JP5846368B2 - Medical image processing apparatus, method, and program - Google Patents

Description

  The present invention relates to a medical image processing apparatus, method, and program, and more particularly, to a medical image processing apparatus, method, and program for extracting a target region from a non-contrast image taken without using a contrast agent.

  In recent years, medical image diagnostic techniques using medical images represented by X-ray CT (Computed Tomography) images, MRI (Magnetic Resonance Imaging) images and the like have been introduced in the medical field.

  At present, in order to extract a target region from these medical images, a contrast agent having high reactivity at the time of radiographing is injected into the target site in advance according to an imaging device such as MRI or CT, and the target site is clearly displayed on the image. It is done to make it appear. In addition, image processing for accurately extracting a target region from such a contrast image has been proposed.

  For example, a plurality of pixel values located in the vicinity of the projection line to each pixel of the projected image are extracted from the pixel value of the original image, and projected to each pixel position of the projected image by interpolation calculation using a spline function from each of the extracted pixel values. There has been proposed an image projection apparatus that interpolates pixel values to be performed and selects a maximum value among values obtained as pixel values to be projected at the same pixel position (see Patent Document 1).

  Further, a curved surface including a designated point group designated by the user is generated, a thickness T is set for a drawing target region obtained by performing hidden surface processing on the curved surface, and a region of interest is determined. A region-of-interest designating method for performing MIP processing has been proposed (see Patent Document 2).

  Also, a functional image representing the function at each position of the heart is generated using the first voxel data constituting the three-dimensional medical image obtained by photographing the subject, and the three-dimensional medical image obtained by photographing the subject. Using the voxel data of the region including the blood vessel along the outer wall of the heart of the heart among the second voxel data constituting the morphological image, a morphological image in which the shape of the blood vessel is drawn is generated, and each position of the heart in the functional image and the morphological image There has been proposed a cardiac function display device that displays a function image and a morphological image in an overlapping manner in correspondence with each position of the heart (see Patent Document 3).

  In addition, a cardiac cavity region is extracted from a three-dimensional functional image representing cardiac function evaluation index data, a cardiac functional bullseye image representing cardiac chamber function is generated, and a coronary artery is derived from a three-dimensional morphological image representing the structure of the heart and coronary artery. Extracting image data, calculating a coronary artery closed curved surface including the extracted coronary artery image data, generating a coronary artery bullseye image from the extracted coronary artery image data, and based on the coronary artery closed curved surface, boundary of the heart chamber region A diagnosis support apparatus that calculates the distance from the surface to the coronary artery closed curved surface has been proposed (see Patent Document 4).

  Also, for the volume data of the area including the thickness of the object to be observed, a drawing method such as the maximum value method or the average value method is determined, an arbitrary surface to be a display surface is created, and an image region corresponding to the arbitrary surface Is obtained, a position on an arbitrary plane (for example, xy coordinates) corresponding to each pixel on the image area is obtained, and a thickness d at the position (x, y) on the arbitrary plane 11 is calculated. It has been proposed (see Patent Document 5).

  When projecting 3D image data to obtain an MIP image or the like, a B-spline is used in the z-axis direction where the density of existing image data is the lowest among the 3 axes x, y, and z in the 3D space. An image processing apparatus for interpolation has been proposed (see Patent Document 6).

Japanese Patent Laid-Open No. 05-207990 Japanese Patent Laid-Open No. 2006-187531 JP 2008-253753 A JP 2010-246776 JP 2006-263078 A JP 2004-173847 A

  However, the techniques of Patent Documents 1 to 6 are applied to a contrast image captured using a contrast agent, and a non-contrast image captured without using a contrast agent is applied because the contrast of the target portion is low. There is a problem that it is difficult.

  Moreover, when using a contrast agent for imaging | photography of a medical image, there exists a problem that it may have a bad influence on the patient who is a to-be-photographed person.

  The present invention has been made in view of the above-described problems, and provides a medical image processing apparatus, method, and program capable of accurately extracting a target region from a non-contrast tomographic image and displaying it three-dimensionally. With the goal.

