JP2008167949A - Radiographic image processing method and apparatus, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve the quality of a radiation image indicating an object without increasing the dose of radiation to be emitted to the object in a radiographic image processing method. <P>SOLUTION: The radiation image 11 for input prepared by using both of a high voltage image and a low voltage image obtained by radiography 10 is prepared for each of the plurality of objects 1P of the same kind, also the radiation image 33 for a teacher whose image quality degradation is less than that of the radiation image 11 for the input, and which emphatically indicates a specified part Px in the object 1P is prepared, and a teacher learning completed filter 40 for which learning is performed with the radiation image 11 for the input as an object and the radiation image 33 for the teacher as the teacher is obtained. Thereafter, the radiation image 21 of the same kind as the radiation image 11 for the input is prepared for a given object 3P, the radiation image 21 is inputted to the teacher learning completed filter 40, and the radiation image 60 whose image quality degradation is compensated and which indicates the object 3P whose specified part Px is emphasized is formed. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a radiographic image processing method, apparatus, and program for obtaining a radiographic image that emphasizes and expresses a specific part in a subject.

  Conventionally, in medical radiography and the like, a high-pressure image and a low-pressure image are obtained by radiography of a subject using radiation having different energy distributions, and the subject is identified by weighted subtraction processing between the high-pressure image and the low-pressure image. There is known a technique for obtaining an energy subtraction image that emphasizes a part exhibiting the radiation absorption characteristics of, for example, a bone part or a soft part of a living tissue (see Patent Document 1). This energy subtraction image is an image formed based on the difference between the high pressure image and the low pressure image.

  In radiography for obtaining the high-voltage image and the low-voltage image, for example, two types of radiations having different energy distributions generated with different tube voltages of the radiation source are applied to the subject twice in total at different timings. The subject is placed on two stimulable phosphor sheets arranged with a copper plate between them, by two-shot radiography for irradiating the high-pressure image and the low-pressure image, or by one-time irradiation of the subject. A one-shot radiography method for simultaneously recording a high-pressure image and a low-pressure image is known.

  The energy subtraction image formed by using the high-pressure image and the low-pressure image is more specific than the radiographic image (hereinafter also referred to as simple radiographic image) obtained by the usual radiographic method (hereinafter referred to as simple radiographic method). Although it is excellent in that it can emphasize this part, the image contains more noise. In the simple radiography method, a radiographic image of the subject is acquired by radiography in which one type of radiation is irradiated to the subject once without using a plurality of types of radiation having different energy distributions.

  Noise generated in the energy subtraction image is mainly caused by a shortage of radiation dose to be irradiated when the high pressure image or the low pressure image is acquired.

  That is, in medical radiography or the like, it is desired to reduce the burden on patients by reducing the dose of radiation used for radiography. For example, in radiography (two-shot radiography) that requires two times of normal radiography, a sufficient dose is not used in any one of the above two radiographs, or radiation is absorbed by a copper plate. An energy subtraction image created using a radiographic image (high-pressure image or low-pressure image) obtained by radiography (one-shot radiography) with attenuated dose is a simple radiographic image obtained by the above-mentioned simple radiography. The image quality is worse than the image.

  In both the one-shot radiography method and the two-shot radiography method, it is demanded to reduce the dose of radiation applied to the subject in the radiography. As described above, the subject is irradiated in the radiography. If an attempt is made to reduce the radiation dose, the amount of noise generated in the radiographic image obtained by this radiography increases and the image quality deteriorates.

  On the other hand, a radiological image that estimates a component representing a bone portion of a subject from one radiographic image obtained by simple radiography and emphasizes the bone portion in the subject without performing weighted subtraction processing of the plurality of radiographic images. There are known methods for forming. This technique is intended to obtain an image similar to the bone image that is the energy subtraction image without performing radiation imaging using radiation having different energy distributions.

  More specifically, this method is intended to obtain an image similar to the bone part image by the following procedure.

That is, a radiographic image for a teacher is formed in advance by emphasizing a bone portion obtained by radiography of a human chest that is a subject. When a learning simple radiographic image representing a human chest that is the same type of subject as the subject is input, a radiographic image using the teacher radiographic image as a teacher, that is, a radiographic image in which a bone portion is emphasized is output. As described above, learning is repeated to create a teacher-learned filter (a filter using an artificial neural network (ANN)). After that, a simple radiation image of the human chest that is the same type of diagnosis subject as the subject is input to the teacher-learned filter to obtain a diagnostic radiation image of the human chest with the bones emphasized. Yes (see Patent Document 2).
JP-A-3-285475 US Patent Publication No. 2005 / 0100208A1

  However, the method using the supervised learned filter is not reliable enough to estimate the bone part in the subject, and the image component representing the soft part in the subject in the image representing the emphasized bone part. May appear as a fake image.

  That is, if an attempt is made to create a radiographic image that emphasizes a specific part in a subject as described above, a false image may be generated in the radiographic image and the image quality may deteriorate. For this reason, there is a demand for a radiographic image that suppresses image quality degradation due to the occurrence of noise, fake images, and the like, and that emphasizes and expresses a specific part in a subject.

  The present invention has been made in view of the above circumstances, and provides a radiographic image processing method, apparatus, and program capable of improving the quality of a radiographic image representing the subject without increasing the dose of radiation applied to the subject. It is intended to provide.

  The first radiographic image processing method of the present invention includes a high-pressure image and a low-pressure image obtained by radiation imaging using radiation having different energy distributions for each of a plurality of similar subjects, and weighting using the high-pressure image and the low-pressure image. Among the radiographic image group consisting of one or more types of energy subtraction images formed by subtraction, (i) high pressure image and low pressure image, (ii) high pressure image and energy subtraction image, (iii) low pressure image and energy subtraction image, ( iv) An input radiation image consisting of any one of (i) to (iv) with only an energy subtraction image is prepared, and the image quality is deteriorated as compared with the input radiation image of the subject obtained by radiography of each subject. Preparing a radiographic image for teacher for each subject with a small amount and expressing a specific part in the subject, Corresponding to the subject so that image quality degradation is compensated for the input radiation image representing the subject and the specific part in the subject is emphasized is output. A teacher-learned filter that trains the teacher radiographic image as a teacher, and then creates a radiographic image of the same type as the input radiographic image for a given subject of the same type as the subject, A radiographic image of the subject is formed by inputting to the teacher-learned filter and compensating for image quality deterioration and emphasizing the same part as the specific part in the given subject. Is.

  It is desirable that the radiation dose used for radiography for creating the teacher radiographic image be larger than the radiation dose used for radiography for creating the input radiographic image.

  The teacher radiographic image can be a so-called energy subtraction image formed by weighted subtraction using a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions.

  The specific part may be a part having specific radiation absorption characteristics different from other parts.

  The subject can be a living tissue, and the specific part can be a bone or soft part of the living tissue.

  The specific part may be a part where the position in the subject has changed between the high-pressure image and the low-pressure image.

