JP2005198798A - Radiation image processing apparatus and processing method - Google Patents

Abstract

An imaging method for further imaging is determined based on a result of an image diagnosis support process to improve diagnostic accuracy.
Imaging processing is started by an imaging start signal (S201). When patient information and an imaging region are input, an imaging image size, an imaging location, an imaging distance, an X-ray aperture, an X-ray tube voltage, Imaging conditions such as tube current and X-ray irradiation time are set (S202). Shooting is started under this shooting condition (S203), and the shot image is displayed after being subjected to preset processing. This image is set as the first image, and analysis processing for diagnosis support is performed (S204).
If a region of interest suspected of being ill is output as a result of diagnosis support (S205), the imaging conditions for detailed imaging of the region of interest are determined (S206), and the states of the aperture means and the prone table are changed. Shooting is performed under the changed conditions (S207), a second image is generated, and shooting ends (S208). If the region of interest is not output (S205), the photographing ends there (S209).
[Selection] Figure 2

Description

  The present invention relates to a radiographic image processing apparatus and a processing method having an image diagnosis support means for analyzing an image captured in medical image capturing.

  When a certain type of phosphor is irradiated with radiation such as X-rays, α-rays, β-rays, γ-rays, electron beams, ultraviolet rays, etc., a part of this radiation energy is accumulated in the phosphor, and visible light etc. It is known that phosphors exhibit stimulated luminescence according to the stored energy when they are irradiated with excitation light, and phosphors exhibiting such properties are stimulable phosphors (stimulable phosphors) and being called.

  Using this stimulable phosphor, radiation image information of a subject such as a human body is once recorded on the sheet of the stimulable phosphor, and this stimulable phosphor sheet is scanned with excitation light such as a laser beam to emit light. Light is generated, and the obtained stimulated emission light is photoelectrically read to obtain an image signal. Based on the image signal, a radiographic image of the subject is displayed on a recording material such as a photographic photosensitive material or a display device such as a CRT. A radiation image information recording / reproducing system for outputting a visible image has already been proposed by the present applicant.

  In recent years, apparatuses for taking X-ray images in the same manner using semiconductor sensors have been developed. These systems have the practical advantage of being able to record images over a very wide radiation exposure range compared to conventional radiographic systems using silver salt photography. That is, for example, as described in Patent Document 1, X-rays with a very wide dynamic range are read by a photoelectric conversion means and converted into an electric signal, and a recording material such as a photographic photosensitive material, a CRT, or the like using this electric signal. The radiation image can be output as a visible image to the display device, and a radiation image that is not affected by variations in radiation exposure can be obtained.

  By the way, in recent years, development of CAD (diagnosis support technology) by a computer has been advanced for an image acquired by such a semiconductor sensor. As a conventional related art, for example, a system is known in which abnormal shadows are extracted by CAD from a reconstructed tomographic image as in Patent Document 2 and a corresponding slice range is reconstructed with a detailed slice pitch.

  Further, for example, as disclosed in Patent Document 3, there is an interpretation support means for extracting an interpretation attention location from a reconstructed tomographic image by CAD and displaying a slice position marker on a slice position image. In addition, a diagnosis support system that aligns the slice positions of the current image and the past image is known. Furthermore, there is Patent Document 4 as related prior art documents.

Japanese Patent Laid-Open No. 9-32267 Japanese Patent Application Laid-Open No. 8-16695 JP 2001-87228 A JP 2001-137230 A

  In clinical imaging of medical images, fine images are not captured from the beginning. As an example of a mammogram, in Japan mammogram screening, one life size MLO view is taken for both breasts.

  However, since only about 0.1% of women who undergo medical examinations have signs of breast cancer, CC and MLO views are not taken for both breasts from the beginning. The captured MLO view image is recalled for a patient who is suspected of having breast cancer based on a doctor's diagnosis, and is re-photographed. Shooting at this point may be both CC view and MLO view shooting, or enlarged shooting of the region of interest.

  The problems with this process are: (1) it takes days to recall and treatment is delayed for fast-growing cancer, (2) the patient needs to revisit the medical institution and the patient's time burden There is big.

  The same problem occurs in general X-ray imaging, CT imaging, and MRI imaging. In other words, at first, the imaging conditions are set so that the imaging time is short so that the exposure dose can be reduced, and initial diagnosis is performed for these relatively coarse images, and the suspicious area, that is, the area of interest If present, a highly accurate image is taken.