  In order to achieve the above object, the medical image processing apparatus of the present invention includes a generating unit that generates a three-dimensional image by stacking a plurality of non-contrast tomographic images obtained by imaging a body including a target region without using a contrast agent. Extracting a region having a pixel value equal to or larger than a predetermined value from the three-dimensional image generated by the generating unit as a bone region, and corresponding the predetermined number of bone regions in the plurality of non-contrast tomographic images in the stacking direction. Setting means for setting a processing region including the target portion for each of the predetermined number of images based on a bone region projection image constituted by a pixel value that is maximum on a projection line when projected; and the setting in the three-dimensional image Based on the distribution of pixel values in the processing region set by the means and the relationship between the target portion and the distribution of pixel values obtained in advance, a predetermined range of pixel values indicating the target portion is set, and the predetermined range Included picture A correction unit that shifts the value and corrects the width of the predetermined range to expand; and the three-dimensional image in which the pixel value is corrected by the correction unit has a corrected pixel value indicating the target portion A target part extracting unit that extracts a region where pixels are connected and has the maximum number of pixels in the region, and a region extracted by the target part extracting unit is displayed so as to be displayed as a three-dimensional image. Display control means for controlling the means.

  According to the medical image processing apparatus of the present invention, the generation unit generates a three-dimensional image by stacking a plurality of non-contrast tomographic images obtained by imaging the body including the target region without using a contrast agent. Then, the setting unit extracts a region having a pixel value equal to or larger than a predetermined value from the three-dimensional image generated by the generation unit as a bone region, and stacks bone regions corresponding to a predetermined number of consecutive images in a plurality of non-contrast tomographic images. A processing region including a target region is set for each predetermined number based on a bone region projection image formed by pixel values that are maximum on the projection line when projected in the direction. In a non-contrast tomographic image, a bone region has a higher pixel value than a region showing an organ or muscle and can be clearly identified. A processing region is set using such a bone region. In addition, since the processing area is set for every predetermined number of consecutive non-contrast tomographic images, it is possible to set the processing area along the human body structure.

  Then, the correction unit determines a predetermined pixel value indicating the target part based on the distribution of the pixel values in the region set by the setting unit in the three-dimensional image and the relationship between the target part and the pixel value distribution obtained in advance. A range is set, pixel values included in the predetermined range are shifted, and correction is performed so that the width of the predetermined range is expanded. Based on the relationship between the target part and the distribution of pixel values obtained in advance, that is, foresight information, it is possible to determine which range of the pixel value distribution obtained from the three-dimensional image is the pixel value indicating the target part. The pixel values in the range are shifted and the range is expanded to differentiate from the surrounding pixels.

  Then, in the three-dimensional image in which the pixel value is corrected by the correction unit, the target part extraction unit is a region in which pixels having corrected pixel values indicating the target part are connected, and the number of pixels in the region is the maximum To extract the region. The area indicating the target part has a certain size, and when the area is small, that is, the area with a small number of pixels is an area corresponding to an error or noise extracted due to the influence of the shadow, etc. It can be determined that the region is not shown. Then, the display control means controls the display means so that the region extracted by the target part extraction means is displayed as a three-dimensional image.

  In this way, the processing region is set based on the bone region, the pixel value indicating the target region in the processing region is corrected so as to shift and expand the range, and the pixel having the corrected pixel value indicating the target region is obtained. Since the region having the maximum number of pixels in the connected region is extracted as a region indicating the target part, the target part can be extracted from the non-contrast tomographic image with high accuracy and displayed three-dimensionally.

  Further, the setting means can interpolate, with a spline curve, a point where the pixels indicating the bone region are discrete in the bone region projection image. In a bone region projection image corresponding to a predetermined number of consecutive images in a plurality of non-contrast tomographic images, pixels indicating the bone region may be discrete when the predetermined number is small, but by interpolating with a spline curve, The processing area can be set with high accuracy.

  In addition, the correction unit may include a distribution of pixel values of a three-dimensional image based on a plurality of contrast tomographic images obtained by imaging a body including a target region using a contrast agent, and a three-dimensional image based on the plurality of non-contrast tomographic images. A predetermined range of pixel values indicating the target portion can be set based on the relationship obtained in advance by comparison with the distribution of pixel values. By using such look-ahead information, it is possible to set a predetermined range of pixel values indicating the target portion with high accuracy.

  The medical image processing apparatus of the present invention extracts a region expanded by the region expansion method from the three-dimensional image, and calculates pixel values of pixels included in the expanded region as the expanded region and the background. Is converted into a distance value indicating the shortest distance to the area, and based on the distance value, the expanded area is separated at a constricted portion of the area, and based on at least one of the size and shape of each separated area, The other part extraction means for extracting a region indicating another part different from the target part may be included, and the display control means includes the area extracted by the target part extraction means and the other part extraction means. The display means can be controlled so that the difference area from the area extracted by the above is displayed as a three-dimensional image. Thereby, even when the target part and the other part are extracted as a close region, the target part can be extracted with high accuracy by removing the other part.