  A radiographic image of the subject formed by the radiological image processing method with the specific portion as a bone portion, the image quality deterioration being compensated, and the bone portion in the given subject being emphasized, The soft part image of the subject can be created by subtracting from the high-pressure image or low-pressure image representing the subject.

  A radiographic image of the subject formed by the radiographic image processing method, the degradation of image quality being compensated, and the noise in the given subject being emphasized is defined as the subject. A radiographic image may be created by subtracting from a bone part image or a soft part image representing.

  The specific part is a part in which the position in the subject has changed in the high-pressure image and the low-pressure image, the image quality deterioration formed by the radiological image processing method is compensated, and the given part in the subject A motion image component generated in the bone image or soft part image can be removed by subtracting the radiographic image of the subject in which the specific part is emphasized from the bone part image or soft part image representing the subject.

  The teacher-learned filter performs learning for obtaining the teacher-learned filter and formation of a radiation image for the given subject for each of a plurality of different spatial frequency bands, and is formed for each spatial frequency band. In addition, each of the radiographic images can be synthesized to obtain one radiographic image.

  The second radiographic image processing method of the present invention is an input radiological image composed of two or more (for example, three) radiographic images obtained by radiography with radiation having different energy distributions for each of a plurality of subjects of the same type. Each of the radiographic images for teachers obtained by radiography of each subject with less image quality degradation than the radiographic image for input of the subject and expressing a specific part in the subject in an emphasized manner. Prepared for each subject and outputs a radiation image of the subject in which image quality deterioration is compensated for the input radiation image representing each subject and the specific part in the subject is emphasized. As described above, a teacher-learned filter obtained by learning the teacher radiographic image corresponding to the subject as a teacher is obtained, and then, for a given subject of the same type as the subject, A radiation image of the same type as the input radiation image is created, the radiation image is input to the teacher-learned filter, image quality deterioration is compensated, and the same part as the specific part in the given subject A radiographic image of this subject formed by emphasizing is formed.

  The radiographic image processing apparatus of the present invention is formed by a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions for each of a plurality of similar subjects, and weighted subtraction using the high-pressure image and the low-pressure image. Among the radiographic image groups composed of one or more types of energy subtraction images, (i) high pressure image and low pressure image, (ii) high pressure image and energy subtraction image, (iii) low pressure image and energy subtraction image, (iv) energy subtraction The input radiation image consisting of any one of (i) to (iv) with only the image, and less deterioration in image quality than the input radiation image of the subject obtained by radiography of each subject, and the subject Each of the subjects is represented using a radiographic image for each subject that emphasizes a specific part of the subject. For the input of the input radiographic image, the radiographic image of the subject corresponding to the subject is output so that the image quality degradation is compensated and the specific portion of the subject is emphasized is output. A filter acquisition unit that obtains a teacher-learned filter that has learned a radiographic image as a teacher; a homogenous image creation unit that creates a radiographic image of the same type as the input radiographic image for a given subject of the same type as the subject; and A part image emphasis is formed by inputting a radiographic image into the teacher-learned filter and forming a radiographic image of the subject in which image quality degradation is compensated and the same part as the specific part in the given subject is emphasized And an image forming unit.

  The program according to the present invention includes a high-pressure image and a low-pressure image obtained by radiation imaging using radiation having different energy distributions for each of a plurality of similar subjects, and one type formed by weighted subtraction using the high-pressure image and the low-pressure image. Among the radiological image group consisting of the above energy subtraction images, (i) high pressure image and low pressure image, (ii) high pressure image and energy subtraction image, (iii) low pressure image and energy subtraction image, (iv) only energy subtraction image (I) to (iv) input radiation image and image quality degradation less than that of the subject input radiation image obtained by radiography of each subject, and identification in the subject For each input representing the subject using the radiographic image for each subject for which the portion of the subject is emphasized The radiographic image for the teacher corresponding to the subject is output so that the radiographic image of the subject in which the image quality deterioration is compensated for the input of the ray image and the specific portion in the subject is emphasized is output. A procedure for obtaining a teacher-learned filter that has been trained as a teacher, a procedure for creating a radiation image of the same type as the input radiation image for a given subject of the same type as the subject, and the radiation image of the teacher learned A radiographic image processing method for performing input to a filter and forming a radiographic image of a subject in which image quality deterioration is compensated and the same portion as the specific portion in the given subject is emphasized Is to make the computer execute.

  The same kind of subject means, for example, a subject having substantially the same size, shape, structure, and radiation absorption characteristics of each part. For example, the human body is the same part, and the chests of different adult males are the same type of subject. The abdomen of different adult women or the heads of different children are also the same kind of subject. For example, in the case of an industrial product, it means a subject having substantially the same size, shape, structure, material, and the like. Further, for example, the same type of subject may be a part of the chest of an adult male different from each other (for example, 1/3 of the chest close to the neck). In addition, the same type of subjects can be different small areas in the same subject.

“To create a radiation image of the same type as the input radiation image for the given subject” means that the given subject is subjected to the same processing as the processing for obtaining the input radiation image. This means that a radiographic image of a given subject is created. That is, for example, when performing radiography of the given subject under the same imaging conditions as when acquiring an input radiographic image, and acquiring the input radiographic image for the radiographic image obtained by this radiography A radiographic image of the given subject can be created by performing image processing equivalent to that described above.

  The emphasis of the specific part is not limited to the case where the specific part is expressed more prominently than the other part, but only the specific part may be expressed.

  The first and second radiographic image processing methods, apparatuses, and programs of the present invention are intended to output a radiographic image of the subject in which a specific portion in the subject is emphasized with respect to an input radiographic image. Obtain a teacher-learned filter that trains a teacher's radiographic image as a teacher, then create a radiation image of the same type as the input radiation image for a given subject, and input the radiation image to the teacher-learned filter Since the radiographic image in which the degradation of the image quality is compensated and the same part as the specific part in the subject is emphasized is formed, the above-mentioned subject can be processed without increasing the dose of the radiation to be applied to the subject. The quality of the radiation image to represent can be improved.

  That is, with respect to noise generated when a radiographic image of the same type as the input radiographic image is created for a given subject, the teacher-learned filter is subjected to the teacher radiographic image in which the generation of noise is suppressed more than the input radiographic image. Therefore, the noise generated in the radiographic image can be compensated by passing a radiographic image of the same type as the radiographic image for input through the teacher-learned filter.

  In addition, regarding the false image generated in the specific part through the supervised learned filter, the image input to the supervised learned filter is not limited to a simple radiation image as in the past, and the high-pressure image and the low-pressure image are not input. Since the image formed using both images is an image to be input to the teacher-learned filter, it is possible to more accurately distinguish a specific part from other parts and suppress the occurrence of the false image. .

  That is, in the conventional method of inputting a simple radiation image to the teacher-learned filter, the false image is generated because the reliability of estimating a specific part in the subject is insufficient. On the other hand, in the present invention, an image formed using both the high-pressure image and the low-pressure image is input to the teacher-learned filter. More image information for distinguishing the part from other parts can be used. Therefore, it is possible to increase the reliability of estimating the specific part by the teacher learned filter, and it is also possible to compensate for a false image generated in the radiation image of the given subject.

