, wherein a reconstruction algorithm for high spatial resolution is used when generating CT image data.   The CT image data having the coordinate information and the three-dimensional modulation transfer function (8) of the CT image data or the function capable of deriving the three-dimensional modulation transfer function (8) are prepared in the common data set (3). 10. Method according to one of the preceding claims, characterized in that   11. Method according to one of the preceding claims, characterized in that the common data set (3) is used as a standard DICOM (medical digital image and communication) object.   12. Method according to one of claims 1 to 11, characterized in that the view file (7) is used as a standard DICOM object.