  It is reasonable to take a coarse image first and to capture the region of interest with high accuracy, but the diagnosis of the coarse image is not performed immediately after the image, and the patient leaves the room and returns home. Therefore, when it is diagnosed that high-accuracy re-imaging is necessary, there are problems that it takes time until re-imaging as described above, and that it is troublesome for the patient to come again.

  An object of the present invention is to provide a radiographic image processing apparatus and a processing method that solve the above-described problems and that can immediately perform diagnostic support for a captured image and perform re-imaging.

  In order to achieve the above object, a radiological image processing apparatus according to the present invention includes an analysis unit that analyzes information on the position and range of a region of interest from a radiographic image, and determines imaging conditions based on the information analyzed by the analysis unit. And a condition determining means.

  The radiological image processing method according to the present invention includes an analysis step of analyzing information on the position and range of a region of interest from a radiographic image, and a condition determination step of determining imaging conditions based on the information analyzed in the analysis step. It is characterized by having.

  According to the radiographic image processing apparatus and processing method of the present invention, it is possible to immediately determine whether or not precise additional imaging is necessary, and to automatically set the necessary imaging conditions. .

  For example, in a mass examination of chest images and breast images, even if it is determined to be suspicious by a doctor's diagnosis, it is necessary for the patient to go to the hospital again until detailed imaging is taken. become. In addition, in conventional CT devices and the like, an engineer may judge whether or not detailed imaging is necessary, but there is a risk of oversight or misdiagnosis of non-illness, but there is a need for re-imaging. It is possible to decide, and the photographing conditions are automatically decided.

Hereinafter, the present invention will be described in detail based on illustrated embodiments.
FIG. 1 shows an X-ray imaging apparatus according to an embodiment. The X-ray imaging apparatus 1 is an X-ray imaging apparatus having a CAD function, and an X-ray generation means 3 is located above a subject P on a prone table 2. The X-ray beam is irradiated toward the subject P via the aperture means 4. In addition, a two-dimensional detector 5 is disposed below the lying table 2.

  Various means are connected to the X-ray generation means 3 and the two-dimensional detector 5 through a CPU bus 6. That is, the CPU bus 6 has an image storage means 11 for storing a photographed image comprising a control means 7, an image processing means 8, an image display means 9, a main memory 10, a computer memory, and the like. An analysis / diagnostic means 12 for analyzing the morphological pathology, an imaging method changing means 13 for changing and storing imaging parameters such as X-ray conditions, and an operator input means 14 are connected. The control means 7, the image processing means 8, the image display means 9, the image storage means 11, the analysis diagnosis means 12, and the imaging method change means 13 are constituted by a computer. Further, an aperture change motor 15 and a table position change motor 16 are connected via the CPU bus 6 to drive the aperture means 4 and the position table 2 respectively.

  The control means 7 controls the operation of the entire apparatus according to the operation from the operator input means 14. First, under the control of the control means 7, the X-ray generation means 3 emits an X-ray beam to the subject P. The X-ray beam passes through the subject P while being attenuated and reaches the two-dimensional detector 5. It is output as an image by the dimension detector 5. At this time, the aperture means 4 can change the shape of the X-ray beam emitted from the X-ray generation means 3, and the supine table 2 relatively changes the positional relationship between the X-ray beam and the subject P.

  Specifically, based on the control from the imaging method changing unit 13, the diaphragm unit 4 is changed by driving the aperture changing motor 15, and the position table 2 is changed by driving the table position changing motor 16. The image collected by the image processing means 8 via the two-dimensional detector 5 is subjected to image processing such as predetermined correction processing and gradation conversion processing and displayed on the image display means 9.

  FIG. 2 is a flowchart showing the operation of the X-ray imaging apparatus 1. In the X-ray imaging apparatus 1 as described above, first, the main memory 10 stores various data necessary for processing by the control means 7 and also includes a work memory for working the control means 7. Yes. The program code according to this flowchart is stored in the main memory 10 or a ROM (not shown), and is read out and executed by the CPU bus 6.