  By the above, it is possible to create a radiation image in which image quality deterioration is compensated and a specific part in the subject is emphasized without increasing the dose of radiation irradiated to a given subject, The quality of the radiographic image representing the subject can be improved.

  In addition, if the radiation dose used for radiography for creating the radiographic image for teachers is made larger than the radiation dose used for radiography for creating the radiographic image for input, The teacher radiographic image can be less deteriorated in image quality than the input radiographic image, thereby improving the quality of the radiographic image representing the subject.

  In addition, if the specific part is a part having a specific radiation absorption characteristic different from other parts, the specific part in the subject and the other part can be more reliably distinguished, and more accurately. A radiographic image formed by emphasizing a specific part in a subject can be formed.

  The radiation image processing method and apparatus and program of the present invention will be described below. The radiation image processing method according to the first embodiment of the present invention employs a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions as input radiation images. FIG. 1 is a diagram showing a procedure for acquiring a supervised learned filter used in the radiographic image processing method according to the first embodiment, and FIG. 2 is a radiographic image for acquiring a diagnostic radiographic image using the supervised learned filter. It is a figure which shows the procedure of a processing method. Note that the hatched portion in the figure indicates an image or image data representing the image.

  As shown in FIG. 1, first, a high-pressure image 11H obtained by radiation imaging 10 with radiation having different energy distributions for each of a plurality of similar subjects 1Pα, 1Pβ (hereinafter collectively referred to as subject 1P). An input radiation image 11 including a low-pressure image 11L is prepared. Further, the image quality is less deteriorated than both the high-pressure image 11H and the low-pressure image 11L, which are the input radiation images 11 of the respective subjects 1P, obtained by the radiographing 30 of the respective subjects 1P, and the respective subjects 1P. A teacher radiographic image 33 representing a specific part Px in the inside is prepared for each subject 1P. Then, for each subject 1P, the teacher learned filter 40 is obtained in which the input radiation image 11 is targeted and the teacher radiation image 33 is learned as a teacher.

  That is, the teacher-learned filter 40 uses the prepared input radiation image 11 and the teacher radiation image 33, and each of the input radiation images 11 created for each of the subjects 1Pα, 1Pβ,. Image quality degradation that has occurred in the input radiation image 11 is compensated, and a radiation image 50 representing the radiation image 50 of each subject 1P that expresses and emphasizes a specific part Px in each subject 1P is output. Thus, it is obtained by learning the radiographic image 33 for teacher corresponding to each subject 1P as a rule.

  More specifically, the teacher-learned filter 40 compensates for image quality degradation that has occurred in the input radiation image 11 when, for example, the input radiation image 11 created for the subject 1Pa is input, Further, the radiographic image 50 representing the radiographic image of the subject 1Pa expressed by emphasizing a specific part Px in the subject 1Pa is output, and thus obtained by learning using the teacher radiographic image 33 representing the subject 1Pa as a reference. Is.

  The teacher-learned filter 40 learns for each set of the input radiation image 11 and the teacher radiation image 33 corresponding to several types of prepared subjects (for example, three types of subjects 1Pα, 1Pβ, and 1Pγ). Can be obtained by:

  Here, the subject is a living tissue, and the specific part Px in the subject is a bone. In addition, a two-shot radiography method or a one-shot radiography method can be adopted for the radiography using radiation having different energy distributions.

  The teacher radiographic image 33 is obtained by radiography 30 on the subject 1P using a dose larger than the radiation dose used in the radiography 10 of each subject 1P when the input radiographic images 11 are created. The bone part image which is the energy subtraction image formed by the weighted subtraction process 32 of the obtained high pressure image 31H and the low pressure image 31L, that is, the energy subtraction process is shown.

  Here, as described above, the sum of the radiation doses used in the individual radiography 30 for each subject 1P when creating the teacher radiation image 33 is the same as that for each subject 1P when creating the input radiation image 11. Greater than the sum of the radiation doses used in individual radiography 10 for.

  As shown in FIG. 2, after obtaining the teacher-learned filter 40, a radiography 20 is performed on one given subject 3 </ b> P of the same type as the subject 1 </ b> P to be diagnosed, and the same type as the input radiation image 11. The radiation image 21 is created. Then, by inputting the radiation image 21 to the supervised learned filter 40 obtained as described above, the image quality degradation that has occurred in the radiation image of the subject 3P is compensated, and a specific part in the subject 3P is obtained. A diagnostic radiation image 60 in which Px is emphasized is formed.

  Note that a radiation image 21 of the same type as the input radiation image 11 is obtained by performing radiation imaging 20 with radiation having different energy distributions on a given subject 3P, that is, by radiation imaging under substantially the same imaging conditions as the radiation imaging 10. It consists of the obtained high-pressure image 21H and low-pressure image 21L. That is, the input radiation image 11 and the radiation image 21 are obtained by radiation imaging in which a subject is irradiated with substantially the same dose of radiation having substantially the same radiation energy distribution.

  The subjects 1Pα, 1Pβ,... Used for the creation of the input radiation image 11 and the teacher radiation image, and the subject 3P given when creating the diagnostic radiation image 60 are of the same type. The subject. That is, each of the subjects 1Pα, 1Pβ,..., 3P is a subject having substantially the same shape, structure, size, radiation absorption characteristic of each part, and the like. For example, the same kind of subjects 1Pα, 1Pβ,..., 3P can be the chest of an adult male.

  As described above, according to the radiological image processing method of the first embodiment, the quality of the radiographic image representing the subject can be improved without increasing the dose of radiation applied to the subject to be diagnosed. .

  Next, a radiation image processing method according to the second embodiment of the present invention will be described with reference to the drawings. In the second embodiment, as an input radiation image, an energy subtraction image formed by weighted subtraction using a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions, and the high-pressure image Is adopted.

  FIG. 3 is a diagram showing a procedure for acquiring a teacher-learned filter used in the radiation image processing method according to the second embodiment, and FIG. 4 is a diagram showing a procedure of the radiation image processing method using the teacher-learned filter. It is.

  As shown in FIG. 3, the radiographic image processing method according to the second embodiment starts with the chests 1Qα, 1Qβ (hereinafter collectively referred to as the chest 1Q) of a plurality of adult women, which are a plurality of similar subjects. In each case, an input radiation image 15 created based on radiation imaging 14 with radiation having different energy distributions is prepared.

  That is, for each chest 1Q that is a subject, a weighted subtraction process using a low-noise high-pressure image 15H obtained by high-dose radiography 14 and a noisy low-pressure image 15L obtained by low-dose radiography 14 An input radiation image 15 comprising two types of radiation images, that is, a bone part image 15K including a lot of noise corresponding to one type of energy subtraction image formed by 16 and the high-pressure image 15H is prepared. The high-dose radiography is radiography that irradiates a subject with a large dose of radiation, and the low-dose radiography is radiography that irradiates a subject with a dose of radiation smaller than the high dose.

  The bone part image 15K is an image mainly representing a specific part in the chest 1Q, that is, a bone part Qx that is a part showing specific radiation absorption characteristics in the chest 1Q.

