WO2011106820A1

WO2011106820A1 - Method and device for processing ct image data of a body region of a patient containing bones

Info

Publication number: WO2011106820A1
Application number: PCT/AT2011/000111
Authority: WO
Inventors: Helmut Ringl; Daniel Toth
Original assignee: Medizinische Universität Wien
Priority date: 2010-03-04
Filing date: 2011-03-04
Publication date: 2011-09-09
Also published as: AT509233B1; AT509233A4

Abstract

The invention relates to a method and device for processing three-dimensional image data of a body region of a patient containing bones (K) generated by means of a computer tomograph (CT), wherein the 3-D image data contain the location-dependent radiation attenuation of the body region, wherein those image data that exceed a specified threshold (A_S1) of the radiation attenuation are determined as the surface (0) of the bones (K) within the body region, and the radiation attenuation of said image data determined as the bone surface (0) and of a number (n) of image data adjacent thereto is preferably reduced to zero.

Description

A method and apparatus for processing CT image data of a bone-containing body region of a patient

The invention relates to a method for processing three-dimensional image data of a bone-containing body region of a patient generated by means of a computer tomograph, the 3D image data containing the respective location-dependent radiation attenuation of the body region.

Likewise, the subject invention relates to an apparatus for performing such an image processing method.

For the sake of completeness, it is noted that the subject invention is directed to the processing of CT images of both humans and animals.

In particular, the subject invention serves to improve the representation of bone metastases and other pathological changes of the human or animal skeleton. One of the most common locations for the establishment of cancer cells (metastases) is the bone marrow, which mainly affects the axial bone skeleton with spine, pelvis and ribs. Bone metastases and other pathological changes in the skeleton of humans and other mammals are primarily detected in the axial plane by conventional image-cutting techniques, such as computed tomography. New generations of so-called multi-detector computed tomographs offer a steadily increasing spatial resolution along the craniocaudal axis of a patient with a proportionately increasing number of axial images. The detection or detection of pathological changes in the bone distributed over several thousand images per examination is particularly difficult and time-consuming for the attending physician, associated with a considerable rate of overlooked metastases. A timely detection of pathological changes, however, is essential for the prognosis and the all ^¬ due further therapy of patients.

Bones of mammals consist essentially of two main components, the so-called corticalis and the cancellous bone, the interior of the bone. The corticalis has usually Cherweise a thickness of 1 to 6 mm and is characterized by particularly dense material. The cancellous bone, however, is spongy and has a correspondingly lower density. The spongiosa contains the bone marrow, where mainly bone metastases accumulate. Due to the high density of the cortical bone, the edges of the bones appear particularly bright in 3D reconstructions or projection methods of CT image data, making it difficult to detect pathological changes in the interior of the bone that appear relatively dark in these imaging procedures. However, these methods (3D visualization or projection method) are useful for shortening the time of diagnosis or facilitating the detection of pathological changes.

In principle, a distinction is made between two types of bone metastases. Osteoplastic bone metastases are characterized by additional bone structure and are therefore denser than the surrounding cancellous bone. Osteolytic metastases have localized bone loss and are therefore less dense than the surrounding cancellous bone. To represent both types of bone metastases different reconstruction methods are necessary.

By appropriate processing of the CT image data, significant improvements can already be achieved today. In ^¬ example, additional three-dimensional reconstructions of the image data can be made, which facilitate, for example, the diagnosis of fractures that are recognizable on the surface of the bones. However, changes that are inside the bone are not detected by such measures and can not be displayed by such methods. Zu ^¬ addition, the preparation of three-dimensional reconstructions, especially if the data must be manually segmented before the reconstruction to allow a representation of the changes at all, time-consuming and computer-consuming and associated with additional costs.

The DE 10 2005 036 999 Al describes a device for automatic detection of abnormalities in medical Bildda ^¬ th, wherein by means of a determining module, the body region, the is detected in order to facilitate the diagnosis by the radiologist.