Further, in the medical image processing method of the present invention, a computer generates a three-dimensional image by stacking a plurality of non-contrast tomographic images obtained by imaging a body including a target region without using a contrast agent. When a region having a pixel value greater than or equal to a predetermined value is extracted from the image as a bone region, the bone region corresponding to a predetermined number of consecutive images in the plurality of non-contrast tomographic images is maximized on the projection line. Based on a bone region projection image composed of pixel values, a processing region including the target part is set for each of the predetermined number, and a distribution of pixel values in the processing region set in the three-dimensional image is obtained in advance. Based on the relationship between the target portion and the distribution of pixel values, a predetermined range of pixel values indicating the target portion is set, the pixel values included in the predetermined range are shifted, and the width of the predetermined range is expanded. So that In the three-dimensional image in which the prime value is corrected, a region where pixels having corrected pixel values indicating the target portion are connected and the number of pixels in the region is maximized is extracted and extracted. This is a method for controlling the display means so that the region is displayed as a three-dimensional image.

  The medical image processing program of the present invention is a program for causing a computer to function as each means constituting the medical image processing apparatus.

  As described above, according to the medical image processing apparatus, method, and program of the present invention, a processing region is set based on a bone region, a pixel value indicating a target part in the processing region is shifted, and the range is expanded. In order to extract the region having the maximum number of pixels as the region indicating the target region in the region where the pixels having corrected pixel values indicating the target region are connected, the target region is accurately extracted from the non-contrast tomographic image. Thus, an effect that three-dimensional display can be performed is obtained.

It is a block diagram showing a schematic structure of a medical image processing device concerning a 1st embodiment. It is a figure for demonstrating the process of a pixel value correction | amendment part. It is a figure for demonstrating a labeling process. It is a figure which shows an example of the number of the neighborhood which is a determination criterion of connectivity. It is a flowchart which shows the content of the medical image processing routine in 1st Embodiment. It is a figure for demonstrating the production | generation of a three-dimensional image. It is a flowchart which shows the content of the process area setting process routine. It is a figure which shows the three-dimensional image of a bone area | region. It is a figure which shows a bone area | region projection image. It is a figure for demonstrating the spline interpolation of a bone area | region projection image. It is a figure which shows the setting of a processing area. It is a flowchart which shows the content of the pixel value correction process routine. It is a figure which shows the histogram of the pixel value for an analysis. It is a figure for demonstrating calculation of the inflection point in the histogram of a pixel value. It is a figure for demonstrating the setting of the processing width WIDE which determines the pixel value of correction object. It is a figure which shows an example of the process result of a pixel value correction process. It is a flowchart which shows the content of the labeling process routine. It is a figure for demonstrating the setting of the pixel value used as the object of a labeling process. It is a figure which shows the pixel number histogram according to label value. It is a figure for demonstrating extraction of the area | region which shows the object site | part from the area | region after a labeling process. It is a figure which shows an example of the process result of the medical image processing apparatus which concerns on 1st Embodiment. It is a figure which shows an example of the process result of the medical image processing apparatus which concerns on 1st Embodiment. It is a figure which shows an example of the process result of the medical image processing apparatus which concerns on 1st Embodiment. It is a block diagram which shows schematic structure of the medical image processing apparatus which concerns on 2nd Embodiment. It is a flowchart which shows the content of the medical image processing routine in 2nd Embodiment. It is a flowchart which shows the content of the other site | part removal process routine. It is a figure for demonstrating the area | region expansion method. It is a figure for demonstrating the area | region expansion method. It is a figure for demonstrating the division | segmentation of the area | region by a constriction location. It is a figure which shows an example of the process result of the medical image processing apparatus which concerns on 2nd Embodiment.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, a case where the present invention is applied to a medical image processing apparatus for extracting and displaying three-dimensionally the blood vessels of the chest and abdomen as target portions will be described as an example.

  As shown in FIG. 1, a medical image processing apparatus 10 according to the first embodiment displays a computer 12 that performs image processing for three-dimensionally displaying a target region extracted from a medical image, and displays an image processing result. And a display device 14 for doing so.

  The computer 12 includes a CPU, a ROM, a RAM, and an HDD, and a program for a medical image processing routine described later is stored in the HDD. The computer 12 is functionally represented by a configuration including a three-dimensional image generation unit 20, a processing region setting unit 22, a pixel value correction unit 24, a labeling processing unit 26, and a display control unit 28. Can do.

  The three-dimensional image generation unit 20 uses a plurality of tomographic images of a part of the human body (here, the chest and abdomen) including the target part without using a contrast agent in a medical image photographing apparatus such as CT or MRI. A three-dimensional image is generated by stacking non-contrast tomographic images in a direction orthogonal to the tomographic direction.