  Further, the image quality is less deteriorated than any of the high-pressure image 15H and the bone portion image 15K obtained by radiation imaging 35 of the chests 1Qα, 1Qβ, and the like serving as the subject, and specific radiation in the subject. A radiological image 36 for teachers mainly representing the bone portion Qx, which is a part showing the absorption characteristics, is prepared for each chest 1Qα, 1Qβ,.

  The teacher subject image 36 representing the bone part Qx is, for example, more than the radiation dose used in individual radiography for each chest 1Qα, 1Qβ... When the input radiation image 15 is created. Can be formed by a weighted subtraction process between the high-pressure image and the low-pressure image obtained by the radiography 35 of the chests 1Qα, 1Qβ,... Using a large dose.

  Next, a teacher-learned filter 41 is obtained by learning the input radiation image 15 composed of the bone part image 15K and the high-pressure image 15H and learning the teacher radiation image 36 as a teacher.

  That is, the teacher-learned filter 41 compensates for image quality degradation when the bone image 15K and the high-pressure image 15H, which are input radiation images 15 for each of the chests 1Qα, 1Qβ,. In addition, the teacher radiographic image 36 is transmitted to the teacher so that a radiographic image 51 of each chest 1Qα, 1Qβ,... Mainly representing the bone portion Qx, which is a specific part in the chest 1Qα, 1Qβ,. Can be obtained by learning as.

  Here, the teacher-learned filter 41, for example, compensates for image quality degradation when the input radiation image 15 of the chest 1Qα is input, and mainly selects the bone portion Qx that is a specific part in the chest 1Qα. It is obtained by learning the teacher radiation image 36 representing the chest 1Qα as a teacher so that the radiation image 51 of the chest 1Qα represented in FIG.

  After obtaining the teacher-learned filter 41, a radiation image 25 of the same type as the input radiation image 15 is created for the breast 3Q of an adult female to be diagnosed, which is the same kind of subject as the chest 1Q, and the radiation image is obtained. 25 is input to the teacher-learned filter 41 to form a diagnostic radiographic image 61 that mainly compensates for the bone part Qx, which is a specific part in the chest 3Q to be diagnosed, with image quality deterioration compensated.

  The radiographic image 25 is obtained by the low-pressure radiography and the low-noise radiography and the high-noise image 25H obtained by the high-dose radiography obtained by the radiography 24 with the radiation having different energy distributions with respect to the chest 3Q. The image includes a bone portion image 25K including a lot of noise, which is an energy subtraction image formed by the weighted subtraction process 26 using the low-noise image 25L with much noise, and the high-pressure image 25H.

  It should be noted that by subtracting the diagnostic radiographic image 61 mainly representing the bone portion with little noise from the high-voltage image 15H with little noise, a soft part image with little noise that is the second diagnostic radiographic image can be created. .

  As described above, according to the second embodiment, the quality of a radiographic image representing the subject can be improved without increasing the dose of radiation applied to the subject.

  Here, the teacher learned filter 41 will be described in detail. In addition, as described below, one image is converted into a plurality of images for different spatial frequency bands, and image processing is performed for each converted image to generate a plurality of processed images for each spatial frequency band. As a technique for combining the plurality of processed images to obtain one processed image, various conventionally known techniques can be employed.

  FIG. 5 is a diagram illustrating a state where an input radiation image is input to a supervised learned filter for each spatial frequency band to obtain a diagnostic radiographic image, and FIG. 6 illustrates a supervised learned filter for each spatial frequency band by learning. FIG. 7 is a diagram showing how to obtain a radiographic image for teacher composed of a plurality of spatial frequency bands. FIG. 13 is a diagram showing how upsampling and addition are performed in the image synthesis filter.

  Here, input radiation images for a plurality of subjects of the same type are formed by high-pressure images and low-pressure images obtained by radiation imaging with radiation having different energy distributions, and weighted subtraction using the high-pressure images and low-pressure images. In addition, it is selected from a group of radiographic images composed of one or more types of energy subtraction images. Here, the input radiation image is a plurality of high-pressure images for different spatial frequency bands and a plurality of bone images which are a plurality of energy subtraction images for different spatial frequency bands. The teacher radiographic image has less image quality degradation than the input radiographic image obtained by radiography of the same type of subject as the subject, and expresses a specific portion in the subject in an emphasized manner. A plurality of radiographic images for teachers for different spatial frequency bands.

  The supervised learned filter is intended for input radiation images composed of a plurality of high-pressure images for the different spatial frequency bands and a plurality of bone images for the different spatial frequency bands. It is assumed that a plurality of teacher radiation images for different spatial frequency bands are learned as a teacher.

  Then, for a given subject of the same type as the subject, a plurality of radiation images for the different spatial frequency bands of the same type as the input radiation image are created, and a plurality of radiations for the different spatial frequency bands are created. An image is input to a teacher-learned filter, and the teacher-learned filter compensates for image quality degradation and emphasizes a specific part in a given subject. A radiographic image is formed. Then, one radiation image is created by combining the plurality of radiation images.

  That is, as shown in FIG. 5, the teacher-learned filter 41 has different spatial frequencies obtained by multi-resolution conversion of the high-pressure image 25H and the bone part image 25K of a given subject 3Q to be diagnosed. Based on the radiographic image input for each band, a plurality of radiographic images 61H, 61M, 61L for each spatial frequency band to be diagnosed are created, and the created radiographic images 61H, 61M, 61L are synthesized. Thus, a diagnostic radiation image 61 is obtained.

  Here, the teacher learned filter 41 is composed of a high frequency band teacher learned filter 41H, a medium frequency band teacher learned filter 41M, a low frequency band teacher learned filter 41L, an image synthesis filter 41T, and the like. .

  In addition, as shown in FIG. 6, the radiographic images 36H, 36M, and 36L for each spatial frequency band representing the chest 1Q prepared for creating the teacher learned filter 41 are compensated for image quality degradation. In addition, the radiographic image 36 (bone high-resolution image) mainly representing the bone that is the specific part is obtained by multi-resolution conversion.

  In addition, bone images 15KH, 15KM, 15KL, which are radiographic images for each spatial frequency band representing the chest 1Q prepared for creating the teacher-learned filter 41, and high-pressure images 15HH, 15HM, 15HL, respectively. Similarly to the case of the teacher radiation image 36, the bone image 15K and the high-pressure image 15H are obtained by multi-resolution conversion.

  More specifically, as the teacher radiation image, the following images for each spatial frequency band obtained by multi-resolution conversion of the teacher radiation image 36, that is, a radiation image representing a high frequency band (hereinafter referred to as a teacher image). High-frequency band image 36H), a radiographic image representing the intermediate frequency band (hereinafter referred to as a teacher medium-frequency band image 36M), and a radiographic image representing a low-frequency band (hereinafter referred to as a teacher low-frequency band image 36L). .