US Pat. No. 6,021,213 A describes an image processing method in which the corticalis and spongiosa of a bone are segmented by different algorithms and used for bone density measurement in osteoporosis therapy.

In DE 10 321 722 AI a method for detecting metastatic bone lesions is described, wherein a bone image from a double or Mehrfachenergiebildsat z is used and the bones are segmented from a background of the bone image.

Finally, DE 10 2006 048 190 A1 also describes a method for improved viewing of rib metastases, wherein the ribs are automatically marked and changes in a structure of the ribs are automatically detected.

The object of the present invention is to provide an above-mentioned image processing method and an entspre ^¬ sponding device whereby containing an improved projection or 3D representation of CT images of a bone body portion of a patient is made possible. In particular, an improved detection of bone metastases is to be achieved and the time required to recognize these changes is shortened. The required calculation and Kostenauf and should be as small as possible. Disadvantages of the prior art should be avoided or at least reduced.

The invention is achieved by an above-mentioned image processing method, wherein those image data are determined as the surface of the bones within the body region, which exceed a predetermined threshold of radiation attenuation, and the radiation attenuation of this determined as bone surface image data and a number of adjacent image data is preferably reduced to zero , Thus, according to the invention, the dense outer shell of the bones is detected and eliminated from the CT image data. The number of image data adjacent to the image data obtained as a bone surface, the are to be deleted, is chosen according to the expected thickness of the corticalis. For example, 2 to 10 image points can be deleted or their radiation ^attenuation can be correspondingly preferably reduced to zero. In addition, the radiation attenuation of preferably all pixels that are outside the bone is also set to zero. Pixels with a radiation attenuation equal to zero are then not displayed in the following projection method or 3D method, or this value is ignored by the projection. The objective procedure is performed after the CT scan and does not require an extension of the duration of the examination or an additional radiation exposure of the patient. In the resulting image data, the corticalis which is troublesome for the diagnosis of bone metastases or other pathological changes within the spongiosa is correspondingly eliminated or suppressed, so that with conventional reconstruction methods a faster and more secure detection of bone metastases or other changes in the bone is made possible. From the processed image data, one or more overview images can be generated, in which both osteolytic and osteoplastic bone metastases can be recognized quickly and reliably. There are no additional processing steps from radiological technological side necessary, but only minor processing power to revise the image data. As a result, there is at least one image, preferably for the entire body region shown, which allows the radiologist to detect pathological bone changes in a few seconds and to significantly increase the detection rate of the changes in comparison to axial raw images according to current practice. In addition, the subject method can reduce the necessary experience of the attending physician for the safe and rapid detection of pathological changes. Since patients have different bone density and can show different contrast agent staining behavior, it may be useful to calculate a density profile in two or three dimensions before the actual calculation in order to define specific adjusted threshold values for the entire patient or even for small volume areas of the patient. These thresholds can be calculated from the number of pixels within the entire patient or a small volume. which are above a threshold value (for example 500-1500 Hounsfield units). Image data representing a ^¬ be Sonders high radiation attenuation, .for example, above 2000 HU, (for example, metal due to prostheses) which will not be considered to exclude a distortion of the calculation.

The image processing method can be performed directly on the three-dimensional CT image data. Alternatively, the 3D image data may be divided into a plurality of 2D image data, and the 2D image data may be processed successively. It is particularly advantageous for the detection of bone metastases, if the two-dimensional image data with a layer thickness between 0.6 and 3 mm are present. Such a low layer thickness, and preferably a very small reconstruction interval of 0.4 to 2 mm, can be used to achieve optimal reconstruction.

