JP5536987B2 - Image processing device - Google Patents

Description

  The present invention relates to an image processing apparatus for supporting diagnosis of a nodule region included in volume data generated by an X-ray computed tomography apparatus.

  In CT examinations for patients suspected of having lung cancer, it is important to compare the volume of nodule areas over time. If the nodular area shows a tendency to expand, it becomes an important material for judging malignancy. As a method for examining changes in the mass region, there is a method of visual comparison. However, a method for examining changes in diameter and volume is important as an objective index. In order to investigate the change of the nodule region, a technique for extracting the nodule region and measuring and displaying the volume has been proposed.

  The simplest method for extracting a lung nodule region from a chest CT image uses simple threshold processing. However, nodules and blood vessels (or bronchi) cannot be distinguished by threshold processing because the nodules are connected with structures having pixel values close to soft tissues such as blood vessels and bronchi.

  In order to distinguish a nodule from a blood vessel, a nodule region extraction method by a level set method (Non-Patent Document 1) and a blood vessel separation method by a morphological operation (Non-Patent Document 2) have been proposed.

  The former extraction method uses an iterative method. Therefore, the extraction process takes a very long time. When used as an analysis function of medical image interpretation software, processing time within several seconds is required, but this extraction processing requires much longer processing time. Therefore, this extraction method is not practical.

In the latter blood vessel separation method, erosion processing and dilation processing are performed on the nodule candidate region obtained by the region growing method. However, if there is a small protrusion of, for example, less than 4 mm in a part of the nodule, the protrusion is cut by mistake. If the erosion distance is reduced and the dilation distance is increased in order not to cut the protrusion, a thick blood vessel cannot be separated, or even if it can be separated, the cutting position becomes far from the base of the nodule. If the erosion distance is increased to separate a thick blood vessel, the small nodule is completely removed and no nodule region is extracted at all.
Segmentation of ground glass opacities by asymmetric multi-phase deformable model Proc.SPIE Vol. 6144 614440-1〜614440-8 Morphologycal segmentation and partial volume analysis of volumetry of solid pulmonary lesions in thoracic CT scans.IEEE Trans. Med. Imag., Vol. 25, No. 4, 2006

  An object of the present invention is to provide an image processing apparatus capable of extracting a nodule region in a short time and with high accuracy.

An image processing apparatus according to a first aspect of the present invention includes a storage unit that stores volume data related to a subject, a nodule candidate region extraction unit that extracts a nodule candidate region included in the volume data based on pixel values, Based on a nuclear region determination unit that determines a nuclear region of the extracted nodule candidate region, a first distance from the determined nuclear region, and a second distance longer than the first distance, the nodule An image processing apparatus comprising: a non-nodule region specifying unit that specifies a non-nodule region from a candidate region; and a nodule region extracting unit that extracts a nodule region by removing the specified non-nodule region from the nodule candidate region. The non-nodule region specifying unit is connected to a region that is separated from the nucleus region by the second distance or more, and a region that is separated from the nucleus region by the first distance or more is the non-nodule region. age Identifying, characterized in that.

  According to the present invention, it is possible to provide an image processing apparatus capable of extracting a nodule region in a short time and with high accuracy.

  Hereinafter, an image processing apparatus according to an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a diagram illustrating a configuration of an image processing apparatus 1 according to the present embodiment. As shown in FIG. 1, the image processing apparatus 1 has a storage unit 12, a target region determination unit 14, a nodule candidate region extraction unit 16, a hole filling unit 18, a nucleus region determination unit 20, and a nodule region extraction with the control unit 10 as a center. Unit 22, image generation unit 24, volume calculation unit 26, display unit 28, and input unit 30.

  The storage unit 12 stores volume data related to the chest of the subject. The volume data is generated by a medical image generation apparatus such as an X-ray computed tomography apparatus (not shown). It is assumed that the volume data includes a nodule region generated in the lung.

  FIG. 2 is a diagram illustrating an example of the nodule region NR. As shown in FIG. 2, the nodule region NR has a small protrusion region TR which is a part of the nodule. Further, the nodule region NR is connected to a blood vessel or bronchial region HR in addition to the small projection region TR. The blood vessel region and the bronchial region HR are not part of the nodule, but are parts that obstruct the observation of the nodule region NR. Here, a blood vessel region or a bronchial region connected to the nodule region NR is referred to as a non-nodule region HR. Further, the small protrusion region TR and the non-nodule region HR have a protrusion shape in appearance. Therefore, the small protrusion region TR and the non-nodule region HR are collectively referred to as a protrusion candidate region.

  The nodule region is roughly divided into a solid region and a GGO region according to the pixel value. The solid region is also called a solid shadow. The solid region is a region having a pixel value close to soft tissue in the nodule region. The GGO region is also called ground glass-like shadow. The GGO region is a region having an intermediate pixel value between the parenchyma and soft tissue of the lung.

  The target area determination unit 14 determines an area to be processed from the volume data. The processing target area is, for example, a cube with a side of 30 mm. The shape and size of the processing target area can be changed.

  The nodule candidate region extraction unit 16 generates a region that is a nodule region candidate from the determined processing target region. The nodule candidate region includes a nodule region and a non-nodule region. The nodule candidate region extraction unit 16 extracts a nodule candidate region from the processing target region using, for example, existing threshold processing, P-tile-thresholding, Mode method, Optimal threshold, and Otsu's method.

  The hole filling unit 18 performs image processing on the generated nodule candidate region, and fills the hole region existing inside the nodule candidate region. The hole region is a region that does not originate from a nodule existing inside the nodule candidate region.

  The nucleus region determination unit 20 determines a nucleus region representing the center of the nodule candidate region after filling in the holes. Specifically, the nucleus region determination unit 20 determines the nucleus region by reducing the nodule candidate region after filling the hole with the erosion distance De (erosion processing of morphological operation). The nucleus region serves as a reference for specifying the projection region and the non-nodule region from the projection candidate regions included in the nodule candidate region after filling in the holes. The nucleus region determination unit 20 can correct the nucleus region based on an instruction input via the input unit 30 by the user.

  The nodule region extraction unit 22 identifies and identifies the nodule region from the nodule candidate region after filling based on the dilation distance Dmin from the determined nucleus region and the dilation distance Dmax longer than the dilation distance Dmin. The nodule region is extracted by removing the non-nodule region from the nodule candidate region after filling the hole. The removal range of the non-nodule region includes an inner portion directed from the dilation distance Dmax toward the nucleus region. The nodule region extraction unit 22 can correct the nodule region based on an instruction input via the input unit 30 by the user.

  The volume calculation unit 24 calculates the volume of the extracted nodule region. The image generator 26 generates CT image data based on the volume data. For example, the image generation unit 26 performs MPR (MultiPlanar Reconstruction) processing on the volume data to generate CT image data related to an arbitrary cross section. The display unit 28 displays a CT image by emphasizing the generated nodule region. The display unit 28 displays the calculated volume.