  Further, as a bone part image, the following image for each spatial frequency band obtained by multiresolution conversion of the bone part image 15K, that is, a radiographic image representing a high frequency band (hereinafter referred to as a bone part high frequency band image 15KH). ), A radiographic image representing the intermediate frequency band (hereinafter referred to as bone frequency band image 15KM) and a radiographic image representing the low frequency band (hereinafter referred to as bone frequency band image 15KL).

  In addition, as a high-pressure image, the following image for each spatial frequency band obtained by multi-resolution conversion of the high-pressure image 15H, that is, a radiation image representing a high frequency band (hereinafter referred to as a high-pressure high-frequency band image 15HH), It is composed of a radiographic image representing a frequency band (hereinafter referred to as a high-voltage medium frequency band image 15HM) and a radiographic image representing a low-frequency band (hereinafter referred to as a high-voltage low-frequency band image 15HL).

  As shown in FIG. 7, for example, the high-voltage high-frequency band image 15HH includes a high-pressure image 15H (high-pressure high-resolution image) that is the high-pressure high-resolution image and a high-pressure medium-resolution image obtained by down-sampling the high-pressure image 15H. It was obtained by upsampling with 15H1.

  In the downsampling, a Gaussian low-pass filter with σ = 1 and 1/2 thinning of the high-pressure image 15H are performed. The upsampling is performed using cubic B-spline interpolation.

  As in the case of the high-voltage high-frequency band image 15HH, the high-voltage medium-frequency band image 15HM is an up-conversion of the high-voltage medium-resolution image 15H1 and the high-voltage low-resolution image 15H2 obtained by down-sampling the high-pressure medium-resolution image 15H1. It was obtained by sampling.

  The high-voltage low-frequency band image 15HL is obtained by down-sampling the high-voltage low-resolution image 15H2 and the high-pressure low-resolution image 15H2, similarly to the case where the high-voltage high-frequency band image 15HH or the high-pressure medium-frequency band image 15HM is acquired. It was obtained by upsampling with the high-pressure ultra-low resolution image 15H3.

  The supervised learned filter 41 is created for each of the three spatial frequency bands. That is, the high frequency band teacher learned filter 41H, the medium frequency band teacher learned filter 41M, and the low frequency band teacher learned filter 41L are obtained by learning for each spatial frequency band.

  Hereinafter, the case where the high frequency band teacher learned filter 41H is obtained by learning will be described with reference to FIG.

  As shown in FIG. 6, each of the bone high-frequency band image 15KH, the high-voltage high-frequency band image 15HH, and the teacher high-frequency band image 36H is a rectangular area of 5 pixels × 5 pixels (25 pixels in total), which is a small region corresponding to each other. A sub-window Sw composed of areas is set.

  Then, with respect to the feature amount composed of 25 pixels each constituting the sub-window Sw of the bone high-frequency band image 15KH and the high-voltage high-frequency band image 15HH, the value of the center pixel of the sub-window Sw in the teacher high-frequency band image 36H is set as the target value. And a plurality of learning samples are extracted while moving the subwindow. The high frequency band teacher learned filter 41H is obtained by learning, for example, 10,000 kinds of learning samples extracted in this manner.

  Note that the high frequency band image 51H, the intermediate frequency band image 51M, and the low frequency band image 71L described later are similar images of the teacher high frequency band image 36H, the teacher medium frequency band image 36M, and the teacher low frequency band image 36L, respectively. It is.

  The high frequency band teacher learned filter 41H is obtained by learning a regression model using support vector regression described later. This regression model is based on the feature amount (image represented by 25 pixels) of the input bone high-frequency band image 15KH and the feature amount (image represented by 25 pixels) of the high-pressure high-frequency band image 15HH. Accordingly, it is a high-frequency band non-linear filter that compensates for image quality degradation and outputs a high-frequency band image 51H that mainly represents the bone that is the specific part.

  Further, the learning filter 41M for the intermediate frequency band is obtained by learning similar to the above using the bone intermediate frequency band image 15KM, the high-voltage intermediate frequency band image 15HM, and the teacher intermediate frequency band image 36M.

  Further, the low frequency band teacher learned filter 41L is obtained by learning similar to the above using the bone low frequency band image 15KL, the high voltage low frequency band image 15HL, and the teacher low frequency band image 36L.

  As described above, the learning of the regression model is performed for each spatial frequency band, and the teacher learned filter 41 including the teacher learned filter 41H, the teacher learned filter 41M, and the teacher learned filter 41L is obtained.

  As shown in FIG. 5, the teacher-learned filter 41 created as described above has the same kind as that of the input radiation image 15 created for the breast 3Q of the given adult woman to be diagnosed. A radiographic image for each spatial frequency unit obtained by multi-resolution conversion of each of the bone image 25K and the high-pressure image 25H, which are the radiographic images 25 of the diagnosis object, is input.

  That is, the bone part high frequency band image 25KH obtained by multi-resolution conversion of the bone part image 25K, the bone part medium frequency band image 25KM, the bone part low frequency band image 25KL, and the high pressure high frequency image 25H obtained by multiple conversion. 25HH, high-voltage medium frequency band image 25HM, and high-voltage low frequency band image 25HL are input to the teacher learned filter 41.

  The teacher-learned filters 41H, 41M, and 41L, to which images for each spatial frequency band obtained by multi-resolution conversion of the bone image 25K and the high-pressure image 25H are input, are diagnosed for each spatial frequency band. The target radiographic images 61H, 61M, and 61L are estimated, and the radiographic images 61H, 61M, and 61L obtained by the estimation are synthesized by the image synthesis filter 41T to obtain the diagnostic radiographic image 61.

  That is, when the bone high-frequency band image 25KH and the high-voltage high-frequency band image 25HH are input to the high-frequency band teacher-learned filter 41H, the image quality deterioration is compensated and the bone that is the specific part is mainly represented. A diagnostic radiographic image 61H in the high frequency band is formed.

  Further, when the bone middle frequency band image 25KM and the high voltage middle frequency band high voltage image 25HM are input to the medium frequency band teacher learned filter 41M, the image quality degradation is compensated for and the bone part which is the specific part A diagnostic radiographic image 61M in the middle frequency band that mainly represents is formed.

  Further, when the bone low-frequency band image 25KL and the high-voltage low-frequency band image 25HL are input to the low-frequency band teacher-learned filter 41L, the image quality degradation is compensated for and the bone that is the specific part is detected. A diagnostic radiation image 61L in the low frequency band mainly represented is formed.

  Then, the formed diagnostic radiographic image 61H in the high frequency band, diagnostic radiographic image 61M in the medium frequency band, and diagnostic radiographic image 61L in the low frequency band are synthesized by the image synthesizing filter 41T, and the diagnostic radiographic image is obtained. 61 is created.

  As shown in FIG. 13, the image synthesis filter 41T repeats upsampling and addition in the order of the diagnostic radiographic image 61L in the low frequency band, the diagnostic radiographic image 61M in the medium frequency band, and the diagnostic radiographic image 61H in the high frequency band. The diagnostic radiation image 61 is obtained.