According to a feature of the invention, it is provided that those image data are determined which ^fall below a predetermined ^threshold value of the radiation attenuation and the radiation attenuation of this determined image data is reduced. In this way, irrelevant soft parts can be eliminated from the CT image data for the subject method. Similarly, disturbances from the image data, such as individual pixels with particularly high radiation attenuation, can also be eliminated from the image data or their radiation attenuation preferably reduced to zero. Of course, the image data can also be ^preprocessed with other conventional filter methods. It has been shown, for example, that the removal of soft tissue from the image data is difficult when intravenous contrast ^mediums have been administered prior to the CT scan. In this case, it may be appropriate to pre-segment the bone and to eliminate any image data that does not correspond to the detected bone. In this way, all non-interested soft parts outside the bone can be deleted from the image data. The pre-segmentation of the bone may be done by a per se known region growing algorithm, selecting a very generous threshold for the radiation attenuation be made. This ensures that the pre-segmentation of the entire bone with corticalis and possibly a few minor adhering to the bone soft tissue remains. In this pre-segmentation by the bone prepared ^¬-adjusted image data image processing method according to the invention is then applied.

The determination of the bone surface or corticalis can be effected by determining those image data from the 3D or 2D image data in which the difference between the value of the radiation attenuation and the value of the radiation attenuation of the preceding image data exceeds a predetermined threshold value. In this way, density differences or density jumps that occur at the boundary layer between the soft tissue and bone are detected. , As mentioned above Of these, detected as bone surface image data, a predetermined number ^¬ An image data following eliminated or their radiation attenuation preferably reduced to zero. So the corticalis is removed from the picture. Experience has shown that the bone surface can be identified first and then the corticalis can be deleted from the image data according to the given thickness. Likewise, this processing can also be performed during the calculation on the original image data. The processed image data set thus only contains the spongiosa areas, possibly cortical bone remains and preferably no soft tissue. An image data record prepared by this method can preferably be optimally represented by a so-called maximum intensity projection (MIP) reconstruction.

Alternatively to the above method also at least a starting point within the 3D image data, or any 2D image data may be selected and from each output point from all 3D image data of the volume or all 2D image data of each layer to each of ^¬ transition point that image data is determined as the bone surface, which exceed a predetermined threshold value of the radiation attenuation, and the radiation attenuation of this image data determined as a bone surface are preferably reduced to zero. This so-called filling algorithm is based on one or more starting points, from which of the environment (for example, the surrounding 8 pixels) is then examined to see whether a certain threshold in the radiation attenuation is exceeded. If this is the case, the pixel is deleted or its radiation attenuation reduced or set to zero, so that it is no longer visible in a MIP reconstruction. Thereafter, the examination is continued in the square or in the circle around the starting point until the entire volume of image data or the entire layer of image data has been processed. In order to prevent superimposition of different bones, individual bones can also be detected and displayed separately by an automatic limit-based filling algorithm or a shape-based algorithm. For example, in the case of image data of the upper body of a patient, it makes sense to segment the scapula initially and to delete it from the main data set. Otherwise, the scapula would overlap the ribs in the presentation and no starting points would be set between the ribs and the scapula because the two bones are very close together. Due to a possible partial lack of corticalis of the ribs, a deteriorated presentation may thus result.

In addition, it is advantageous if the values of the radiation attenuation of all image data outside the image data determined as the bone surface are preferably reduced to zero. In this way, the areas outside of the bones that are irrelevant to the detection of bone metastases or other pathological changes within the cancellous bone can be eliminated.

Finally, the detection and elimination of the corticalis can also be effected by selecting at least one starting point ^within the 3D image data or each 2D image data and from each starting point all 3D image data of the volume or all 2D image data of the layer one behind the other along a line are ana ^¬ lysed, and those image data are determined along this line as a bone ^¬ chenoberfläche which a predetermined