  The input unit 30 inputs various instructions from the operator. As the input unit 30, a pointing device such as a mouse or a trackball, or an input device such as a keyboard can be used as appropriate.

  The control unit 10 includes, for example, a CPU and a memory, and develops an application program on the memory, and the CPU executes processing according to the application program. By executing the application program, the control unit 10 controls each unit constituting the image processing apparatus 1 and performs the nodule region extraction / display processing according to the present embodiment.

(Nodule area extraction / display processing)
The operation of the nodule region extraction / display process performed under the control of the control unit 10 will be described below. FIG. 3 is a diagram showing the flow of the nodule region extraction / display processing.

  First, upon receiving a nodule region extraction / display processing start instruction, the control unit 10 causes the processing target region determination unit 14 to perform center point determination processing.

  In the center point determination process, the processing target region determination unit 14 determines the center point of the nodule region (step SA). Specifically, the processing target area determination unit 14 determines a point designated by the user via the input unit 30 as a central point. Alternatively, the processing target area determination unit 14 may calculate the center point by existing CAD processing.

  After the center point is determined, the control unit 10 causes the processing target region determination unit 14 to perform processing target region determination processing (step SB). In the processing target area determination process, the processing target area determination unit 14 determines a certain range of areas centered on the center point as the processing target area. For example, a cubic region having a side of 30 mm with the center point as the center is determined as the processing target region. The determined processing target area is cut out from the volume data.

  After the processing area determination processing, the control unit 10 causes the nodule candidate region extraction unit 16 to perform nodule candidate region extraction processing (step SC). In the nodule candidate region extraction process, the nodule candidate region extraction unit 16 uses the pixel value of the boundary between the pixel value of the nodule candidate region and the pixel value other than the nodule candidate region (for example, −700 HU) as a threshold value from the processing target region. Extract candidate areas. More specifically, a nodule candidate region is extracted from the volume data by replacing a voxel having a pixel value of −700 HU or less with a value “1” and a voxel having a pixel value of −700 or more with a value “0”. . Typically, by using a plurality of threshold values, the solid candidate area and the GGO candidate area are extracted separately from the volume data. The threshold value may be set in advance, but there are non-uniformity in concentration within the lung area and differences among individuals. Therefore, the threshold value is determined adaptively by taking these into consideration and determining using a histogram. The way to do it is good. When the threshold processing is performed, the nodule candidate region extraction unit 16 calculates a logical sum of all the extracted solid candidate regions and the GGO region, and generates a sum region. The generated sum area is referred to as a nodule candidate area.

  After the nodule candidate region extraction processing, the control unit 10 causes the hole filling unit 18 to perform hole filling processing (step SD). In the hole filling process, the hole filling unit 18 performs image processing on the nodule candidate area and fills the hole area existing in the nodule candidate area. The nodule candidate region in which the hole region is filled is referred to as a post-filling nodule candidate region. Details of the hole filling process will be described later.

  After the hole filling process, the control unit 10 causes the nucleus region determination unit 20 to perform a nucleus region determination process (step SE). In the nuclear region determination process, the nuclear region determination unit 20 performs erosion processing on the nodule candidate region after filling with the erosion distance De, and determines the post-filling nodule candidate region subjected to erosion processing as the nuclear region.

  FIG. 4 is a diagram illustrating the nodule candidate region AR and the nucleus region CR after filling in holes. As shown in FIG. 4, the nodule candidate region AR after filling has projection candidate regions such as a projection region TR and a non-nodule region HR. The nucleus region CR is a region obtained by reducing the nodule candidate region AR after filling by the erosion distance De by the erosion process. Since the projection region TR and the non-nodule region HR are removed by the erosion process, the nucleus region CR does not have a projection shape derived from the projection region TR and the non-nodule region HR. As will be described later, the nucleus region CR serves as a base point of a distance Dmin and a distance Dmax for distinguishing the projection region TR and the non-nodule region HR from the projection candidate region. Details of the nuclear region setting process will be described later.

  After the nuclear region determination process, the control unit 10 causes the nodule region extraction unit 22 to perform a nodule region extraction process (step SF).

  The outline of the nodule region extraction processing by the nodule region extraction unit 22 will be described below with reference to FIGS. 5 and 6. As shown in FIG. 5, the core region CR is expanded (dilation processing of morphological operation) using the nodule candidate region AR after filling in the holes and the dilation distance Dmin. Of the nodule candidate region AR after filling in, the region that is separated from the nucleus region CR by the distance Dmin or more is identified as the projection candidate region KR. Further, the nodule region extraction unit 22 dilates the nucleus region CR with the dilation distance Dmax. Here, an area between the distance Dmin and the distance Dmax is referred to as a nodule margin area ZR. If the region protruding outside the nodule margin region ZR is the non-nodule region HR, the region that does not protrude is specified as the small protrusion region TR. That is, among the projection candidate regions, the region connected to the region outside the distance Dmax is determined as the non-nodule region HR, and the region not connected is determined as the small projection region TR. More specifically, the position portion of the non-nodule region HR is specified from the nodule candidate region AR after filling based on the pixel value of the voxel located at the distance Dmax from the nucleus region CR. Here, when the value “1” is assigned to the voxel at the dilation distance Dmax, it is specified as a part of the non-nodule region HR. And the area | region connected with the specified part among processus | protrusion candidate areas is specified as non-nodule area | region HR. The nodule region NR is extracted by removing the identified non-nodule region HR from the nodule candidate region AR after filling in. This removal range includes a portion inside the dilation distance Dmax.

  In order to remove the non-nodule region from the nodule candidate region after filling in the hole, the nodule region extraction processing specifically includes a nodule mask region generation process and a logical product calculation process. In the nodule mask region generation processing, the nodule region extraction unit 22 performs image processing on the nodule candidate region after filling and generates a nodule mask region used as a mask for removing the non-nodule region from the nodule candidate region after filling. The nodule mask region is a sum region of the region within the dilation distance Dmin and the small protrusion region TR. The nodule mask area generation process is a main process of the nodule area extraction process. The nodule mask region generation process will be described later.

  In the logical product calculation process, the nodule region extraction unit 22 calculates a logical product of the nodule mask region and the nodule candidate region to generate a product region. The generated product area is a nodule area obtained by removing the non-nodule area from the nodule candidate area. More specifically, the nodule region extraction unit 22 calculates a logical product of the nodule mask region and the solid candidate region to generate a product region. The generated product area is a solid area of the nodule area. Similarly, the nodule region extraction unit 22 calculates a logical product of the nodule mask region and the GGO candidate region to generate a product region. The generated product area is a GGO area in the nodule area.

  When the nodule region is generated, the control unit 10 causes the display unit 28 to perform a nodule region display process (step SG). In the nodule region display processing, the display unit 28 emphasizes the generated nodule region with a color or the like and displays a CT image. Note that the emphasis method may be enhanced not only by color but also by brightness or brightness.