  The processing region setting unit 22 extracts a region having a pixel value equal to or greater than a predetermined value from the three-dimensional image generated by the three-dimensional image generation unit 20 as a bone region. Also, based on a bone region projection image obtained by applying MIP (Maximum Intensity Projection) to a three-dimensional image of a bone region corresponding to a predetermined number of consecutive non-contrast tomographic images stacked when generating a three-dimensional image. Thus, a processing area is set for every predetermined number of non-contrast tomographic images. The processing area is an area that is a target of pixel value correction processing and labeling processing described later. The processing region setting here is intended to separate regions such as chest and abdominal muscles and fat from the organ region. Since internal organs and muscles are composed of very close pixel values in the image, separation based on pixel value information is very difficult. Therefore, here, the pixels existing outside the annular structure of the spine, ribs, and sternum constitute fat and muscle, and are based on the clinical characteristic that blood vessels that are target sites in this embodiment are not included. Thus, the processing area is set. Note that the processing for setting the maximum value on the projection line from an arbitrary direction as a pixel value for a three-dimensional image is referred to as MIP. Conventionally, it is mainly used for visualization of contrast blood vessels, and pixels having no maximum value are not extracted. .

  The pixel value correction unit 24 creates a histogram of pixel values of pixels in the processing region set by the processing region setting unit 22 in the three-dimensional image generated by the three-dimensional image generation unit 20. In addition, based on the relationship between the target portion and the pixel value histogram obtained in advance, a predetermined range of pixel values indicating the target portion is set in the created histogram. Correction is performed by applying a transmission function to pixel values included in the predetermined range. Here, the transmission function used in the pixel value correction unit 24 will be described. Here, in order to differentiate between a pixel indicating a blood vessel that is a target site to be extracted and its surrounding pixels, the pixel value to which the transmission function is applied is corrected. This means that the pixel value is corrected by multiplying the histogram of the pixel value as shown in FIG. 2 by a functional expression that deforms the histogram around the pixel value of the pixel indicating the blood vessel. The transmission function is a function defined so that the pixel value of the pixel indicating the target region is shifted to the high luminance side, and the corrected pixel value range is expanded from the predetermined range set before correction, By applying this function, the pixel indicating the target region and the peripheral pixels are differentiated.

  The concept of a transmission function for correcting such pixel values will be described. A pixel value at each coordinate (i, j, k) of the three-dimensional image IN is denoted as IN (i, j, k). Here, considering the transmission function for the pixel value IN (i, j, k), when the processing width is WIDE and the center of the processing width is CENTER, the processing width start point α and the processing width end point β are the following (1) The corrected three-dimensional image OUT (i, j, k) is expressed by the following expression (2).

  This processing depends on the size of the processing width WIDE and the setting of the processing width center CENTER. The setting of WIDE and CENTER is determined based on the foresight information indicating in which range the pixel value indicating the target part appears in the pixel value histogram. This foresight information is, for example, a histogram of pixel values obtained from a contrast image using a contrast image obtained by imaging the same target region using a contrast agent and a non-contrast image obtained without using a contrast agent. And the difference between the histogram of pixel values obtained from the non-contrast image.

  In the present embodiment, a range (WIDE) for correcting the pixel value as described above is automatically set in a pixel value correction process described later.

  The labeling processing unit 26 assigns a label to each region where pixels having corrected pixel values indicating the target portion are connected in the three-dimensional image whose pixel values are corrected by the pixel value correcting unit 24 (labeling processing). A region having the maximum number of pixels in the region for each label is extracted as a region indicating the target part. The labeling process is a process for giving a group attribute to independent structures in an image. When there are multiple structures in a single image, each labeling process identifies It is processing. FIG. 3 shows an image of labeling processing and group attributes for each region. In the upper diagram of FIG. 3, shaded pixels are pixels having corrected pixel values indicating the target portion. Then, as shown in the lower diagram of FIG. 3, a different label is assigned to each region where the shaded pixels are connected. In FIG. 3, for ease of explanation, the labeling process is shown for a two-dimensional image. However, in the present embodiment, the labeling process is performed on a three-dimensional image. In the labeling process for a three-dimensional image, for example, a connectivity determination criterion as shown in FIG. 4 is used. In the example of FIG. 4, 6 neighborhood processing that considers only vertical and horizontal directions, 18 neighborhood processing that emphasizes connectivity of planes for each three-dimensional axis, and 26 neighborhoods that emphasize connectivity in all directions including diagonal directions. Processing is shown. In particular, in the 18 neighborhood processing, the same value label is added to pixels that are continuous in the diagonal direction, whereas in 6 neighborhood processing, different labels are added to the continuous pixels in the diagonal direction. The number of neighbors to perform is important in the labeling process. Since the amount of calculation increases with an increase in the number of neighbors for connection determination, the type of connection determination to be used may be selected according to the shape of the target part to which a label is to be added. In this embodiment, since there is little difference in the results obtained between the 18-neighbor process and the 26-neighbor process, the 18-neighbor process is used from the viewpoint of reducing the amount of computation and shortening the computation time.