  For example, when an X-ray chest front image of the subject P is taken as an example of the imaging target, the imaging process is started by an imaging start signal input from the operator input means 14 (S201). When patient information such as a patient name, sex, and age is input from the operator input unit 14 and an imaging region is input, the imaging method changing unit 13 refers to a table (not shown), and the imaging image size, imaging location, imaging distance, Imaging conditions such as the X-ray aperture, X-ray tube voltage, tube current, and X-ray irradiation time are determined and stored (S202).

  Next, the imaging conditions determined and stored by the imaging method changing means 13 are called into the control means 7, and the control means 7 performs imaging according to the X-ray irradiation start signal from the operator input means 14 according to the called imaging conditions. The process is started (S203). An image photographed by the two-dimensional detector 5 is input to the image processing means 8, subjected to preset processing, and displayed on the image display means 9. The image obtained by the first photographing is set as a first image, and the first image is simultaneously input to the analysis / diagnostic means 12 to perform analysis processing for diagnosis support (S204).

  If it is determined whether or not re-imaging is required before the patient leaves the imaging room, and it is determined that re-imaging is necessary, the imaging conditions for re-imaging are set in the apparatus and the radiographer (operator) is re-established. Notification that shooting is necessary is performed.

  As for re-imaging, of course, automatic imaging is possible. However, for devices using X-rays, since the start of X-ray irradiation may be limited to the imaging technician, only notification of re-imaging is performed.

  If a region of interest that is a suspicious part of the disease is output as a result of the diagnosis support by the analysis diagnostic unit 12 (S205), the region of interest is transferred to the imaging method changing unit 13, and the imaging conditions for detailed imaging are set. The shooting conditions are determined and written in the shooting method changing means 13 (S206). The states of the aperture means 4 and the position table 2 are changed according to the written photographing condition. By the X-ray irradiation start operation from the operator input means 14, imaging is performed under the changed conditions (S207), a second image is generated, and imaging is terminated (S208). If the region of interest is not output (S205), shooting is terminated (S209).

  The analysis / diagnostic means 12 has different means depending on the purpose of diagnosis support and the object to be imaged, and uses the processing in each means properly. For example, when the subject to be imaged is a chest image, the shadow extraction means 21 shown in FIG. 3 and the processing of the texture disease extraction means described later are used properly. The shadow extraction means 21 is used to detect calcification in a chest image, a tumor in the form of a tumor, or a tumor (MASS) in a mammogram image, and the texture disease extraction means 21 is an interstitial pneumonia found in the chest image. It is used for area extraction processing.

  FIG. 3 is a block circuit configuration diagram for executing the shadow extraction processing. The shadow extraction means 21 for calculating a specific frequency band image and extracting circular shadow candidates is a high-pass filter means 21a and a low-pass filter means for the input image. 21b, and these outputs are obtained as a rectangle of a certain region surrounding a positive shadow candidate through a pathological feature amount extraction unit 22 that calculates a feature amount and a determination unit 23 that determines whether the extracted shadow is positive or negative. It is connected to the positive region determining means 24 to be determined.

  FIG. 4 is a flowchart of this shadow extraction process. First, the first image is input to the shadow extraction unit 21 (S401), and a high-frequency image and a low-frequency image are created from the input image by the high-pass filter unit 21a and the low-pass filter unit 21b, respectively (S402). Here, a shadow candidate image is created from the difference image of the two images.

  FIGS. 5A and 5B show characteristic diagrams of the high-pass filter unit 21a and the low-pass filter unit 21b, respectively. FIG. 5C shows a characteristic diagram of a matched filter that detects a circular pattern of a certain size, which is a difference between a high-frequency image and a low-frequency image obtained by the high-pass filter unit 21a and the low-pass filter unit 21b. Yes. The circular pattern has a specific frequency band, and the size of the circular pattern to be extracted can be adjusted by adjusting the characteristics of the high-pass filter unit 21a and the low-pass filter unit 21b.

  In general, in the case of a chest front image, it is extremely difficult to select and extract a tumor having a diameter of 5 mm or less from the image. The reason is that the image contains many signals of the same size, and many of them are not sick signals. Therefore, in this embodiment, a matched filter corresponding to a circular pattern having a diameter of 10 mm is used.

  However, the diameter of the circular pattern to be extracted by the matched filter is not limited to one type, but a plurality of types of filters such as 8 mm, 12 mm, and 16 mm are sequentially applied, and the shadow candidates extracted by the respective filters are subsequently used. It is possible to flow through the process.