  If necessary, when the nodule region is generated, the control unit 10 causes the volume calculation unit 24 to perform a volume calculation process (step SH). In the volume calculation process, the volume calculation unit 24, most simply, counts the number of voxels in the nodule region, and calculates the volume of the nodule region based on the counted number of voxels. The volume calculation process is performed individually for the solid area and the GGO area.

  When the volume is calculated, the control unit 10 causes the display unit 28 to perform volume display processing (step SG). In the volume display process, the display unit 28 displays the calculated volume value in a predetermined layout.

  This is the end of the nodule region extraction / display processing. Next, details of the hole filling process (step SD), the nucleus area determination process (step SE), and the nodule mask area generation process (step SF) will be described in order.

(Clogging process)
FIG. 7 is a diagram showing a flow of the hole filling process performed by the hole filling unit 18 in step SD. As shown in FIG. 7, the hole filling unit 18 dilates the nodule candidate region generated by the nodule candidate region generation process with the dilation distance D1 (step SD1). The dilation distance D1 is preset and is 1 mm, for example. By the dilation processing, the nodule candidate region is expanded by the dilation distance D1, and a hole region smaller than the distance D1 is filled.

  After the dilation processing, the hole filling unit 18 identifies the background region (the voxel having the value “0”) included in the nodule candidate region subjected to the dilation processing, and replaces it with the foreground region (the voxel having the value “1”). (Step SD2). By this replacement process, it is possible to fill a hole area that cannot be filled with the dilation distance D1 or a hole area connected to the background area.

  After the replacement process, the hole filling unit 18 performs an erosion process on the nodule candidate area subjected to the replacement process with the erosion distance D2 (step SD3). The erosion distance D2 is set in advance and is longer than the dilation distance D1, for example, 1.5 mm. That is, the nodule candidate region subjected to the erosion process is smaller than the original nodule candidate region (the nodule candidate region generated by the nodule candidate region generation process).

  After the erosion process, the hole filling unit 18 calculates a logical sum of the original nodule candidate area and the nodule candidate area subjected to the erosion process, and generates a sum area (a nodule candidate area after filling) (step SD4). The generated post-filling nodule candidate region has the shape of the original nodule candidate region and does not have a hole region.

(Nuclear region determination processing)
Next, details of the determination process for the nuclear region performed by the nuclear region determination unit 20 in step SE will be described.

  FIG. 8 is a diagram showing the flow of the nuclear region determination process. As shown in FIG. 8, the nucleus region determination unit 20 performs a distance conversion process on the volume data relating to the post-filling nodule candidate region generated by the filling process (step SE1). The distance conversion process in step SE1 is a process of assigning a distance value from the background area (the voxel assigned the value “0”) to each voxel constituting the volume data related to the nodule candidate area after filling. By the distance conversion process in step SE1, distance data A in which the distance from the background area is assigned to each voxel is generated.

  When the distance data A is generated, the nuclear region determination unit 20 sets a seed point based on the distance data A and the center point determined by the center point determination process (step SE2). Specifically, the seed point is a voxel assigned a maximum distance around the center point.

  When the seed point is determined, the nuclear region determination unit 20 determines an erosion distance De suitable for the size of the nodule candidate region after filling (step SE3). Specifically, the erosion distance De is the smaller of a preset distance (for example, 3 mm) and half of the distance value assigned to the seed point. That is, when the nodule candidate region after filling is smaller than a preset distance (3 mm in the above example), the erosion distance De is determined to be half of the distance value assigned to the seed point. Thus, by providing the lower limit of the erosion distance De, it is possible to prevent disappearance of the nodule candidate region after hole filling by erosion processing. On the other hand, when the post-filling nodule candidate region is larger than a preset distance, the erosion distance De is set to a preset distance. By thus setting the upper limit of the erosion distance De, it is possible to prevent the core region from being greatly deformed by excessive erosion processing.

  When the erosion distance De is determined, the nucleus region determination unit 20 generates a nucleus candidate region by performing erosion processing on the nodule candidate region after filling with the determined erosion distance De (step SE4). The generated area may include isolated points generated by erosion processing or the like.

  In order to remove these isolated points, the nucleus region determination unit 20 removes a region that is not connected to the seed point from the nucleus candidate regions, and determines a region that is connected to the seed point as a nucleus region (step SE5). Specifically, the candidate region is expanded from the seed point determined in step SE2, and a region connected to the seed point is specified. The identified area is determined as a nuclear area. Of the nuclear candidate regions, regions that are not connected to the seed points are removed.

(Nodule mask area generation processing)
Next, the nodule mask region generation process executed by the nodule region extraction unit 22 in step SF will be described.

  FIG. 9 is a diagram illustrating a flow of a nodule mask region generation process. As shown in FIG. 9, the nodule region extraction unit 22 performs distance conversion on the volume data related to the nodule candidate region after hole filling (step SF1). The distance conversion in step SF is a process of assigning a distance value from the nucleus region to each voxel constituting the nodule candidate region after hole filling. By the distance conversion process in step SF1, distance data B in which the distance from the nucleus region to the outside of the nucleus region is assigned to each voxel is generated.

  After the distance conversion process, the nodule region extraction unit 22 determines the dilation distance Dmin based on the distance data B (step SF2). The dilation distance Dmin is slightly larger than or equal to the erosion distance De.

  When the erosion distance Dmin is determined, the nodule region extraction unit 22 calculates a logical product of the region related to the distance equal to or greater than the erosion distance Dmin and the nodule candidate region after filling, and generates a product region (step SF3). The product area to be generated is a protrusion candidate area.

  FIG. 10 is a diagram illustrating the process of step SF3. In FIG. 10, the process of step SF3 is expressed two-dimensionally for simplicity, but actually step SF3 is processed three-dimensionally. As shown in FIG. 10, the region DOR related to the distance greater than or equal to the distance Dmin is a region that is separated from the nucleus region CR by the dilation distance Dmin or more. That is, the voxels constituting the region DOR have a distance value equal to or greater than the distance Dmin in the distance data B. By calculating the logical product of the region DOR and the post-filling nodule candidate region AR, the product region, that is, the projection candidate region CR in the post-filling nodule candidate region AR is extracted. That is, the dilation distance Dmin is a reference for determining whether or not it is a projection candidate region.

  When the projection candidate area is generated, the nodule area extraction unit 22 determines the dilation distance Dmax (step SF4). The distance Dmax can be arbitrarily set by the user via the input unit 30 even if it is set in advance. When arbitrarily setting the distance Dmax, the user inputs, for example, the width W of the nodule margin region (the difference between the distance Dmax and the distance Dmin). Then, the distance Dmax is determined according to the calculation formula of the distance Dmax, that is, the distance Dmax = the distance Dmin + the width W of the nodule margin region.