  The display control unit 28 performs control so that the region indicating the extracted target part is displayed three-dimensionally on the display device 14.

  Next, the operation of the medical image processing apparatus 10 according to the first embodiment will be described. In a medical imaging apparatus such as CT or MRI, a part of the human body (here, the chest and abdomen) including the target part is tomographically imaged at predetermined intervals without using a contrast agent, and a plurality of non-contrast tomographic images (two-dimensional) When the image is obtained, the medical image processing apparatus 10 executes a medical image processing routine shown in FIG.

  In step 100, a plurality of non-contrast tomographic images taken are acquired.

  Next, in step 110, the plurality of non-contrast tomographic images acquired in step 100 are stacked in a direction orthogonal to the tomographic direction as shown in FIG. 6A, and as shown in FIG. A three-dimensional image is generated.

  Next, in step 120, processing area setting processing is executed. Here, the processing region setting processing routine will be described with reference to FIG.

  In step 122, in the three-dimensional image generated in step 110, a pixel having a pixel value equal to or greater than a predetermined threshold is extracted as a bone region. When CT values of non-contrast tomographic images are distributed between −2000 (HU) and +2000 (HU), almost all regions having CT values of 110 (HU) or more can be regarded as bone regions. it can. Therefore, using this value as a threshold, a three-dimensional image in which only the bone region as shown in FIG. 8 is extracted is created.

  Next, in step 124, the MIP from the image stacking direction is applied to the three-dimensional image of the bone region created in step 122. Here, considering that the tomographic areas from the abdomen to the chest of the human body are sequentially different, the MIP is calculated for each part of the bone region corresponding to a predetermined number (for example, five) of non-contrast tomographic images continuous in the stacking direction. Apply. Thereby, it is possible to set a processing region as close to the body structure as possible. When MIP is applied to a three-dimensional image of a bone region corresponding to a predetermined number of non-contrast tomographic images, a two-dimensional image (bone region projection image) as shown in FIG. 9 is created. In order to apply MIP to such a small number of images, there are locations where the bone regions are discrete on the bone region projection image, as shown in FIG.

  Therefore, in step 126, discrete points are interpolated with spline curves as shown in FIG. 10 for the bone region projection image created in step 124. In the figure, black circles are known points, and shaded circles are points interpolated with a spline curve based on the known points.

  Next, in step 128, as shown in FIG. 11, the inside of the annular structure connecting the bone regions of the bone region projection image after interpolation is set as a processing region for pixel value correction processing and labeling processing described later, The process returns to the medical image processing routine of FIG.

  Next, in step 140, pixel value correction processing is executed. Here, the pixel value correction processing routine will be described with reference to FIG.

  In step 142, a histogram of pixel values of the pixels in the processing region set in step 128 is created in the three-dimensional image generated in step 110. When the target region is a blood vessel as in the present embodiment, as shown in FIG. 13, the pixel value around the rightmost peak (shaded portion in the figure) of the pixel value histogram is the pixel value indicating the blood vessel. It is known to correspond to the pixel value. This is obtained, for example, by obtaining a pixel value histogram based on a contrast image and a pixel value histogram based on a non-contrast image by imaging a plurality of patients and comparing them. Therefore, in the following steps, this range is automatically extracted and a transmission function is set.

  Next, in step 144, preprocessing is performed on the histogram of pixel values created in step 142. The preprocessing here is to delete pixel values in an unnecessary range and to apply a low-pass filter. First, since it is known that the pixels around the pixel value 0 [HU] are regions that reflect devices such as air and a bed, information of 500 [HU] or less is deleted, for example. The range to be deleted may be appropriately set according to the target part. The low-pass filter is for removing a portion (high frequency component) that repeats minute fluctuations in the histogram and facilitating calculation of an inflection point, which will be described later. For example, a low-pass filter with a normalized angular frequency ω = 0.05 [rad / s] can be applied.