Exceed threshold of radiation attenuation. The Strah ^¬ lung attenuation of these determined as bone surface data ^¬ image and the corresponding number of the following picture data is pre- preferably zero, and thereafter the image data of the volume or layer is analyzed along another line until finally all the image data has been analyzed. In this so-called intelligent circular beam method, a line is drawn from each starting point and each pixel is scanned along this line and, finally, when a certain radiation attenuation is exceeded, this pixel is identified as corticalis. In this case, the corticalis is erased along a predetermined thickness and the pixels in front of the corticalis are also eliminated. Thereafter, another line is drawn in a different direction from the same starting point and the pixels provided along this line are also processed. Thus, each pixel in the course of the line from inside to outside is set to zero when the pixel exceeds a certain threshold value, which amounts to recognition of the corticalis. After deleting the corticalis, the deletion mode on this line is ended and the next line is started. This corticalis erase mode is also terminated when the radiation attenuation or the density value of the pixel being examined falls below a certain threshold value. This is an indication that the end of the corticalis and thus the cancellous bone is reached. This prevents the adjacent spongiosa from being erased if there is a thin corticalis. If a sufficient number of starting points are selected, an image results within a very short time and without high computing power, which only shows the cancellous bone of the bones of the body region shown, in which metastases can be recognized more quickly and reliably. An improvement can be achieved by adapting the threshold value of the radiation attenuation from which the pixel is identified as corticalis to the local density ratios. In this case, a local calculation of the threshold value of the radiation attenuation by density measurement of the surrounding pixels can be carried out for each individual line along which the pixels are scanned.

Preferably, from each starting point all image data of the volume or layer are analyzed in succession along a plurality of lines, preferably at a regular angular distance of 1 ° to 10 ° to one another. Advantageous it has further proved when the situation will ever ^¬ the starting point determined randomly. Alternatively, the starting points can be set specifically. By such a deliberate setting of the starting points, the length of the extinguishing rays emanating from the starting points can be reduced and with the same number of extinguishing jets a significantly improved speed and uniformity of the extinction can be achieved.

The number of detected as to the bone surface Bildda ^¬ th adjacent image data whose radiation attenuation is reduced to zero, preferably, may be chosen constant, for example 3 to 4. In this way, regardless of the illustrated body region and the respective patient always the same thickness of the corticalis is eliminated from the image data.

Since the thickness of the corticalis depends not only on the particular bone but also, for example, on the age of the patient, it is advantageous if the number of image data to be removed from the surface of the bone is adjustable. In this case, an adjustment of this number n, which is valid for all image data, for example in the range between 3 and 5, can be carried out.

Even better results are obtained when the number n is set depending on the spatial position of the image data, preferably automatically angren ^¬ collapsing to the determined as a bone surface image data is image data, the radiation attenuation should be reduced to preferably zero. In this way, the anatomical situation can be taken into account and, for example, of vertebral bodies, which usually have a thinner corticalis, fewer pixels are eliminated than pelvic bones, which usually have a thicker corticalis. As a result, also prevents the ^¬ that too many image data are eliminated or their radiation attenuation is decreased, and therefore areas of interest of the cancellous bone to be deleted.

Preferably, the processed 3D image data is stored so that it can be used for subsequent reconstructions and representations be available.

In addition to the processed image data even those image ^¬ data can be stored, the values of the radiation attenuation are preferably reduced to zero. The additional Spei ^¬ manuals such deleting matrix can be used for quality assurance, but also for other subsequent statements.

From the processed 3D image data may be a three-dimensional image of the body region or at least a two-dimensional sectional image, preferably a series of two-dimensional

Cut images are automatically reconstructed and displayed as a movie. Also, a certain number of relevant views can be created in the form of a gallery, with projections chosen that represent one or more similarly oriented bones in a single plane. This allows the treating radiologist a particularly rapid detection of bone metastases or other pathological changes within the cancellous bone of the considered body area.

When the reconstruction by means of a Maximum Intensity projek ^¬ tion (MIP) is made osteoplastic bone metastases can be displayed particularly well in particular.

On the other hand, reconstruction by means of a minimum intensity projection (MinlP) makes it possible in particular to better display osteolytic metastases.