  That is, an image obtained by upsampling the diagnostic radiographic image 61L in the low frequency band and the diagnostic radiographic image 61M in the medium frequency band is obtained, and an image obtained by upsampling the image is obtained. The diagnostic radiographic image 61 is obtained by adding the diagnostic radiographic image 61H in the high frequency band.

  As described above, the teacher learned filter can be obtained by learning in a plurality of different spatial frequency bands.

  The feature amount input in the regression learning will be described in detail below. FIG. 8 is a diagram showing an example of regions constituting the feature amount.

  The feature amount may not be the pixel value itself in the radiation image for each spatial frequency band, but may be obtained by performing a special filter process. For example, as shown in FIG. 8, the average value of each pixel value of the region U1 or the region U2 composed of three pixels adjacent to each other in the vertical direction or the horizontal direction in the radiation image in a specific spatial frequency band is a new feature amount. It is good. In addition, wavelet transform may be performed and wavelet coefficients may be used as feature amounts. Further, pixels over a plurality of frequency bands may be used for the feature amount.

  The contrast normalization performed in the regression learning will be described below.

  A standard deviation is calculated for the pixel value of each pixel included in the sub-window Sw (see FIG. 6) set in the image for each spatial frequency band. The pixel value of the band image is multiplied by a coefficient so that the standard deviation matches a predetermined target value.

I '= I × (C / SD)
Here, I is the pixel value of the original image, I ′ is the pixel value after contrast normalization, SD is the standard deviation for the pixel value in the sub-window Sw, C is the target value of the standard deviation (the above C is set in advance) Constant).

  The sub-window Sw is scanned so as to cover the entire region in each radiation image, and the pixel value in the sub-window Sw is set so that the standard deviation approaches the target value for all sub-windows Sw that can be set in each image. Normalize by multiplying by a predetermined coefficient.

  As a result of the normalization, the magnitudes (contrasts) of the amplitudes of the image components for each spatial frequency band are aligned. Thereby, since the variation of the image pattern in the radiographic image for every spatial frequency band input into the said teacher learning filter 41 reduces, the effect which improves the estimation precision of the said bone part is acquired.

  In the step of learning the supervised learned filter that is the non-linear filter, the high-pressure image is subjected to the above-described contrast normalization process, and the multiplied coefficient is also multiplied to the bone part image without deterioration. A learning sample is prepared from a pair of normalized high-pressure image and bone part image to learn a nonlinear filter.

  In the step of estimating the diagnostic radiographic image mainly representing the bone part of the diagnostic object, the input high-voltage image is subjected to contrast normalization, and the pixel value of each normalized spatial frequency band is set as a teacher-learned filter. To enter. The output value of the teacher-learned filter is multiplied by the reciprocal of the coefficient used when the normalization is performed to obtain an estimated value of the bone.

  Next, support vector regression (regression by support vector machine (SVR)) will be described.

  FIG. 9 is a diagram showing how an approximate function is obtained by support vector regression.

Regarding the problem of learning a function that approximates a real value y corresponding to a d-dimensional input vector x, first consider the case where the approximation function is linear.

  The ε-SVR algorithm proposed by Vapnik finds f that minimizes the following loss function.

  The following documents can be referred to for the ε-SVR algorithm proposed by Vapnik.

Nello Cristianini (Author), John Shawe-Taylor (Author), Takeshi Ohkita (Translation), titled “Introduction to Support Vector Machine”, Kyoritsu Shuppan, published on March 25, 2005, p. 149 to P.I. 156.

<W · w> is a term representing the complexity of a model that approximates data, and Remp [f] is expressed as follows.

  Here, | y−f (x) | ε = max {0, | y−f (x) | −ε}, which means that errors smaller than ε are ignored. ξ and ξ * are relaxation variables that allow an error exceeding ε in the positive direction and the negative direction, respectively. C is a parameter that sets a trade-off between the complexity of the model and relaxation of constraints.

The above main problem is equivalent to solving the following dual problem, and a global solution can always be obtained from the characteristics of the convex quadratic programming problem.

The regression model obtained by solving this dual problem is expressed by the following equation.

This function is a linear function, but in order to extend it nonlinearly, if the input X is mapped to a higher-order feature space Φ (X) and the vector Φ (X) in that feature space is regarded as the previous input X, Good (X → Φ (X)). In general, mapping to a high-dimensional space is accompanied by a significant increase in computational complexity, but the inner product terms appearing in the formula to be optimized are expressed as K (x, y) = <Φ (x), Φ (y)>. If it is replaced with a kernel function that satisfies, the same result as that calculated after mapping to a higher dimension in the calculation of the input dimension can be obtained. As the kernel function, an RBF kernel, a polynomial kernel, a sigmoid kernel, or the like can be used.

  Next, a radiation image processing method according to a third embodiment of the present invention will be described with reference to the drawings. In the third embodiment, as an input radiation image, an image consisting only of an energy subtraction image formed by weighted subtraction using a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions is used. Adopted.

  FIG. 10 is a diagram showing a procedure for acquiring a teacher-learned filter used in the radiation image processing method according to the third embodiment, and FIG. 11 is a diagram showing a procedure of the radiation image processing method using the teacher-learned filter. It is.

  In the radiographic image processing method according to the third embodiment of the present invention, first, the energy of each of the adult female breasts 1Rα, 1Rβ (hereinafter collectively referred to as the chest 1R), which are a plurality of subjects of the same kind, is mutually determined. A weighted subtraction process 77 using a high-pressure image 72H with low noise obtained by radiography of high dose obtained by radiography 71 with radiation having a different distribution and a low-pressure image 72L with high noise obtained by radiography of low dose. Thus, a soft part image 73A including a large amount of noise, which is one type of energy subtraction image, is formed. Then, a low-pass filter process 74 is applied to the soft part image 73A to obtain a soft part image 73B from which high-frequency components have been removed. Further, the subtraction process 75 for subtracting the soft part image 73B from which the high-frequency component is removed from the high-pressure image 72H with little noise increases the mixing of the soft part component toward the high frequency side. An input radiation image 76 that is a bone image with little mixing is prepared.

The high-frequency component means a component having a high spatial frequency in the image, and the low-frequency component means a component having a low spatial frequency in the image. Although more noise components are included on the side of, the noise components are removed by the low-pass filter processing 74. Therefore, the input radiation image 76, which is the bone image, is an image with less noise as a whole.

  Further, image quality deterioration is less than that of the input radiation image 76 obtained by the radiation imaging 37 of the breasts 1Rα, 1Rβ,... A teacher radiation image 38 which is a bone image representing only a specific part in the chest 1R is prepared.

  Then, the teacher learned filter 42 is obtained by learning the input radiation image 76 and learning the teacher radiation image 38 as a teacher.

  That is, the teacher-learned filter 42 uses a set of an input radiation image 76 and a teacher radiation image 38 prepared for each chest 1R as a subject, and each input radiation image 76 for each chest 1R Learning the radiographic image 38 for each chest 1R as a teacher so that the radiographic image 52 representing the radiographic image of the chest 1R representing only the bone portion is output when the image quality is compensated. To obtain.