  When the dilation distance Dmax is determined, the nodule region extraction unit 22 sets a region related to a distance equal to or greater than the dilation distance Dmax as a seed region (step SF4). The seed region includes, for example, a portion of the projection candidate region having a distance Dmax or more in addition to the background region having a distance Dmax or more.

  When the seed region is set, the nodule region extraction unit 22 identifies a region connected to the seed region among the projection candidate regions as a non-nodule region (step SF5).

  FIG. 11 is a diagram illustrating the process of step SF5. In FIG. 11, for the sake of simplicity, the process of step SF is expressed two-dimensionally. However, step SF is actually processed three-dimensionally. As shown in FIG. 11, an area that is longer than the dilation distance Dmax is set as a seed area. The non-nodule region HR is specified by subjecting the projection candidate region KR to the region expansion process from the seed region. Of the projection candidate regions KT, a region that is not connected to the seed region is identified as the small projection region TR. In other words, of the regions constituting the projection candidate region KR, a region where the distance from the nucleus region CR reaches the dilation distance Dmax is regarded as a non-nodule region HR, and the distance from the nucleus region CR reaches the distance Dmax. The area not to be treated is regarded as the projection area TR. That is, the dilation distance Dmax is a reference for determining whether or not it is a non-nodule region.

  When the non-projection candidate region is specified, the nodule region extraction unit 22 removes the non-nodule region from the projection candidate region and generates a small projection region (step SF6). Specifically, the non-nodule region is removed by assigning the value “0” to each voxel constituting the non-nodule region.

  When the small protrusion area is generated, the nodule area extraction unit 22 calculates a logical sum of the generated small protrusion area and an area related to a distance equal to or less than the dilation distance Dmin, and generates a sum area (step SF7). The generated sum area is a nodule mask area.

  FIG. 12 is a diagram illustrating the process of step SF7. In FIG. 12, for the sake of simplicity, the process of step SF is expressed two-dimensionally. However, step SF is actually processed three-dimensionally. As shown in FIG. 12, the region DMR related to the distance equal to or less than the distance Dmin is a region within the distance Dmin from the nucleus region CR. In other words, the voxels constituting the region DMR have a distance value equal to or smaller than the distance Dmin in the distance data B. A sum area, that is, a nodule mask area MR is generated by calculating a logical sum of the area DMR and the small protrusion area TR.

  This completes the nodule mask region generation process. After generating the nodule mask region, the nodule region is removed from the nodule candidate region and the nodule region is extracted by calculating the logical product of the nodule mask region and the nodule region. At this time, the non-nodule region can be removed from the distance Dmin instead of the dilation distance Dmax determined to be the non-nodule region.

  As described above, the nodule region extraction processing according to the present embodiment includes simple processing such as threshold processing, distance conversion, dilation processing, erosion processing, logical operation, and region expansion processing. Therefore, this extraction process can be executed at a relatively high speed. Further, this extraction process is not an iterative process as described in Non-Patent Document 1. Therefore, the nodule region can be extracted in a shorter time than the method described in Non-Patent Document 1. In the trial software by the present inventors, the time required for the entire extraction process is approximately 1.5 seconds.

  Moreover, according to this extraction process, a small protrusion can be completely restored to its original shape. In the conventional method described in Non-Patent Document 2, when the erosion distance is increased, the nodule region is deformed. In this extraction process, the upper limit of the erosion distance is expanded. Even if the erosion distance for obtaining the nucleus region is increased, the blood vessel can be cut from the root. Therefore, the erosion distance can be set longer than the conventional method described in Non-Patent Document 2. If the erosion distance is set longer, a thicker blood vessel can be separated from the nodule region.

  Next, various application examples of the nodule region extraction / display processing according to the present embodiment will be described. Application examples include a process of determining a spicula from a non-blood vessel area, a process of correcting a nodule area, and a process of correcting a process target area. Hereinafter, these processes will be described in order.

(Spikura judgment process)
In addition to blood vessels and bronchi, a spicule is connected to the nodule. Since the spicula is a growth of tumor cells, it is preferably recognized as a part of a nodule, that is, a protruding region. The origin of the spicula varies, but even the longest one rarely exceeds 20 mm. On the other hand, when blood vessels and bronchi are connected to nodules, they are connected to the hilar region. Therefore, it is possible to distinguish a spicula from a blood vessel (or bronchus) depending on whether it exceeds about 20 mm.

  Spikler determination processing is incorporated into the nodule mask region generation processing. Hereinafter, a process for generating a nodule mask region in which a spiker determination process is performed, which is performed by the nodule region extraction unit 22, will be described with reference to FIG. FIG. 13 is a diagram illustrating a flow of a process for generating a nodule mask region in which a spiker determination process is incorporated. As shown in FIG. 13, steps SJ1 to SJ5 are the same as steps SF1 to SF5 of the nodule mask region generation process of FIG.

  As illustrated in FIG. 13, when the seed region is determined, the nodule region extraction unit 22 identifies a region connected to the seed region among the projection candidate regions as a non-nodule candidate region (step SJ6). The specified non-nodule candidate region includes not only a blood vessel and a bronchus to be included in the non-nodule region, but also a spiker region that should not be included in the non-nodule candidate region. The non-projection candidate area is the same as the projection candidate area in FIG.

  When the non-nodule candidate region is specified, the nodule region extraction unit 22 generates a non-nodule region by removing the region having the length Ds or less from the non-nodule candidate region (step SJ7). The length Ds is a reference for determining whether or not each region is a spicula region. Hereinafter, the length Ds is referred to as a blood vessel determination distance Ds. The process of step SJ7 will be specifically described. First, the length of each area constituting the non-nodule candidate area is calculated. The length is a major axis or a diagonal line of a figure generated by approximating each region. Of each region, a region having a blood vessel determination distance Ds (for example, 20 mm) or less is determined to be a spicula region, and a region having a blood vessel determination distance Ds or more is determined to be a non-nodule region such as a blood vessel or bronchus. Then, the spiker region is removed from the non-projection candidate region. By removing the spiker region from the non-nodule candidate region, a non-nodule region is generated.

  When the non-nodule region is generated, the nodule region extraction unit 22 generates a small projection region by removing the non-nodule region from the projection candidate region (step SJ8). The microprojection regions that are generated include not only small projections but also spicula.

  When the small projection region is generated, the nodule region extraction unit 22 generates a nodule mask region which is a sum region of the region related to the distance less than the dilation distance Dmin and the small projection region (step SJ9).

  This completes the nodule mask region generation process in which the spikler determination process is incorporated. Note that the method for determining a spicula is not limited to the above method. For example, by determining the dilation distance Dmax as the dilation distance Dmax = the erosion distance De + the blood vessel determination distance Ds, the spiker region is included in the nodule mask region in the same process as the nodule mask area generation process of FIG. Can do. In this method, it is not strictly determined whether or not the length of each region constituting the non-nodule candidate region is actually within 20 mm. However, when each region extends from the nodule at a substantially right angle, the result is almost the same as that determined by the length of the protruding structure.