  Next, in step 146, all inflection points (marked with ● in the figure) are calculated from the histogram pre-processed in step 144 as shown in FIG. Then, the rightmost peak is searched from these inflection points, and the inflection point indicating the rightmost peak is set as CENTER. Further, as shown in FIG. 15, the first inflection point in the minus direction from CENTER is point P1, and the first inflection point in the plus direction is P2, and a predetermined width between these two points (for example, the width between the two points). 1/2) is set as WIDE. The predetermined width may be the entire width between two points, or may be 1/3 between two points, 3/4, or the like, and is set as appropriate according to the target site.

  Next, in step 148, based on the setting in step 146, a transmission function shown in the following equation (3) is set.

  The transmission function shown in the equation (3) has a characteristic of shifting the pixel value around CENTER to a higher luminance side and expanding the corrected range of the pixel value within the WIDE range. To what extent the luminance is increased is determined by the value of RATE. The higher the RATE value, the easier it is to differentiate between the pixel to be corrected and its surrounding pixels. On the other hand, since a pixel value exceeding the 16-bit dynamic range may be taken with a value of RATE = 10 or more, an appropriate value is determined in advance according to the target region. Here, for example, RATE = 1.5 can be set.

  Next, in step 150, for the pixel having the pixel value within the WIDE range set in step 146, the pixel value is corrected using the transmission function set in step 148, and the medical image processing routine of FIG. Return to

  FIG. 16 shows an example of the processing result at the stage where the transmission function is applied. (A) is a simple non-contrast CT image without application of the transmission function of this embodiment, (b) is a result of applying the transmission function of this embodiment to a non-contrast image, and (c) is substantially equivalent. It is the contrast image which image | photographed the site | part. When (a) and (b) are compared, the contrast of the part that was unclear in (a) is clear in (b) because the pixel values of blood vessels such as the aorta and vena cava are corrected. It can be seen that In addition, the pixel values of the muscles and fat portions of the outer edge of the abdomen are outside the processing region, so the pixel values are not corrected, and the pixel values of the target region are corrected while differentiating from other regions. Has been. Comparing (b) and (c) in the figure, it can be confirmed that a region equivalent to a portion whose contrast is clarified by contrast is a correction target.

  Next, in step 160, a labeling process is executed. Here, the labeling processing routine will be described with reference to FIG.

  In step 162, a pixel value to be labeled is set. Here, as shown in FIG. 18, in the histogram created using the pixel values after the correction in step 150, the newly generated inflection point P3 is CENTER and has the same width as WIDE set in step 146. WIDE is set, and pixels having pixel values within this WIDE range are set as labeling targets. Note that a new WIDE may be set from the pixel value after correction of the WIDE start point pixel value before correction to the pixel value after correction of the WIDE end point pixel value before correction.

  Next, in step 164, a label having the same value is added to the connected pixels having the pixel value set in step 162, which are pixels in the processing region of the three-dimensional image. The connectivity determination condition may be 18 neighborhood processing, for example.

  Next, in step 166, as shown in FIG. 19, a label value-specific pixel number histogram representing the number of pixels to which the label of the same value is assigned is created, and the label value that maximizes the number of pixels is selected. The fact that the number of pixels with the same value label is large means that there is a high probability that the area represents one structure, and when the number of pixels with the same value label is small, the influence of shadows, etc. It is likely that the area does not represent a structure. In particular, since the blood vessel that is the target portion of the present embodiment has a structure that extends vertically, the number of pixels increases when viewed three-dimensionally. Therefore, a region is extracted based on the number of pixels.

  Next, in step 168, the region to which the label value selected in step 166 is added is extracted, and the process returns to the medical image processing routine of FIG.

  For example, when a label is given as shown in FIG. 20A and the number of pixels of label 1 is the maximum, a region as shown in FIG. 20B is extracted. In FIG. 20, for simplicity of explanation, a two-dimensional image is displayed, but actual processing is performed on the three-dimensional image. For this reason, in FIG. 5A, it appears that there are a plurality of regions of the label 1, but when viewed three-dimensionally, these regions are connected at any point.

  Next, in step 180, the region extracted in step 168 is displayed three-dimensionally on the display device 14, and the process ends.

  The processing results by the medical image processing layer of the first embodiment are shown in FIGS. In each figure, (a) shows the processing result according to the present embodiment, and (b) shows the extraction result from the contrast image for reference. (A) was performed automatically from image input to output, and the contrast image (b) was extracted as a reference material with manual lettering. From this result, it was confirmed that the heart, the surrounding arch aorta and the abdominal aorta were extracted in the same manner as the contrast image, and the effectiveness of the present embodiment was confirmed.