Likewise, the reconstruction can also be carried out by means of a so-called average method. The reconstruction methods mentioned can also be combined, so that all Ver ^¬ changes can be detected within the cancellous bone, in particular both osteoly ^¬ tables and osteoblastic bone metastases, with entspre ^¬ assurance.

Likewise, the reconstruction can be carried out by means of a VRT (Volume Rendering Technique) method. Here, the processed before ^¬ pixels for individual density ranges are assigned to individual color values and transparency values to also represent Metasta ^¬ sen different species on a single image. In addition, the above-mentioned reconstruction methods can be combined with each other in a combination projection method, so that, for example, the maxima of a MIP and the minima of a MinIP are color-coded into an image. In this case, osteoplastic metastases can obtain a given color value range (for example red) as well as the osteolytic metastases a predetermined color value range (for example yellow), so that they are clearly recognizable on a single projection image or SD image. Important in the above-mentioned reconstruction methods is that those pixels whose radiation attenuation was preferably reduced to zero are not included in the calculation of the images. Especially with a reconstruction by means of a minimum intensity projection (MinIP), such pixels with a radiation attenuation equal to zero would render the representations unusable. For this reason, by the above process identi ^¬ fied pixels must be ignored in accordance with the reconstruction.

Advantageously, the raw image data prior to reconstruction or a noise suppression blur procedural ^¬ ren be subjected. Various known from Bildverarbei ^¬ processing noise reduction filters and blur can be a ^¬ set. In particular, the presentation of osteolytic metastases is more difficult since their radiation attenuation or

Density lies within the noise of cancellous bone. In this case, it is advantageous to have corresponding noise suppression. Apply soft-focus method and then make reconstructions using a MinlP or VRT to represent the osteolytic metastases can.

The object according to the invention is also achieved by an abovementioned device for carrying out an image processing method described above. Such a device is preferably formed by a corresponding computer stored with the CT image data. Due to the relative simplicity of the present image processing method, the method can also be carried out on conventional personal computers or notebooks in a very short time without the need for a workstation with correspondingly high computing power.

The present invention will be explained with reference to the accompanying Zeichnun ^¬ gene showing various steps of the procedural ^¬ Rens and application examples of the invention, in more detail.

Show:

Figure 1 is a CT image of the cross-section of the upper body of a patient and an explanatory diagram of the CT image.

FIG. 2 shows the CT image and a diagram according to FIG. 1 after the soft tissue removal; FIG.

3 shows the diagram of a method according to the invention for the detection and elimination of the corticalis;

4 shows the resulting two-dimensional image after removal of the soft tissues and the corticalis;

5 shows a variant of the method according to the invention in detail;

6 and 7 are flowcharts of an embodiment of the method according to the invention;

Figures 8 and 9 show a coronal and sagittal ΜΙΡ projection of conventional CT images of a patient; and

10 and 11, the coronal and sagittal ΙΡ Proj ection according to FIGS. 8 and 9 after application of the method according to the invention.

Fig. 1 shows a left CT image of the cross section of the upper body of a patient, wherein in addition to the body outline and the Weichtei ^¬ len also a section through a vertebral body as bone K is clearly visible. The right image shows a schematic view of the CT image for explanation. The dense corticalis C of the bone K or vertebral body can be seen in the CT image by particularly bright appearing outlines. Inside the vertebral body are bone metastases M, which are superimposed by the bright-appearing corticalis C of the bone K in a projection or a 3D visualization process.

FIG. 2 shows the CT image and the diagram according to FIG. 1, in which all pixels whose radiation _{attenuation falls} below a predetermined threshold value A _{s2 have been} eliminated or their radiation attenuation has been reduced or set to zero. This filtering method removes the soft tissues that are less dense and absorb less contrast. The method is also referred to as Highlightnin-g. When Highlightning the picture is scanned points. All pixels whose radiation attenuation is below a certain limit (eg 300 HU) can be deleted. This structures that lie deleted ^¬ gen below the typical values of the cancellous and cortical bone C,. Furthermore, all pixels can be deleted whose radiation attenuation is above this limit, in the immediate vicinity (eg 1 field in each direction, 8 Fel ^¬ the total) but not at least a certain number (eg: 1) is present at other pixels (neighbors) whose radiation attenuation is above the limit. This suppresses the noise (dots with randomly high radiation attenuation values). Nevertheless, the very bright outer layer of the bone K is still dominant part of the image.