  The feature quantity is calculated by the pathological feature quantity extraction means 22 for one or a plurality of shadow candidates extracted by the shadow extraction means 21 (S403). The feature quantities calculated here are area, circularity, and threshold sensitivity. However, although three types of feature quantities are used in the present embodiment, it is not limited to three types.

When calculating the area S and the circularity C, the shadow candidate region and the other region are binarized and distinguished. The binarization threshold value is, for example, the upper 10% value of the histogram of the matched filter output image.
Area S = number of pixels included in shadow candidate / area of one pixel

  Circularity C = A / S, where A represents an area where the circle and the shadow candidate overlap, that is, overlap when the circle having the effective diameter D is arranged at the center of gravity of the shadow candidate.

  Threshold sensitivity = | 2 · S (10) −S (5) −S (15) | is a change in the shaded area when the threshold is changed. Here, S (5), S (10), S (15) represents the area when the threshold is changed to 5%, 10%, and 15%, and || represents the absolute value.

  Next, on the basis of the feature amount calculated by the pathological feature amount extraction unit 22, a determination as to whether each shadow candidate is positive or false positive is input to the neural network for determination by the determination unit 23 (S404). ). A false positive is a shadow extracted because it is pathologically normal but has a circular pattern. The determination means 23 is constituted by a neural network, and the neural network has a learning phase and a use phase, and therefore will be described separately for both phases. However, the determination means 23 may use any means as long as pathological determination can be performed according to a certain rule. For example, when separation is possible depending on the purpose of support, linear separation may be used. Good.

  For a shadow whose positive or false positive result is known in the learning phase, while inputting the feature quantity calculated by the pathological feature quantity extraction means 22 while presenting the result to the neural network, the internal coefficient of the neural network is inputted. To learn. Various neural networks have been developed. For example, a feedforward type error back propagation neural network developed by Rammel Heart of Non-Patent Document 1 and a Radial Basis Function neural network of Non-Patent Document 2 (simply RBF-NN).

D.E. Rumelhart and J.L. McCelland, Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Vol. 1: Foundation. Cambridge: The MIT Press, 1986. C. Bishop, "Improving the Generalization Properties of Radial Basis Function Neural Networks," Neural Comp., Vol. 3, pp. 579-588, 1991.

  In this embodiment, a 3-input RBF-NN as shown in FIG. 6 is used. In terms of structure, a feedforward type is adopted, which has a three-layer structure of an input layer, an intermediate layer, and an output layer. In the input layer, input nodes corresponding to the number of feature amounts extracted by the pathological feature amount extraction unit 22 are arranged. However, for convenience, the number of input NODEs is 3, the number of intermediate neurons is 4, and the number of output NODEs is 1. The RBF neurons arranged in the intermediate layer have an output characteristic that has a Gaussian distribution as a non-linear element. The number of RBF neurons depends on the number of learning cases and the complexity of the problem, but about 100 is appropriate for setting the calculation time appropriately. The number of outputs of the neural network is 1. If it is positive, 1 is output, and if it is negative, 0 is output.

  In this embodiment, the neural network is prepared for each purpose of diagnosis support. However, since it is configured by software, the coefficient for each diagnosis support is actually stored, and the coefficient for each diagnosis support is used in use. set.

  In the use phase, after setting the internal coefficient learned in the learning phase in the neural network, the feature amount calculated by the pathological feature amount extraction means 22 from the image taken by the two-dimensional detector 5 is input to this neural network, Get the output of the neural network. For example, when the positive output is learned as 1 and the negative output is 0, the determination is positive if the output of the neural network is 0.5 or more, and negative if it is 0.5 or less (S405).

  When a plurality of circular patterns (shadows) determined to be positive are close to each other, a single positive area is formed, and the positive area is rectangularized (S406). However, although it is not necessary to limit the designation of the positive region to a rectangular shape, in general, a rectangular shape is often used for the aperture for X-ray irradiation. If the aperture for X-ray irradiation is circular, it is desirable to make it circular.

  When there are a plurality of shadow candidates extracted from the first image, the rectangular area may not be limited to one place. In this case, the number of detailed images to be described later may be plural (S407).