  As described above, according to the medical image processing apparatus of the first embodiment, the processing region is set based on the bone region, and the pixel value indicating the target part in the processing region is shifted to the high luminance side. In addition, by using a transmission function that expands the range and performing a labeling process using the pixel values in the corrected processing region, the region indicating the target region is extracted, so accuracy can be obtained from the non-contrast tomographic image. The target part can be well extracted and displayed three-dimensionally.

  Next, a second embodiment will be described. In addition, about the medical image processing apparatus of 2nd Embodiment, about the structure same as the medical image processing apparatus 10 of 1st Embodiment, the same code | symbol is attached | subjected and detailed description is abbreviate | omitted.

  As illustrated in FIG. 24, the medical image processing apparatus 210 according to the second embodiment includes a computer 212 and a display device 14. Functionally, the computer 212 includes a three-dimensional image generation unit 20, a processing region setting unit 22, a pixel value correction unit 24, a labeling processing unit 26, another part extraction unit 30, and a display control unit 228. It can be expressed in a configuration that includes it.

  As shown in FIGS. 21 to 23, in the medical image processing in the first embodiment, although the target part can be extracted with high accuracy, other parts different from the target part are also extracted. The second embodiment aims to remove such other parts.

  The other part extraction unit 30 roughly extracts a region indicating the other part to be removed from the three-dimensional image generated by the three-dimensional image generation unit 20 by a region expansion method, and the distance value of each pixel of the extracted region By separating the regions based on the above, regions indicating other parts are extracted.

  The display control unit 228 indicates another region from the region extracted by the labeling processing unit 26 by taking a difference region between the region extracted by the labeling processing unit 26 and the region extracted by the other region extracting unit 30. The region is removed, a region indicating the target part is extracted, and control is performed so that the region is three-dimensionally displayed on the display device 14.

  Next, the operation of the medical image processing apparatus 210 according to the second embodiment will be described. In the second embodiment, as shown in FIG. 25, the other part removal processing step (step 200) is only added to the medical image processing routine in the first embodiment, so refer to FIG. The other part removal processing routine will be described.

  In step 202, a region extension method is applied to the three-dimensional image generated in step 110 to extract a rough region region. Specifically, as shown in FIG. 27A, a seed point is set on a three-dimensional image, and as shown in FIG. X64 pixels) is set, and the average value μ and standard deviation σ of the pixel values in the window are calculated. Then, as shown in FIG. 28A, it is determined whether or not the condition of the following expression (4) is satisfied for pixels adjacent in the predetermined extension direction from the seed point (for example, pixels in the vicinity of 6). If the condition is satisfied, the area is expanded.

  Next, as shown in FIG. 4B, the expanded pixel is newly set as a search pixel, a window centered on the search pixel is set, and the process is repeated, so that there is no pixel that satisfies the condition. The expanded pixel group at the stage is extracted as an expanded region ((c) in the figure). 27 and 28 are shown as two-dimensional images for ease of explanation.

  Next, in step 204, the pixel value of each pixel included in the extended region indicating the rough other part extracted in step 202 is set to the shortest distance between the pixel and the background (region other than the extracted extended region). Is converted into a distance value indicating.

  Next, in step 206, based on the distance value converted in step 204, the region is divided at the constricted portion of the region as shown in FIG. When different parts are extracted as one extended region due to close contact with each other, a constriction occurs in the close contact portion, that is, there is a portion where the path is narrowed in the region. Different parts can be divided by dividing by the constricted part. A label is assigned to each divided area.

  Next, in step 208, for each region to which a label is assigned, it is determined whether or not the region is a region indicating another part to be removed based on the size and shape of the region. An area determined not to be another part is deleted, and an area indicating the other part is extracted.

  Next, in step 209, the region extracted in step 208 is removed from the region extracted in step 168, that is, a difference image between the image after labeling and the image extracted from other parts is generated. Then, the process returns to the medical image processing routine of FIG.

  The processing result by the medical image processing apparatus 210 of the second embodiment is shown in FIG. (A) is an extraction result by the labeling processing unit 26, (b) is a processing result after the removal of other parts in the second embodiment, and (c) is an extraction result from a contrast image as a reference example. It is.

  As described above, according to the second embodiment, even when the target part is extracted together with the other part that is in close contact with each other, the constriction between the other part and the target part is generated. Since the part can be extracted and removed, the target part can be extracted from the non-contrast tomographic image with higher accuracy.

  In the first and second embodiments, the case where blood vessels in the chest and abdomen are extracted as target portions has been described. However, other organs, tumors, and the like may be used as target portions. Moreover, although the said embodiment demonstrated the case where a process area | region was set based on the cyclic | annular structure obtained by a bone area | region, it is not limited to this. If the rough positional relationship between the bone region and the target region is known, the processing region can be set based on this. For example, for arms and legs, a predetermined range centered on the bone region can be set as the processing region.