As mentioned above, problems with soft tissue removal may occur, for example, when using intravenously administered contrast agents prior to CT acquisition. In this case it may be expedient to perform a pre-segmentation of the bones, for example by means of a region growing algorithm. This can be done, for example, such that within ^¬ within the bone K, a starting point is selected, from which all 3D image data for a generously chosen

Threshold for the radiation attenuation can be sampled. By means of this algorithm, the bone K can be generously recognized, ie together with the corticalis and any adhering soft tissue, and all soft tissue lying outside the bone can be eliminated. Thus, the disturbing image data can be elimi ^¬ ned. After removal, for example, very strong con- With trastised organs, the image data can be filtered with more sensitive thresholds for radiation attenuation, as only corticalis need be considered. As a result, less dense cortical areas remain over.

FIG. 3 shows the application of the method according to the invention, wherein several (here 4) starting points Si to S are selected within the image data. From each starting point Si, the image data are analyzed along lines Li and those image data are determined as the bone surface O or the beginning of the corticalis C, which exceed a predetermined threshold value A _S i of the radiation attenuation. The radiation attenuation of these determined as bone surface 0 image data and a corresponding number n of the following image data is preferably reduced to zero and then drive ^¬ fortge with the image data along a further line Li. According to the invention, therefore, a type of extinguishing jets is emitted from the starting points Si, which extinguish the soft tissues and the corticalis C.

Fig. 4 shows the result in accordance with processed image data, wherein the soft tissue and the cortical bone C of K were ent ^¬ removed. For example, the image data is ready for MIP projection and detection of metastases M is facilitated.

FIG. 5 shows the right-hand illustration according to FIG. 3 in detail.

Accordingly, the ausgewähl ^¬ th output points preferably randomized Si are set within the 3D or 2D image data and set by each base Si lines Li along which the image data is processed accordingly. The deletion of the image data is indicated in FIG. 5 as soon as the surface 0 of the bone K has been recognized (see arrows X). In this way, both the soft tissues and the cortical bone K can be removed, thus facilitating the diagnosis of the internal area of the bone K, the cancellous bone.

6 shows a flowchart of the method according to the invention. According to block 100, the computed tomographic investigation is carried out, resulting in a certain number of three-dimen ^{¬-dimensional} image data. According to block 101, the 3D image data is stored in a plurality of 2D image data, for example 100 to 200 2D image data, preferably with a respective layer thickness between 0.6 and 3 mm, divided. This 2D image data is starting with _. Step 102 is processed in succession according to the method described below. According to block 103, the removal of the soft parts, ie of all pixels with a radiation attenuation below a predefined threshold value A _{s2, takes place} . This so-called Highlightning eliminates disturbing soft tissue or artifacts outside the bony structures. According to block 104, a starting point Si within the 2D image data is preferably selected randomly. In step 105, it is checked whether this point within the 2D image data has already been deleted or its radiation attenuation has been reduced or set to zero, or is bone-tight, that is within a bone K. If appropriate, a selection of a new starting point Si according to block 104 is continued in accordance with step 106. In accordance with method step 107, a line Li from the selected starting point Si is selected one after the other and the surface 0 of a bone K is detected and all pixels before and corresponding to the number n are deleted after the surface 0 of the bone K. This loop is repeated according to the number of 2D images via step 108. After completing the process on the last set of 2D image data, finally, a suitable projection method, for example, a Maximum Intensity Projection (MIP) reconstruction method, is applied (block 109) and a corresponding image is displayed. According to the number of desired images, the process is repeated as often (block 110).