(Nodule area correction processing)
The above-described nodule region extraction processing cannot always distinguish a small projection region and a non-nodule region for any nodule. Therefore, nodule region correction processing is incorporated in the nodule region extraction processing.

  First, the display unit 28 displays a CT image in which the nodule region is emphasized in a predetermined layout. And the control part 10 receives the input of the correction point via the input part 30. FIG. The user designates a point to be corrected via the input unit 30.

  FIG. 14 is a diagram illustrating an example of a screen for designating correction points. Once the nodule region is extracted, the display unit 28 displays the CT image I1 in which the nodule region is emphasized, and an operation screen I2 for performing various processing start instructions, volume display, parameter setting, and the like. . On the operation screen I2, a width setting / display column D1 of the nodule margin region, a setting / display column D2 of the blood vessel determination distance Ds, and a volume display column D3 are displayed. In addition, on the operation screen I2, a start button BA for extracting a nodule region, a call button BB for a correction menu, and a start button BC for a volume calculation process are displayed.

  When the user presses the correction menu call button BB via the input unit 30, the correction menu MN is displayed on the CT image I1. The correction menu MN displays various buttons related to the correction process. Specifically, the correction menu MN includes an Add button B1, a Delete button B2, and a Limit button BC. The Add button B1 is a button for designating a correction point for adding a protrusion to the nodule region. A correction point designated on the screen in a state where the Add button B1 is selected is referred to as a nodule inner point. The Delete button B2 is a button for designating a correction point for removing the protrusion from the nodule region. A correction point designated on the screen in a state where the Delete button B2 is selected is referred to as an extranodal point. The Limit button B3 is a button for designating a correction point for correcting the cutting position of the protrusion. A correction point designated on the screen in a state where the Limit button B3 is selected is referred to as a nodule boundary point. The code indicating the type of the specified correction point and the coordinate value are associated with each other and stored in the storage unit 12 in a list format.

  When the specification of the correction point is completed and the extraction start button BA is selected, the correction process corresponding to the type of correction point is started.

  Hereinafter, correction processing when the nodule inner point, the nodule outer point, and the nodule boundary point are designated will be described. Since the correction processing related to the nodule inner point and the nodule outer point are almost the same, the correction processing related to the nodule boundary point will be described.

[Nodal inner point and outer nodal point]
FIG. 15 is a diagram illustrating a flow of a process of generating a nodule mask region regarding the nodule inner point and the nodule outer point. As shown in FIG. 15, steps SK1 to SK7 are the same as steps SF1 to SF7 of the nodule mask region generation process of FIG.

  When the user specifies the nodule inner point or the nodule outer point via the input unit 30 and the extraction start button is pressed, the nodule region extraction unit 22 performs the projection correction processing A related to the nodule inner point or the nodule outer point. Start (step SK8).

  FIG. 16 is a diagram illustrating a flow of the protrusion correction process A performed by the nodule region extraction unit 22 in step SK8. As illustrated in FIG. 16, when the extraction start button is pressed, the nodule region extraction unit 22 refers to the code of the type of correction point to be processed stored in the storage unit 12 and determines the correction point to be processed. It is determined whether or not the point is a nodule inner point (step SK8-1).

  When it is determined that the correction point is a nodule inner point (step SK8-1: YES), the nodule region extraction unit 22 performs region expansion processing from the nodule inner point, and specifies a region connected to the nodule inner point (step). SK8-2). The identified area is called an inner area. When the inner region is specified, the nodule region extracting unit 22 includes the inner region in the small protrusion region (step SK8-3). Specifically, the inner area is included in the small protrusion area by calculating the logical sum of the inner area and the small protrusion area.

  If it is determined in step SK8-2 that the correction point is not a nodule inner point (step SK8-2: NO), the nodule region extraction unit 22 determines whether the correction point is an outer nodule point (step SK8-4). ). When it is determined that the point is a nodule outside point (step SK8-4: YES), the nodule region extraction unit 22 performs region expansion processing from the nodule outside point and specifies a region connected to the nodule outside point (step SK8-5). ). The area specified in step SK8-4 will be referred to as an outer area. When the outer region is specified, the nodule region extracting unit 22 removes the outer region from the small protrusion region (step SK8-6). Specifically, volume data obtained by inverting the pixel values of all voxels constituting volume data relating to the outer area, and calculating the logical product of the inverted volume data and the volume data relating to the small protrusion area, the protrusion area Remove outer area from

  If it is determined in step SK8-4 that the correction point is not a nodule outside point, the nodule region extraction unit 22 proceeds to step SK8-7.

  When step SK8-3 or step SK8-6 is performed, or when it is determined in step SK8-4 that it is not a nodule outside point, the nodule region extraction unit 22 lists correction points stored in the storage unit 12 To determine whether there is an unprocessed correction point (step SK8-7). When it is determined that there is an unprocessed correction point (step SK8-7: YES), the nodule region extraction unit 22 proceeds to step SK8-1.

  Hereinafter, the nodule region correction process for the nodule inner point and the nodule outer point will be described in detail. As shown in FIG. 17, the nodule inner point is included in the small protrusion area before the protrusion correction process A at one point of the area R1 that is not included in the small protrusion area before the protrusion correction process A in the protrusion candidate area KR. Assume that a nodule point is designated as one point in the region R2. First, in step SK8-2, the region R1 including the nodule inner point is specified as the inner region. In step SK8-3, as shown in FIG. 18, the inner region R1 is removed from the non-nodule region and included in the small protrusion region. Therefore, the region R1 that has been deleted as the non-nodule region before the projection correction process A can be included in the small projection region. On the other hand, in step SK8-5, the region R2 including the nodule outer point is specified as an outer region as shown in FIG. In step SK8-6, the outer region R2 is removed from the small protrusion region and included in the non-nodule region as shown in FIG. Accordingly, the region R2 included in the small protrusion region before the protrusion correction process A can be removed from the small protrusion region.

  On the other hand, when it is determined that there is no unprocessed correction point (step SK8-7: NO), the nodule region extraction unit 22 ends the protrusion correction processing A.

  As shown in FIG. 15, when the projection correction processing A is completed, the nodule region extraction unit 22 sums the region related to the distance equal to or less than the distance Dmin and the small projection region after the projection correction processing A (corrected nodule mask region). Is generated (step SK9). By applying the generated nodule mask region after correction to the nodule candidate region, the nodule region extraction unit 22 includes the inner region connected to the nodule inner point in the projection candidate region, The outer region connected to the nodule outer point can be removed from the nodule region.

  As described above, when the user designates the nodule inner point and the nodule outer point via the input unit 30, the area determined as the small protrusion area before the protrusion correction process A is removed from the nodule area as desired by the user. Or a region that has been determined to be a non-nodule region can be included in the nodule region.