  FIG. 7 shows the input chest front image, and FIG. 8 shows the result of applying the shadow extraction means to this image. FIG. 9 shows the result of extracting pathological feature amounts for each of the four shadow candidates in FIG. 8 and determining them by the determination means 23 using a neural network based on the feature amounts. Here, the rectangular shadow pattern of the right lung is excluded as negative. FIG. 10 shows the result of determining the positive area by the positive area determining means 24 for this positive shadow pattern. The positive region is set so as to include all positive shadows, and the positive region is set by integrating the shadows of the left lung into one.

  FIG. 11 is a block circuit configuration diagram for executing the texture disease diagnosis support process described above. The texture disease extraction means 31 includes a segmentation means 32 and a region division means 33, and the segmentation means 32 further includes a segment feature quantity extraction means 32a and a discrimination means (neural network means) 32b. The output of the texture disease extraction means 31 is connected to the positive region determination means 24 via the pathological feature quantity extraction means 22 and the determination means 23 as in FIG.

  Here, exemplifying an interstitial lung disease with respect to the chest front image, the texture disease extraction unit 31 extracts a region candidate having a texture disease as a region of interest (ROI) for the first image. First, the first image is input to the segmentation means 32 and the lung field region is segmented. Since interstitial lung disease is a disease determined in a local region, the lung region is extracted to define the local region of the lung field.

  The segmentation means 32 is constituted by a neural network, and has a learning phase and a use phase as described above. In the following phases, segmentation is performed in units of pixels. However, segmentation can also be performed by a method of tracking an outline from an image shown in Non-Patent Document 3, for example.

O. Tsujii, M. T. Freedman, and S. K. Mun, "Lung contour detection in chest radiographs using 1-D convolution neural networks," Electronic Imaging, 8 (1), pp. 46-53, January 1999

  For each pixel of the first image, the segment feature quantity extraction unit 32a calculates a feature quantity. The calculated feature amount includes, for example, a calculation based on a pixel value disclosed in Non-Patent Document 4, a calculation based on a texture, and a calculation based on a relative address from an anatomical structure. . These feature amounts indicate the texture characteristics of the image. Note that the feature amount is not limited to these, and any feature amount may be used as long as it indicates the texture feature of the image. Here, as shown in FIG. 12, the first image is divided into a grid-like rectangular ROI, and a feature amount is calculated for each ROI.

O. Tsujii, MT Freedman, and SK Mun, "Automated Segmentation of Anatomic Regions in Chest Radiographs using Adaptive-Sized Hybrid Neural Network," Med. Phys., 25 (6), pp. 998-1007, June 1998.MF McNitt -Gray, HK Huang and JW Sayre, "Feature Selection in the Pattern Classification Problem of Digital Chest Radiograph Segmentation," IEEE Trans. Med.Imag. 14, 537-547 (1995)

  Various neural networks have been developed as described above. In this embodiment, RBF-NN is used. In terms of structure, a feed-forward type is adopted, and a three-layer structure of an input layer, an intermediate layer, and an output layer as shown in FIG.

  In the input layer, input Nodes corresponding to the number of feature amounts extracted by the segment feature amount extraction unit 32a are arranged. The RBF neurons arranged in the intermediate layer have output characteristics that have a Gaussian distribution as a non-linear element. The number of RBF neurons depends on the number of learning cases and the complexity of the problem, but about 5000 is appropriate for setting the calculation time appropriately. The number of outputs of the neural network corresponds to the number of segments for each part.

  For example, in the case of an image of the front of the chest as shown in FIG. 7, since there are two anatomical segments S1 and S2 as shown in FIG. 13, two output NODEs are prepared. In this embodiment, the neural network is prepared for each imaging region, that is, for each target image. However, since the neural network is configured by software, the coefficients for each region are actually stored and used for each region. Set the coefficient of.

  In the learning phase, an answer example as a segmentation result is presented to the determination unit 32b together with the feature amount for each pixel. FIG. 13 shows an example of a teacher output image with respect to an input image. The teacher image is created so that the teacher image has output values 1 and 2 so as to correspond to the segment classification corresponding to the chest front image. However, the number of areas other than the segment, for example, zero is set. The teacher image means can be implemented by graphic software that can designate an area in the image, but commercially available image processing software can also be used. In the case of RBF-NN, the learning of the neural network can be obtained analytically by performing the least square method that minimizes the output error, and the calculation of the internal coefficient can be performed in a short time.