Before the projection of the image data, these can still be filtered and smoothed by ^various methods. Example ^¬ example, can be smoothed with a filter operating in the three-dimensional space method, which also includes image data of the surrounding layers, the edges of the resulting spongiosa. By including the surrounding layers, it can be ensured that no relevant image data is removed. In this way, image data that has been erased, but where image data are present in the layer above or below it, can also be replenished in layers and thus erroneously deleted cancellous bone components again reconstructed and gaps in the image are prevented.

FIG. 7 shows a further flowchart of the method described in FIG. 6 in detail. According to method step 200, the so-called highlighting takes place, ie all pixels with a radiation attenuation below a predetermined threshold value A _S 2 are deleted or their radiation attenuation is reduced to preferably zero. Starting from a specific starting point Si, according to step 201, the next pixel along the line Li is selected and proceeded along the line Li pixel by pixel (step 202). According to block 203, if the maximum length of the line L _it has reached the end of the 2D image data, the sub-process is ended and the same procedure is continued with a new line Li. In block 204, it is examined whether the radiation attenuation of the pixel has exceeded a predetermined threshold value A _sl . If this is not the case (see block 205), the process returns to step 200. If the radiation attenuation is above the threshold value A _s i, the system switches to the so-called corticalis erase mode and the pixel is deleted. Thereafter, the new point is proceeded according to block 207 and, according to 208, a check is made as to whether the preset number n of those pixels which are to be erased from detection of the surface 0 of the bone K has been reached. Thereafter, it is checked in step 209 whether the radiation attenuation of this pixel is also above the threshold A _S1 . If not, the method is terminated according to block 210. If, on the other hand, this pixel also lies above the threshold A _S1 , this pixel is also deleted according to block 211 and jumped to method step 212, where a new pixel along the line Li is selected and the method is continued.

In some cases, it may be advantageous to arrange the detected corticalis for protection against extinction. This protects all pixels in the vicinity of a bone and also inside the bone from being accidentally erased. Only after applying such a protection against extinction, the region's growing algorithm is used, which deletes all unprotected pixels and their surroundings from the periphery. This results in image data of the bone with a correspondingly broad (in For example 10-20 pixels) so-called soft tissue cuff. Optimal results can be achieved if starting points are set to the limit of this soft tissue cuff next to each other at a distance of, for example, 10 to 20 pixels. By such a targeted setting of the starting points for the image processing method, the length of the extinguishing rays emanating from the starting points can be reduced and with the same number of extinguishing jets a significantly improved speed and uniformity of the extinction can be achieved.

Figures 8 and 9 now show a coronal and sagittal MIP projection of CT images of the same patient without the use of the method of the invention. The bone metastases M are not clearly recognizable.

By contrast, FIGS. 10 and 11 show the same image data as FIGS. 8 and 9 after application of the method according to the invention, with metastases M being much more clearly recognizable in pelvic bones and vertebral bodies.

Due to the fact that the environment of the bone, the corticalis and also preferably the soft tissues have been removed from the image data, any osteoplastic bone metastases in a maximum intensity (MIP) projection and osteolytic metastases in a minimum intensity projection (MinlP) reconstruction come out clearly and can be detected more easily and quickly by the diagnosing radiologist.

Claims

Claims:

A method for processing computed tomography (CT) generated three-dimensional image data of a bone (K) containing body region of a patient, the SD image data containing the respective location-dependent radiation attenuation of the body region, characterized in that those image data as the surface (0) of Bones (K) are determined within the body region, which exceed a predetermined threshold value (Asi) of the radiation attenuation, and that the radiation attenuation of this as bone surface (0) determined image data and a number (n) of adjacent image data is preferably reduced to zero.

2. The image processing method according to claim 1, characterized in that the 3D image data is divided into a plurality of 2D image data, preferably with a layer thickness between 0.6 and 3 mm, and the 2D image data is processed successively.