[Nodal boundary point]
The nodule boundary point is a correction point for designating the cutting position of the projection region. First, the clinical significance of the nodule boundary point will be described with reference to FIG. FIG. 19 is a diagram showing a nodule region NR to which a non-nodule region HR ′ derived from a long and long blood vessel or bronchus is connected. As shown in FIG. 19, when a blood vessel or bronchus is long and long, the shape of the blood vessel or bronchi remains even by erosion processing. Therefore, the nucleus region CR ′ has a protrusion shape derived from this long and long blood vessel or bronchus. Thus, if the nucleus region CR ′ has a protrusion shape derived from a long blood vessel or bronchus, the blood vessel or bronchus that should originally be deleted cannot be deleted. In such a case, by specifying a nodule boundary point at an arbitrary position on a long blood vessel or bronchus, the nucleus region CR ′ is corrected according to the specified position, and is identical to the specified nodule boundary point (or It is possible to cut blood vessels and bronchi at the substantially same position.

  Further, there are cases where it is necessary to cut the projection area from the middle position, for example, when it is not a small projection area but is erroneously determined as a small projection area. In such a case, by specifying the nodule boundary point at an arbitrary position on the small protrusion region, the small protrusion region can be cut at the position of the specified nodule boundary point.

  FIG. 20 is a diagram illustrating a flow of a nuclear region setting process when a nodule boundary point is designated. As shown in FIG. 20, steps SL1 to SL4 are the same as steps SE1 to SE4 of the setting process according to the first embodiment, and a description thereof will be omitted.

  The nucleus area determination unit 20 removes an area included within the erosion distance De from the designated nodule boundary point from the nucleus candidate area (step SL5). The region generated by this removal process is called a nucleus candidate region after the removal process.

  After the generation of the nuclear candidate region after the removal process, the nuclear region determination unit 20 removes the region that is not connected to the seed point determined in step SL2 from the candidate nuclear region after the removal process, and connects the seed point to the seed point. The region is determined as a modified nuclear region (step SL6).

  Hereinafter, step SL5 and step SL6 will be described with reference to FIG. 21 and FIG. 21 and 22 express the process of step SF two-dimensionally for simplicity, but in practice, step SF is processed three-dimensionally. As shown in FIG. 21, the first nodule boundary point KP1 is designated as a non-nodule region HR ′ having such a thickness and length that it cannot be removed even by the erosion process at the erosion distance De. It is assumed that the second nodule boundary point KP2 is specified in the middle of the small protrusion region TR ′. The nucleus candidate region CCR has a projection region derived from a long and long blood vessel. A portion of the protrusion of the nucleus candidate region CCR is included within the radius De centered on the nodal boundary point KP1. Further, the nucleus candidate region CCR is not reached within the radius De centered on the nodal boundary point KP2.

  FIG. 22 is a diagram showing nucleus candidate regions ZCR1 and ZCR2 after the removal process. As shown in FIG. 22, when the region within the radius De is removed, at least a part of the projection region of the nucleus candidate region CCR is removed. It is assumed that the nucleus candidate region is separated into each nucleus candidate region ZCR1 after the first removal processing and each nucleus candidate region ZCR2 after the second removal processing by being removed. Since each nucleus candidate region ZCR1 after the first removal processing includes the seed point SP determined in step SL2, it is set as a modified nucleus region in step SL6. On the other hand, each candidate nuclear region ZCR2 after the second removal process does not include the seed point SP, and is thus removed in step SL6.

  This completes the nuclear region setting process when a nodule boundary point is designated. After the nucleus region setting process, the control unit 10 causes the nodule region extraction unit 22 to perform a nodule mask region generation process in which the protrusion correction process B is incorporated.

  FIG. 23 is a diagram illustrating a flow of a process of generating a nodule mask area regarding a nodule inner point, an outer nodule point, and a nodule boundary point. As shown in FIG. 23, the nodule region extraction unit 22 performs distance conversion on the volume data relating to the nodule candidate region after filling (step SM1). The distance conversion in step SM1 is a process of assigning a distance value from the corrected core region to each voxel constituting the post-filling nodule candidate region. The distance conversion process in step SM1 generates distance data C in which the distance from the corrected nuclear region to the outside of the nuclear region is assigned to each voxel.

  After the distance conversion process, the nodule region extraction unit 22 determines the dilation distance Dmin based on the distance data C (step SM2).

  FIG. 24 is a diagram illustrating a positional relationship among the post-filling nodule candidate region AR, the corrected nuclear region ACR, and the dilation distance Dmin. As shown in FIG. 24, the corrected nuclear region ACR does not have a projection region derived from a long and long blood vessel. It can be seen that the dilation distance Dmin from the corrected nuclear region ACR passes through the vicinity of the root of a long and long blood vessel. As described above, the distance Dmin is the same as or slightly longer than the distance De. Therefore, the nodule boundary point KP1 is within a distance Dmin from the corrected nuclear region ACR.

  When the erosion distance Dmin is determined, the nodule region extraction unit 22 calculates a logical product of the region related to the distance equal to or greater than the erosion distance Dmin and the nodule candidate region after filling, and generates a product region (step SM3). The product area to be generated is a projection candidate area AKR after correction as shown in FIG.

  When the projection candidate area is generated, the nodule area extraction unit 22 determines the dilation distance Dmax (step SM4).

  When the dilation distance Dmax is determined, the nodule region extraction unit 22 sets a region related to a distance equal to or greater than the dilation distance Dmax as a seed region (step SM5).

  When the seed region is set, the nodule region extraction unit 22 identifies the region connected to the seed region among the corrected projection candidate regions as the corrected non-nodule region (step SM6). In the case of FIG. 25, the protrusion region AKR1 of the correction word derived from the long and long blood vessel is specified as the non-nodule region after correction.

  When the corrected non-projection candidate region is specified, the nodule region extraction unit 22 removes the corrected non-nodule region from the corrected projection candidate region and generates a corrected small projection region (step SM7).

  FIG. 26 is a diagram illustrating the modified small projection area ATR generated in step SM7. As shown in FIG. 26, the corrected non-nodule region derived from a long and long blood vessel is removed with the dilation distance Dmin as a boundary. However, since the nodule boundary point KP2 is longer than the dilation distance Dmin, the corrected small protrusion region ATR2 in which the nodule boundary point KP is specified has not been cut yet.

  In order to cut the corrected small protrusion region at the nodule boundary point, the nodule region extraction unit 22 performs a protrusion correction process B (step SM8). FIG. 27 is a diagram showing the flow of the protrusion correction process B. As shown in FIG. 27, steps SM8-1 to SM8-6 are the same as steps SK8-1 to SK8-6 of the protrusion correction process shown in FIG.

  If it is determined in step SM8-4 that the correction point is not a nodule outside point, the nodule region extraction unit 22 refers to the code of the type of correction point to be processed stored in the storage unit 12, and the correction point Is a nodule boundary point (step SM8-7).