  Extraction of feature amounts for each pixel in the image relating to the use phase is performed in the same manner as in the learning phase. As the internal coefficient for the neural network, the coefficient for the target imaging region is loaded before use. The feature quantity for each pixel is presented to the neural network, and the network outputs an output value to the output NODE. In the case of the chest front image, there are two output NODEs, and the input pixels are classified into segments corresponding to the NODEs that output the closest to YES among these output NODEs. FIG. 14 shows segment images S3 and S4 in which the classification for all the pixels in the image is completed in this way.

  In general, the segment image output from the discriminating means 32b contains noise, and for example, a lung field region having a small area may be generated by separating from the lung field region. Such a small noise area is removed in the post-processing process, but this technique is described in Non-Patent Document 3 described above.

  FIG. 15 shows an example of a segmentation output image of the discrimination means 32b. Region segmentation means 33 is applied to this segmentation output image. Region segmentation is to define a local region for calculating texture disease parameters. Specifically, as shown in FIG. 12, the target region which is the lung field in this case of the target segmentation output image is to be divided into rectangles so as to have an area of a certain value or less. The roughly rectangular region of interest defined in this way is called a texture disease candidate ROI.

  For each of the plurality of texture disease candidate ROIs output by the texture disease extraction unit 31, pathological feature amounts are calculated by the pathological feature amount extraction unit 22, and determination is made based on the feature amounts as positive or negative. . The process after the feature amount calculation is the same as the shadow diagnosis support process shown in FIG. However, the feature amount calculation is calculated according to the purpose of the diagnosis support process. The following non-patent documents 5 and 6 can be referred to for the feature values used for texture disease diagnosis support.

Shigehiko Katsurakawa et al .: Possibility of computer diagnosis support for interstitial diseases, Journal of Japanese Society of Medical Radiology, 50: 753-766, 1990. Yasuo Sasaki et al .: Quantitative evaluation by texture analysis of pneumoconiosis standard photographs, Japanese Society of Medical Radiology, 52: 1385-1393, 1992.

  The region of interest is output by the analysis / diagnosis unit 12, and this output is input to the imaging method changing unit 13 and used for determining the imaging condition of the second image. The photographing method changing means 13 is displayed on the operator input means 14. In the case of an X-ray apparatus, the start of imaging is limited to the photographer, but in the case of CT or MRI, the condition setting and the position table 2 are automatically moved, and the imaging starts automatically. It is also possible.

  FIGS. 16A and 16B show the photographing positions of general supine chest photographing of the subject P. FIG. (A) shows the state seen from the direction of the head, and (b) shows the state seen from the side. The conditions of X-ray imaging by the X-ray generation means 3 are tube voltage 120 KV, tube current 200 mA, imaging time is about 10 msec, imaging distance is about 1.8 m from the X-ray focal point to the patient, and the imaging size is 43 cm / 43 cm. is there. The irradiation time is automatically determined when an X-ray exposure monitor (not shown) is used.

  In this state, when the first chest image of the subject P is photographed, and the region of interest Q as shown in FIG. Since the chest image of 1 is lying on the lying position table 2, X of the lying position table 2 so that the region of interest Q indicated by a rectangle in FIG. The movement in the Y direction and the elevation table 2 are moved up and down to set an appropriate enlargement ratio.

  FIGS. 16D and 16E show a state in which the position table 2 has been moved in order to capture the second image. (D) and (e) correspond to the states (a) and (b), respectively, and the position of the supine table 2 is moved so that the lower part of the left lung is enlarged.

  Note that it is not necessary to strictly enlarge the region of interest Q detected in FIG. 16 (c), but the X-ray beam stop means 4 and the X-rays are only applied to the necessary portions when the imaging size is changed. The beam is being irradiated. In this case, since the imaging distance to the two-dimensional detector 5 is not changed, the tube voltage and tube current of the X-ray generation means 5 are maintained as they are. In addition, although the internal structure of the X-ray exposure monitor is not strictly illustrated, which of the X-ray light receiving parts formed of a plurality of segments corresponding to the imaging region may be used to block the X-rays. An appropriate one is automatically used.

  Further, since the imaging time until the X-ray beam is interrupted can be acquired from the X-ray control device when the first image is captured, X-ray exposure can be performed by setting the irradiation time in the imaging control unit. It is also possible to take a picture by specifying the irradiation time without using a monitor for the camera.