  After obtaining the teacher-learned filter 42, as shown in FIG. 11, the same kind of radiation as the input radiation image 76 based on the radiation imaging 71 ′ of the breast 3R of an adult female as a given subject. An image 76 'is created. Then, the radiation image 76 ′ is input to the teacher learned filter 42 to form a radiation image 62 in which image quality deterioration is compensated and only the bone portion in the chest 3 R is represented. Thereby, the quality of the radiographic image representing the subject can be improved without increasing the dose of radiation applied to the subject.

  Here, the radiographic image 76 ′ is created by following substantially the same procedure as that for forming the input radiographic image 76 for the chest 3R that is the given subject. This radiographic image 76 'is a bone image that is equivalent to the input radiographic image 76 and has a soft portion that increases as it approaches the high-frequency component side, but is less contaminated with noise from the high-frequency component side to the low-frequency component side as a whole. is there.

  The specific part in the subject may represent a motion artifact caused by a timing difference between the high-pressure image shooting and the low-pressure image shooting. A specific part in the subject representing a motion artifact component that is a position change component between the two images is taken from when the high-pressure image (or low-pressure image) is taken until the low-pressure image (or high-pressure image) is taken (for example, , For 0.1 seconds), it can represent a part moved in the subject (a part whose position in the subject has changed). For example, in the case where the subject is a chest of a living tissue, the specific part in the subject is taken after the high-pressure image (or low-pressure image) is taken until the low-pressure image (or high-pressure image) is taken. It can be set as the site | part which moved according to the heartbeat.

  FIG. 12 is a diagram showing motion artifacts that occur in the bone image representing the chest.

  As shown in FIG. 12, in a bone image FK representing an adult female breast that is an energy subtraction image formed by weighted subtraction using a high-pressure image and a low-pressure image obtained by radiation imaging with different energy distributions. In some cases, a motion artifact Ma generated in response to the heartbeat may occur. Since such motion artifacts are required to be removed from the radiographic image, a radiographic image that emphasizes the motion artifact Ma that is the specific part is formed through the supervised learned filter, and so on. By subtracting the radiographic image formed in this manner from the bone part image FK, a bone part image from which the motion artifact component, which is a component representing the motion artifact Ma, is removed can be created.

  As described above, the specific part can be a part where the display position in the subject has changed between the high-pressure image and the low-pressure image. In addition, as described above, the specific part that is emphasized may be an unnecessary part (defect part) in the radiographic image. In such a case, by subtracting the radiation image representing the unnecessary part from the radiation image including both the unnecessary part and the necessary part, the desired radiation image representing only the necessary part is removed from the unnecessary part. Obtainable.

  Further, as a method for acquiring the radiographic image, either a one-shot method or a two-shot method may be used.

  In radiography for acquiring a high-pressure image and a low-pressure image, the radiation dose used for acquiring the low-pressure image relative to the dose of radiation used for acquiring the high-pressure image may be increased or decreased. When the purpose is to suppress noise, it is preferable that the radiation dose used for acquiring the high-pressure image is larger than the radiation dose used for acquiring the low-pressure image.

  Further, the regression learning method may use a neural network, a Relevance Vector Machine, or the like in addition to the support vector machine.

  In addition, when acquiring radiographic images for teachers of each subject based on radiography using a dose larger than the dose of radiation used for radiography for each subject when obtaining an input radiographic image, There is a case where the dose of radiation irradiated to one subject in the radiography of the subject exceeds an allowable value. However, the above large dose is used by limiting the sum of the doses of radiation irradiated to the subject during a predetermined period. Radiation imaging can be performed on a subject for teachers.

  Hereinafter, the radiographic image processing method representing each embodiment as described above will be described again.

  The radiological image processing method representing each of the embodiments described above uses a high-pressure image and a low-pressure image of the subject obtained by radiography of the subject with radiation having different energy distributions to identify the subject. The radiographic image which emphasizes and represents the site | part of this is acquired.

  In the above method, first, for a plurality of subjects, for each subject, an input radiation image composed of two or more types of radiation images obtained by radiography of the subject with radiation having different energy distribution, or a high pressure image and a low pressure One or more types of input radiographic images created using both images are prepared, and image quality degradation is less than that of the input radiographic images generated from the radiographic images obtained by radiography of the respective subjects. In addition, a radiographic image for teacher that expresses and emphasizes the specific part in the subject is prepared, and when the input radiographic image for each subject is input, image quality degradation is compensated for, and The teacher has learned the teacher radiation image corresponding to the subject as a teacher so that a radiation image in which the specific part of the subject is emphasized is output. Get a filter.

  Thereafter, for the given subject of the same type as the subject, a radiation image of the same type as the input radiation image is created by the same processing as when the input radiation image is created. That is, substantially the same as when the input radiographic image is created for the given subject under the same radiographing conditions as when the input radiographic image is created, and for the radiographic image obtained by the radiography. A radiographic image corresponding to the input radiographic image for the given subject is created by image processing. Then, a radiographic image of the subject corresponding to the created radiographic image for input is input to a teacher-learned filter, image quality deterioration is compensated, and a specific part in the given subject is emphasized. A radiographic image representing the radiographic image of the subject is obtained.

  The input radiation image includes a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions, and one or more types of energy formed by weighted subtraction using the high-pressure image and the low-pressure image. Among the radiological image group consisting of subtraction images, (i) high pressure image and low pressure image, (ii) high pressure image and energy subtraction image, (iii) low pressure image and energy subtraction image, (iv) image consisting only of energy subtraction image, etc. Can be adopted.

  As shown in FIGS. 1 and 2, a radiographic image processing apparatus 110 that performs the radiographic image processing method of the present invention performs radiography with radiation having different energy distributions for each subject 1P constituting a plurality of subjects 1P of the same type. 10 and an input radiation image 11 comprising the high-pressure image 11H and the low-pressure image 11L obtained in 10 and the high-pressure image 11H that is the input radiation image 11 of each subject 1P obtained by the radiography 30 of each subject 1P. And the low-pressure image 11L has less image quality degradation, and the radiographic image 33 for each input of the subject 1P is used to emphasize the specific part Px in the subject 1P. A radiographic image of the subject is output in which image quality degradation is compensated for the input and the specific part in the subject is emphasized. As described above, the filter acquisition unit Mh1 (see FIG. 1) for obtaining the teacher-learned filter 40 in which the teacher radiation image 33 corresponding to the subject 1P is learned as a teacher, and the subject 1P to be diagnosed Radiation imaging 20 is performed on a given subject 3P of the same kind, and the same kind of image creation unit Mh2 (see FIG. 2) for creating the same kind of radiation image 21 as that of the input radiation image 11, and the radiation image 21 as described above. Input to the teacher-learned filter 40 obtained in this manner, the image quality degradation that has occurred in the radiation image of the subject 3P is compensated, and the specific radiation P60 in the subject 3P is emphasized. And a part-enhanced image forming unit M3 (see FIG. 2).

  Since the operation of the radiographic image processing apparatus 110 is the same as that of the radiographic image processing method already described, a description thereof will be omitted. Note that each image handled by the filter acquisition unit Mh1, the same kind of image creation unit Mh2, and the part-enhanced image formation unit Mh3 can be an image itself or image data representing an image.