  When it is determined that the correction point is a nodule boundary point (step SM8-7: YES), the nodule region extraction unit 22 specifies a region connected to the nodule boundary point among the corrected small protrusion candidate regions. (Step SM8-8). The specified area is called a boundary area.

  When the boundary region is specified, the nodule region extraction unit 22 specifies a region related to a distance greater than the distance at the nodule boundary point among the specified boundary regions (step SM8-9). The identified area is called an out-of-boundary area. For example, if the distance value assigned to the voxel relating to the nodule boundary point is 5 mm, the voxel assigned the distance value of 5 mm or more is specified from the voxels constituting the boundary region. In addition, a surface composed of voxels assigned a distance value of 5 mm is a cut surface.

  When the out-of-boundary region is specified, the nodule region extracting unit 22 deletes the out-of-boundary region from the small protrusion region (step SM8-10). Specifically, volume data obtained by inverting the pixel values of all voxels constituting the volume data related to the out-of-boundary area is generated, and the logical product of the inverted volume data and the volume data related to the modified small projection area is calculated. Thus, the out-of-boundary region is removed from the corrected small projection region. The small protrusion region is cut at the cut surface that is equidistant from the corrected nucleus region.

  When the inner area is included in the small protrusion area in step SM8-3, or when the outer area is removed from the small protrusion area in step SM8-6, the outside boundary area is changed from the corrected small protrusion area in step SM8-10. If it has been removed, or if it is determined in step SM8-7 that the correction point is not a nodule boundary point, the nodule region extraction unit 22 refers to the list of correction points stored in the storage unit 12 and is not yet processed. It is determined whether there is any correction point (step SM8-11).

  When it is determined that there is an unprocessed correction point (step SM8-11: YES), the nodule region extraction unit 22 proceeds to step SM8-1. When it is determined that there is no unprocessed correction point (step SM8-11: YES), the nodule region extraction unit 22 ends the protrusion correction process B.

  As shown in FIG. 23, when the projection correction process B is finished, the nodule region extraction unit 22 adds the region related to the distance less than the dilation distance Dmin and the small projection region after the projection correction processing B (the nodule after correction). (Mask area) is generated (step SM9).

  FIG. 28 is a diagram illustrating the nodule mask region AMR after correction generated in step SM9. As shown in FIG. 28, the protrusion of the corrected nodule mask region AMR is cut at the position of the nodule boundary point KP2. Further, a non-nodule region derived from a long and long blood vessel is removed at a position substantially the same as the nodule boundary point KP1.

  By applying the generated nodule mask region after correction to the nodule candidate region, the nodule region extraction unit 22 has a nucleus region equal to or more than the distance between the designated nodule boundary point and the nucleus region among the corrected projection candidate regions. The area away from the nodule area can be removed from the nodule area. Thus, by specifying the nodule boundary point, the non-nodule region derived from a thick blood vessel can be separated from the nodule region. That is, the user can cut the protrusion of the nodule region at an arbitrary position by specifying the nodule boundary point via the input unit 30.

(Processing area correction processing)
When the processing target area determined in step SB is smaller than the size of the nodule, the extracted nodule area does not fit in the processing target area. The processing area correction process is to determine whether or not the generated nodule area fits in the processing area after completion of the nodule area extraction process, and corrects the processing area if it does not fit.

  FIG. 29 is a diagram illustrating a flow of the process for correcting the processing target area, which is executed under the control of the control unit 10. As shown in FIG. 29, the control unit 10 controls each unit to perform a nodule region extraction process in a processing target region of a predetermined size (step SB1). The predetermined size is assumed to be 30 mm on a side.

  When the nodule region is generated, the control unit 10 causes the processing target region determining unit 14 to determine whether or not the size of the processing target region is equal to or larger than a preset upper limit value (for example, one side of 100 mm) (step SB2). .

  When it determines with it being below an upper limit (step SB2: NO), the control part 10 makes a process target area | region determine whether the nodule mask area | region has reached the end surface of the process target area (step SB3). Specifically, it is determined whether or not any of the plurality of voxels constituting the end face of the processing target area includes a voxel related to the foreground area (a voxel having a value “1”). If there is a voxel related to the foreground area among the voxels constituting the end face, the processing target area determination unit 14 determines that the nodule mask area has reached the end face of the processing target area. On the other hand, when there is no voxel related to the foreground area among the voxels constituting the end face, the processing target area determination unit 14 determines that the nodule mask area has not reached the end face of the processing target area.

  When it is determined that the nodule mask area has not reached the end face of the processing target area (step SB3: NO), the processing target area determination unit 14 determines whether the projection area shorter than the blood vessel determination distance Ds has reached the end face of the processing target area. It is determined whether or not (step SB4).

  When it is determined that a projection region shorter than the blood vessel determination distance Ds (for example, 20 mm) has reached the end surface of the processing target region (step SB4: YES), or it is determined that the nodule mask region has reached the end surface of the processing target region. In the case (step SB3: YES), the control unit 10 causes the processing target region determination unit 14 to enlarge the processing target region by a predetermined size (step SB5). The predetermined size is, for example, 10 mm. In this case, enlargement is performed by 5 mm in all directions of the six surfaces of the processing target area. Further, it may be enlarged by 5 mm only in the direction of the end face where the nodule mask region has reached. The predetermined size may be set in advance or may be set by the user via the input unit 30.

  When the processing target area is enlarged by a predetermined size, the control unit 10 controls each unit to cause the nodule area to be extracted again in the enlarged processing target area.

  On the other hand, when the size of the processing target region is 100 mm or more (step SB2: YES), or when it is determined that the projection region shorter than the blood vessel determination distance Ds (for example, 20 mm) has not reached the end surface of the processing target region (step SB4). : NO), the control unit 10 ends the processing area correction processing.

  If a large processing target area is set from the beginning (for example, 100 mm on a side), the problem that the nodule area does not fit in the processing target area does not occur. However, the process can be executed at high speed by performing the nodule region extraction process at the beginning with a small size as in this embodiment.

  In the present embodiment, a plurality of (generally) projection candidate regions of a nodule are obtained, a non-nodule region among the plurality of projection candidate regions is specified based on the dilation distance Dmax from the nucleus region, and the nodulation distance Dmax is determined. Since the non-nodule region is cut based on the short dilation distance Dmin, the small protrusion is not removed. Therefore, the erosion distance Dmin can be made longer than in the conventional method, a thicker blood vessel can be separated from the nodule region, and the small protrusions are not removed, resulting in less deformation of the nodule shape. Therefore, the extraction accuracy of the nodule region is improved, and the calculation accuracy of the volume value of the nodule region is improved as the extraction accuracy of the nodule region is improved.

  According to this embodiment, compared with the conventional method, a thick blood vessel can be separated in a short time, without cutting a small projection of a part of the nodule region, and without removing the small nodule region, Can be extracted. Thus, it is possible to provide an image processing apparatus capable of extracting a nodule region in a short time and with high accuracy.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. In addition, various inventions can be formed by appropriately combining a plurality of components disclosed in the embodiment. For example, some components may be removed from all the components shown in the embodiment. Furthermore, constituent elements over different embodiments may be appropriately combined.