  In the above embodiment, the case where the pixels within the WIDE range are shifted to the high luminance side by the transmission function has been described. However, since it is only necessary to be able to differentiate from the pixels other than the target region, the low luminance side You may shift to.

  Moreover, although the said embodiment demonstrated the case where the pixel value of the predetermined range in the histogram of a pixel value was demonstrated, it is not limited to this. For example, a distribution pattern of pixels indicating a target part in the generated three-dimensional image may be obtained as foreseeing information, and a pixel at a position that matches the pattern may be set as a correction target pixel.

  In the above-described embodiment, the case where the program is preinstalled in the apparatus has been described. However, the medical image processing program of the present embodiment may be stored in a computer-readable recording medium and provided. It may be provided via a network.

10, 210 Medical image processing device 12, 212 Computer 14 Display device 20 3D image generation unit 22 Processing region setting unit 24 Pixel value correction unit 26 Labeling processing unit 28, 228 Display control unit 30 Other part extraction unit

Claims (6)

Generating means for generating a three-dimensional image by stacking a plurality of non-contrast tomographic images obtained by imaging a body including a target region without using a contrast agent;
A region having a pixel value greater than or equal to a predetermined value is extracted as a bone region from the three-dimensional image generated by the generating unit, and the bone regions corresponding to a predetermined number of consecutive images in the plurality of non-contrast tomographic images are projected in the stacking direction. Setting means for setting a processing region including the target part for each of the predetermined number based on a bone region projection image configured by a pixel value that is maximum on the projection line when
Based on the distribution of pixel values in the processing area set by the setting means in the three-dimensional image and the relationship between the target part and the distribution of pixel values obtained in advance, a predetermined range of pixel values indicating the target part is obtained. Correction means for setting and shifting the pixel values included in the predetermined range and correcting the width of the predetermined range to expand;
In the three-dimensional image in which the pixel value is corrected by the correction unit, an area where pixels having corrected pixel values indicating the target portion are connected and the number of pixels in the area is maximized is extracted. Target part extraction means to perform,
Display control means for controlling the display means so that the region extracted by the target part extraction means is displayed as a three-dimensional image;
A medical image processing apparatus.   The medical image processing apparatus according to claim 1, wherein the setting unit interpolates, with a spline curve, a location where pixels indicating bone regions are discrete in the bone region projection image.   The correction means includes a distribution of pixel values of a three-dimensional image based on a plurality of contrast tomographic images obtained by imaging a body including a target region using a contrast agent, and a pixel value of a three-dimensional image based on the plurality of non-contrast tomographic images. The medical image processing apparatus according to claim 1, wherein a predetermined range of pixel values indicating the target portion is set based on the relationship obtained in advance by comparison with the distribution of the medical image. A region expanded by a region expansion method is extracted from the three-dimensional image, and pixel values of pixels included in the expanded region are converted into distance values indicating the shortest distance between the expanded region and the background. , Based on the distance value, the expanded region is separated at a constricted portion of the region, and based on at least one of the size and shape of each separated region, a region indicating another part different from the target part Including other part extracting means for extracting
The display control means controls the display means so that a difference area between the area extracted by the target part extracting means and the area extracted by the other part extracting means is displayed as a three-dimensional image. The medical image processing apparatus according to any one of claims 3 to 4. Computer
Laminating a plurality of non-contrast tomographic images of the body including the target region without using a contrast agent to generate a three-dimensional image,
Projected when a region having a pixel value equal to or greater than a predetermined value is extracted from the generated three-dimensional image as a bone region, and the bone regions corresponding to a predetermined number of consecutive non-contrast tomographic images are projected in the stacking direction Based on a bone region projection image composed of pixel values that are maximum on a line, a processing region including the target part is set for each predetermined number of images,
Based on the distribution of pixel values in the processing region set in the three-dimensional image and the relationship between the target portion and the distribution of pixel values obtained in advance, a predetermined range of pixel values indicating the target portion is set, The pixel value included in the predetermined range is shifted and corrected so that the width of the predetermined range is expanded.
In the three-dimensional image in which the pixel value is corrected, an area where pixels having a corrected pixel value indicating the target portion are connected and the number of pixels in the area is maximized is extracted,
A medical image processing method for controlling display means so that an extracted region is displayed as a three-dimensional image.   The medical image processing program for functioning a computer as each means which comprises the medical image processing apparatus of any one of Claims 1-4.