  FIG. 17 is an explanatory diagram of a process for calculating the movement amount of the position table 2 calculated inside the photographing method changing means 13. If the relationship between the focal point of the X-ray beam and the two-dimensional detector 5 is fixed, the distance from the center of the first image to the center of the region of interest Q is the amount of Y in the body axis direction of the subject P, and the horizontal axis method. Calculated with X quantity. Specifically, it can be obtained by calculating the distance by the number of pixels and multiplying by the pixel size.

  Ascending / descending of the saddle position table 2 is usually doubled in the case of magnified photographing, so that the lateral position table is placed between the focal point of the X-ray beam and the two-dimensional detector 5 so as to obtain a double geometric distance. The position of 2 is determined. Then, the size of the aperture means 4 is calculated in consideration of image enlargement. Of course, the enlargement size is not limited to twice, and other multiples may be determined by referring to the table according to the type of disease. For example, it is conceivable that the enlargement rate is doubled for cancer and is set to triple for interstitial pneumonia in order to observe the symptoms in more detail.

  Although an example of changing the captured image size, imaging location, imaging distance, and X-ray aperture opening by the imaging method changing unit 13 has been described, imaging conditions such as the X-ray tube voltage, tube current, and X-ray irradiation time of the X-ray generating unit 3 May be changed. In general, when the tube voltage is lowered to, for example, 110 KV, it is suitable for observing soft tissue, and an example of a disease as soft tissue is interstitial pneumonia. When calcification is suspected, the contrast of other organs can be reduced by, for example, imaging with the tube voltage increased to 140 KV.

  As a factor for changing the tube current, in general, when a soft tissue such as interstitial pneumonia is observed, it is necessary to improve the SN ratio. The X-ray irradiation time is determined in relation to the movement of the body during imaging and the movement of an organ such as the heart. For example, in the case where the region suspected of being positive is close to the heart, In order to reduce this, a high-speed shooting time of 1.25 msec, which is half of the normal 2.5 msec, is preferable. However, if the imaging time is halved to obtain the same dose, it is necessary to perform imaging with double tube current.

  In the above description, the X-ray chest imaging apparatus is taken as an example, but the X-ray breast image can be similarly subjected to enlarged imaging using the diagnosis support result. Further, in the above example, both the first image and the second image are still images, but based on the diagnosis support result in the first still image, a moving image of the region of interest is captured as the second image. It is also possible to capture the detailed image of the region of interest as the second image based on the result of the diagnosis support for the first moving image.

  FIG. 18 is an explanatory diagram taking the case of CT as an example. (A) represents the place where the first image is shot and the shooting conditions. Imaging is performed on a wide area from the lungs to the abdomen. For this reason, the photographing is performed with a large slice thickness of 10 mm and a coarse photographing interval of 10 mm. Diagnosis support means is applied to the plurality of first images, and as a result, as shown in (b), what appears to be a tumor is detected at slice numbers 3 and 4.

  In this state, since it is difficult to judge malignancy and benignity and the degree of progression, the second image is taken with the slice thickness set to 5 mm and the slice pitch set to 5 mm. In this case, the change in the photographing condition is that the position table 2 is moved so that photographing is performed around the positions 3 and 4 of the first image as shown in FIG. The slice thickness is changed and the slice pitch is changed, and a change such as an increase in tube current is performed in accordance with the change of the slice thickness. The change of the slice thickness is realized by changing the interval between the slits installed on the front surface of the X-ray tube. In this example, CT is used, but this technique can be applied to MRI as well.

  In the above-described embodiment, the imaging devices for the first image and the second image are the same. However, the imaging devices need not be the same, and the X-ray image diagnosis support result is used, and the CT scanner performs fine imaging. It is also possible to determine imaging conditions such as the position of the prone position table, slice thickness, and X-ray conditions, and set the imaging conditions to CT.

  In addition, the analysis means, the imaging method, and the change of the imaging conditions can also be applied outside the medical field. That is, in capturing a natural image or an analysis image, it may be necessary to analyze an automatically captured image and capture an important region of interest in more detail.

  Further, the present invention is not limited to only the apparatus and method for realizing the above-described embodiment, and the program code of software for realizing the embodiment is supplied to the CPU or MPU computer in the apparatus. The case where the above-described embodiment is realized by causing the computer of the system or apparatus to operate various devices according to the program code is also included in the scope of the present invention.