3. The image processing method according to claim 1 or 2, characterized in that those image data are determined, which fall below a predetermined threshold value (A _S2 ) of the radiation attenuation, and that the radiation attenuation of this determined image data is reduced.

4. Image processing method according to one of claims 1 to 3, characterized in that those image data are determined as the surface (0) of the bones (K), in which the difference between the value of the radiation attenuation and the value of the radiation attenuation of the preceding image data has a predetermined threshold value ( ÄA _S ).

5. An image processing method according to any one of claims 1 to 3, characterized in that at least a starting point (R ^) is selected in ^¬ nerhalb the 3D image data, or any 2D image data and from any starting point (Ri) from all 3D image data of the volume or all 2D image data of each layer to each output point (Ri) are determined as the surface of the bone (0) which has a pre-given threshold ^¬ (a _sl) of the radiation attenuation exceed and that the radiation attenuation of this image data determined as bone surface (0) is reduced.

6. An image processing method according to claim 5, characterized in that the determined values of the radiation attenuation of all the image data ^¬ outside the bone surface as a (0) ^¬ image data is reduced.

7. The image processing method according to claim 1, wherein at least one starting point (Si) is selected within the 3D image data or each 2D image data, and from every starting point (Si) all 3D image data of the volume or all 2D image data are selected. Image data of the layer in succession along a line (L _x ) are analyzed and those image data along this line (Li) are determined as bone surface (0), which exceed a predetermined threshold (A _S1 ) of the radiation attenuation, and that the radiation attenuation of this as a bone surface ( 0) and, corresponding to the number (n) of subsequent image data, is preferably reduced to zero and thereafter the image data of the volume or of the layer is analyzed along another line (Li).

8. An image processing method according to claim 7, characterized characterized labeled in ^¬ that from each starting point (Si) from all the image data in the volume or .Schicht sequence along a multi ^¬ number of lines (Li), preferably at a regular angular ^¬ distance of 1 ° up to 10 ° to each other.

9. image processing method according to any one of claims 5 to 8, characterized in that the position of each starting point (Ri, Si) is determined randomly.

10. An image processing method according to any one of claims 1 to 9, characterized in that the number (n) of the determined at as bone ^¬ chenoberfläche (0) image data adjacent image data whose radiation attenuation is reduced to preferably zero, is constant, and preferably 3 to 5 is.

11. The image processing method according to any one of claims 1 to 9, characterized in that the number (n) of the as Kno chenoberfläche (0) determined image data adjacent image data whose radiation attenuation is reduced to preferably zero, is set.

12. An image processing method according to claim 11, characterized in that the number (n) of the image surface of the bone (0) determined image data adjacent image data whose radiation attenuation is reduced to preferably zero, depending on the spatial position of the image data, preferably automatically, is set ,

13. An image processing method according to any one of claims 1 to 12, characterized in that the processed 3D image data are stored.

14. An image processing method according to claim 13, characterized in that in addition to the processed 3D image data and those image data are stored, the values of the radiation attenuation were reduced.

15. The image processing method according to claim 1, wherein from the processed SD image data, a three-dimensional image or at least a two-dimensional slice image of the body region, preferably a plurality of two-dimensional slice images is automatically reconstructed and displayed as a film.

16. The image processing method according to claim 15, characterized in that the reconstruction is carried out by means of a maximum intensity projection (MIP).

17. An image processing method according to claim 15 or 16, characterized in that the reconstruction is carried out by means of a minimum intensity projection (MinlP).

18. An image processing method according to any one of claims 15 to 17, characterized in that the reconstruction is performed by means of an average method.

19. Image processing method according to one of claims 15 to 18, characterized in that the reconstruction is carried out by means of a VRT method.

20. Image processing method according to one of claims 15 to

19, characterized in that the image data is subjected to noise suppression or soft-focus procedure before the recon ^¬ constructive tion.

21. An apparatus for performing an image processing method according to any one of claims 1 to 20.