1 is a diagram showing a configuration of an image processing apparatus according to an embodiment of the present invention. The figure which shows an example of the nodule area | region contained in the volume data memorize | stored in the memory | storage part of FIG. The figure which shows the flow of the extraction / display process of a nodule area | region performed by control of the control part of FIG. The figure which shows the nodule candidate area | region after a hole produced | generated in the hole-filling process (step SD) of FIG. 3, and the nucleus area | region produced | generated in the nucleus area | region determination process (step SE). The 1st figure which shows the outline | summary of the production | generation process (step SF) of the nodule mask area | region of FIG. The 2nd figure which shows the outline | summary of the production | generation process (step SF) of the nodule mask area | region of FIG. The figure which shows the flow of the hole-filling process performed by the hole-filling part in step SD of FIG. The figure which shows the flow of the nucleus area | region determination process performed by the nucleus area | region determination part in step SE of FIG. The figure which shows the flow of the nodule mask area | region production | generation process performed by the nodule area | region extraction part in step SF of FIG. The figure for demonstrating the process of step SF3 of FIG. The figure for demonstrating the process of step SF5 of FIG. The figure for demonstrating the process of step SF7 of FIG. The figure which shows the flow of the production | generation process of the nodule mask area | region in which the spiker determination process incorporated by the nodule area extraction part of FIG. The figure which shows an example of the screen for designating the correction point displayed on the display part of FIG. The figure which shows the flow of the nodule mask area | region production | generation process regarding the nodule inside point and the nodule outside point performed by the nodule area extraction part of FIG. The figure which shows the flow of the protrusion correction process A performed by the nodule area | region extraction part in step SK8 of FIG. The 1st figure for demonstrating the protrusion correction process A of FIG. The 2nd figure for demonstrating the protrusion correction process A of FIG. The figure which shows the nodule area | region which non-nodule area | regions, such as a thick long blood vessel and a bronchi, contained in the volume data memorize | stored in the memory | storage part of FIG. It is a figure which shows the flow of the determination process of a nucleus area | region performed when a nodule boundary point is designated by the nucleus area | region determination part of FIG. The 1st figure for demonstrating step SL5 and step SL6 of FIG. The 2nd figure for demonstrating step SL5 and step SL6 of FIG. The figure which shows the flow of the production | generation process of the nodule mask area | region regarding the nodule inside point, the nodule outside point, and the nodule boundary point performed by the nodule area extraction part of FIG. The figure which shows the positional relationship of the nodule candidate area | region after filling, the nucleus area | region after correction | amendment, and the dilation distance Dmin regarding step SM2 of FIG. The figure which shows the processus | protrusion candidate area | region after correction | amendment produced | generated in step SM3 of FIG. The figure which shows the small protrusion area | region after correction | amendment produced | generated in step SM7 of FIG. The figure which shows the flow of the small protrusion correction process B performed by the nodule area | region extraction part in step SM8 of FIG. The figure which shows the nodule mask area | region after correction | amendment produced | generated in step SM9 of FIG. The figure which shows the flow of the correction process of a process target area | region performed by control of the control part of FIG.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Image processing apparatus, 10 ... Control part, 12 ... Memory | storage part, 14 ... Processing object area | region determination part, 16 ... Nodule candidate area extraction part, 18 ... Filling part, 20 ... Nuclear area | region determination part, 22 ... Nodule area extraction part 24 ... Volume calculation unit, 26 ... Image generation unit, 28 ... Display unit, 30 ... Input unit.

Claims (11)

A storage unit for storing volume data relating to the subject;
A nodule candidate region extraction unit that extracts a nodule candidate region included in the volume data based on a pixel value;
A nuclear region determination unit for determining a nuclear region of the extracted nodule candidate region;
A non-nodule region specifying unit for specifying a non-nodule region from the nodule candidate region based on a first distance from the determined nucleus region and a second distance longer than the first distance;
A nodule region extraction unit that extracts a nodule region by removing the identified non-nodule region from the nodule candidate region;
An image processing apparatus comprising:
The non-nodule region specifying unit is connected to a region separated from the nucleus region by the second distance or more, and identifies a region separated from the nucleus region by the first distance or more as the non-nodule region,
An image processing apparatus. An image generator for generating image data based on the volume data;
A display unit for displaying the generated image by highlighting the extracted nodule region;
The image processing apparatus according to claim 1, further comprising:   The nucleus region determining unit contracts the nodule candidate region at a third distance shorter than or the same as the first distance, and determines the nodule candidate region subjected to the contraction processing as the nucleus region. The image processing apparatus described.   The image processing apparatus according to claim 3, wherein the nucleus region determining unit determines the third distance based on a size of the nodule candidate region. Further comprising a generation unit for generating the first nodule mask area is the sum region between a region within a distance of said from the nodule candidate region and the non-nodule regions projection region that is removed the nodule candidate region,
The nodule region extraction unit extracts the nodule region from the nodule candidate region by calculating a logical product of the generated nodule mask region and the nodule candidate region.
The image processing apparatus according to claim 1. The nodule candidate region extraction unit extracts a plurality of shadow candidate regions included in the nodule candidate region from the volume data based on a plurality of pixel values,
The nodule region extraction unit extracts a plurality of shadow regions from the plurality of shadow candidate regions as the nodule region by calculating a logical product of the plurality of extracted shadow candidate regions and the nodule mask region. ,
The image processing apparatus according to claim 5.   The image processing apparatus according to claim 1, further comprising a calculation unit that calculates a volume of the extracted nodule region. A first designating unit for designating a first point on the nodule candidate region;
The nodule region extraction unit includes, in the nodule region, a region connected to the designated first point among the projection candidate regions separated from the nucleus region by the first distance or more.
The image processing apparatus according to claim 1. A second designating unit for designating a second point on the nodule candidate region;
The nodule region extraction unit removes, from the nodule region, a region connected to the designated second point from among the projection candidate regions separated from the nucleus region by the first distance or more.
The image processing apparatus according to claim 1. A third designating unit for designating a third point on the nodule candidate region;
The nucleus region determining unit determines a region obtained by removing the predetermined region including the designated third point from the nucleus region as a new nucleus region;
The non-nodule region specifying unit includes a non-nodule region included in the nodule candidate region based on a first distance from the determined new nucleus region and a second distance longer than the first distance. Identify
The image processing apparatus according to claim 1. A fourth designating unit for designating a fourth point on the nodule candidate region;
The nodule region extraction unit is configured to select, from among the projection candidate regions separated from the nucleus region by the first distance or more, a region separated from the nucleus region by a distance between the nucleus region and the fourth point or more. Removing from the nodule region,
The image processing apparatus according to claim 1.

Images