  In this case, the program code of the software itself realizes the functions of the embodiment, and the program code itself and means for supplying the program code to the computer, specifically, a memory storing the program code The medium is included in the scope of the present invention.

  As a storage medium for storing such a program code, for example, a floppy disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

  In addition to the case where the functions of the embodiments are realized by the computer controlling various devices according to only the supplied program code, an OS (operating system) in which the program code is running on the computer, or other Such a program code is also included in the scope of the present invention when the embodiment is executed in cooperation with the application software.

  Further, after the supplied program code is stored in the memory of the function expansion board of the computer or the function expansion unit connected to the computer, the program code is stored in the function expansion board or function storage unit based on the instruction of the program code. A case where the CPU or the like provided performs part or all of the actual processing and the embodiment is realized by the processing is also included in the scope of the present invention.

It is a block diagram of the X-ray imaging apparatus of an Example. It is a flowchart figure of operation | movement. It is a block circuit block diagram of a shadow extraction process. It is a flowchart figure of shadow extraction. It is a characteristic view of the difference of a low pass, a high pass filter, and both filters. It is explanatory drawing of a discrimination means. It is explanatory drawing of a chest front image. It is explanatory drawing of a shadow candidate image. It is explanatory drawing of a determination means output. It is explanatory drawing of a positive area | region. It is a block circuit block diagram of a texture disease diagnosis assistance process. It is explanatory drawing divided | segmented into an image and a rectangular ROI in texture feature-value extraction. It is explanatory drawing of the segmentation result of a lung field. It is explanatory drawing of the segmentation result of a lung field. It is explanatory drawing of a segmentation output. It is explanatory drawing of imaging | photography method change. It is explanatory drawing of the calculation process of imaging | photography method change. It is explanatory drawing in the case of imaging | photography a 2nd image based on a 1st image.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 X-ray imaging apparatus 2 Position table 3 X-ray generation means 4 Aperture means 5 Two-dimensional detector 7 Control means 8 Image processing means 9 Image display means 10 Main memory 12 Analysis diagnostic means 13 Imaging method change means 14 Operator input means 21 shadow extraction means 21a high-pass filter means 21b low-pass filter means 22 pathological feature quantity extraction means 23 determination means 24 positive area determination means 31 texture disease extraction means 32 segmentation means 32a segment feature quantity extraction means 32b discrimination means 33 area division means Q region of interest

Claims (9)

  A radiographic image processing apparatus comprising: an analyzing unit that analyzes information on a position and a range of a region of interest from a radiographic image; and a condition determining unit that determines an imaging condition based on the information analyzed by the analyzing unit.   2. The radiographic image according to claim 1, wherein the condition determining unit determines at least one of a position of a radiation generating unit, a position of a table on which a subject is loaded, a range of an irradiation aperture, and a position of a sensor. Processing equipment.   The condition determining means includes a position of a radiation generating means, a position of a table for loading a subject, a range of an irradiation aperture, and a sensor position so that only the region of interest calculated by the analyzing means is irradiated with X-rays. The radiographic image processing apparatus according to claim 1, wherein at least one is determined.   The condition determining means includes at least one of the position of the radiation generating means, the position of the table for loading the subject, the range of the irradiation aperture, and the position of the sensor so that the region of interest calculated by the analyzing means is magnified. The radiation image processing apparatus according to claim 1, wherein one is calculated.   The condition determining means includes a position of a radiation generating means, a position of a table for loading a subject, a range of an irradiation aperture, a position of a sensor, a position of a sensor, so that the region of interest calculated by the analyzing means is photographed. The radiographic image processing apparatus according to claim 1, wherein at least one of the driving methods is determined.   The radiographic image is a CT or MR reconstructed image, and the condition determining means determines an imaging range and / or imaging pitch based on the position and range of the region of interest calculated by the analyzing means. The radiographic image processing apparatus according to claim 1.   A radiographic image processing method comprising: an analysis step of analyzing information on a position and a range of a region of interest from a radiographic image; and a condition determination step of determining an imaging condition based on the information analyzed in the analysis step.   A program for executing the radiation image processing method of claim 7 on a computer.   A storage medium storing a program for executing the radiation image processing method according to claim 7 on a computer.