  Note that the teacher-learned filter is not learned for each small region, but only one type is prepared for each frequency, and all small regions are processed by one filter. In the learning of the filter, learning samples are taken out from various small regions of a single (or a small number) of radiographic images, and these many samples are treated as a set at the same time. That is, for example, learning samples consisting of Mr. A's collarbone, Mr. A's collarbone, Mr. A's rib contour, Mr. A's rib center, etc. are collectively learned. In addition, although the feature quantity to be input to the filter is 25 pixels, the teacher who is the output corresponding to the 25 pixels is not 25 pixels but one pixel at the center of the small area.

  Further, a program for executing the function of the radiation image processing apparatus of the present invention can be installed in a personal computer, and the same operation as in the above embodiment can be executed in the personal computer. That is, a program for causing a computer to execute the radiographic image processing method of the above embodiment corresponds to the program of the present invention.

The figure which shows the procedure which acquires the teacher learning completed filter used for the radiographic image processing method of 1st Embodiment. The figure which shows the procedure of the radiographic image processing method of 1st Embodiment The figure which shows the procedure which acquires the teacher learning completed filter used for the radiographic image processing method of 2nd Embodiment. The figure which shows the procedure of the radiographic image processing method of 2nd Embodiment. The figure which shows a mode that the image which consists of several spatial frequency bands is obtained from the radiographic image for teachers The figure which shows a mode that the teacher learned filter for every spatial frequency band is obtained by learning The figure which shows a mode that the radiographic image for an input is input into a teacher learned filter for every spatial frequency band, and the radiographic image for a diagnosis is obtained It is a figure which shows the area | region which comprises a feature-value. Diagram showing how approximate function is calculated by support vector regression The figure which shows the procedure which acquires the teacher learning completed filter used for the radiographic image processing method of 3rd Embodiment. The figure which shows the procedure of the radiographic image processing method of 1st Embodiment It is a figure which shows the motion artifact which arose in the bone part image showing a chest. Diagram showing upsampling and addition in image synthesis filter

Explanation of symbols

1P Subject 3P Subject 10 Radiography 11H High pressure image 11L Low pressure image 11 Radiation image for input 21 Radiation image 33 Radiation image for teacher 40 Teacher-trained filter 60 Radiation image

Claims (12)

A high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions for each of a plurality of similar subjects, and one or more types of energy subtraction images formed by weighted subtraction using the high-pressure image and the low-pressure image (I) high pressure image and low pressure image, (ii) high pressure image and energy subtraction image, (iii) low pressure image and energy subtraction image, and (iv) energy subtraction image only (i) ( iv) Prepare an input radiation image consisting of either
Prepared for each subject is a teacher radiation image obtained by radiographing each subject with less image quality degradation than the input radiation image of the subject and representing a specific part in the subject with emphasis. And
Obtaining a teacher-learned filter that trains the teacher radiation image corresponding to the subject as a teacher with respect to the input radiation image representing each subject,
Then, for a given subject of the same type as the subject, create a radiation image of the same type as the input radiation image,
Inputting the radiographic image into the teacher-learned filter to form a radiographic image of the subject in which image quality degradation is compensated and the same part as the specific part in the given subject is emphasized A radiation image processing method.   The radiation dose used for radiography for creating the teacher radiographic image is larger than the radiation dose used for radiography for creating the input radiographic image. The radiation image processing method according to 1.   3. The radiation according to claim 1, wherein the teacher radiation image is formed by weighted subtraction using a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distributions. Image processing method.   The radiographic image processing method according to claim 1, wherein the specific part is a part having specific radiation absorption characteristics different from other parts.   The radiographic image processing method according to claim 1, wherein the subject is a living tissue, and the specific part is a bone part or a soft part of the living tissue.   The radiographic image processing method according to claim 1, wherein the specific part is a part in which a position in a subject is changed between the high-pressure image and the low-pressure image. The specific part is a bone part;
A radiographic image of the subject formed by the radiographic image processing method, in which the image quality deterioration is compensated and the bone portion in the given subject is emphasized, is represented as a high-pressure image or a low-pressure image representing the subject. 6. The radiographic image processing method according to claim 1, wherein a soft part image of the subject is created by subtracting from the image. The specific part is a part where the position in the subject has changed in the high-pressure image and the low-pressure image,
A radiographic image of the subject formed by the radiographic image processing method, in which the deterioration in image quality is compensated, and the specific part in the given subject is emphasized, is a bone part image or a soft part representing the subject. 6. The radiographic image processing method according to claim 1, wherein motion artifact components generated in the bone image or soft image are subtracted from the image to remove the motion artifact component.   The teacher-learned filter performs learning for obtaining the teacher-learned filter and formation of a radiation image for the given subject for each of a plurality of different spatial frequency bands, and is formed for each spatial frequency band. 9. The radiographic image processing method according to claim 1, wherein each radiographic image is synthesized to obtain one radiographic image. For each of the same type of subjects, prepare an input radiation image consisting of two or more types of radiation images obtained by radiation imaging with radiation having different energy distributions.
Prepared for each subject is a teacher radiation image obtained by radiographing each subject with less image quality degradation than the input radiation image of the subject and representing a specific part in the subject with emphasis. And
Obtaining a teacher-learned filter that trains the teacher radiation image corresponding to the subject as a teacher for the input radiation image representing each subject,
Then, for a given subject of the same type as the subject, create a radiation image of the same type as the input radiation image,
Inputting the radiographic image into the teacher-learned filter to form a radiographic image of the subject in which image quality degradation is compensated and the same part as the specific part in the given subject is emphasized A radiation image processing method. From a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distribution for each of a plurality of similar subjects, and one or more types of energy subtraction images formed by weighted subtraction using the high-pressure image and the low-pressure image (I) high-pressure image and low-pressure image, (ii) high-pressure image and energy subtraction image, (iii) low-pressure image and energy subtraction image, and (iv) energy subtraction image only (iv) to (iv) ) And an input radiation image formed by any of the above and a radiographic image of each subject, the image quality degradation is less than that of the input radiation image of the subject, and a specific part in the subject is emphasized. The input radiographic image representing each subject is input using the teacher radiographic image for each subject represented. A filter obtaining means for obtaining a teacher trained filter to learn the teacher radiation image corresponding to the object as a teacher,
For a given subject of the same type as the subject, the same kind of image creating means for creating a radiation image of the same kind as the input radiation image;
The radiation image is input to the teacher-learned filter to form a radiation image of the subject in which image quality degradation is compensated and the same part as the specific part in the given subject is emphasized A radiation image processing apparatus comprising: an enhanced image forming unit. From a high-pressure image and a low-pressure image obtained by radiation imaging with radiation having different energy distribution for each of a plurality of similar subjects, and one or more types of energy subtraction images formed by weighted subtraction using the high-pressure image and the low-pressure image (I) high-pressure image and low-pressure image, (ii) high-pressure image and energy subtraction image, (iii) low-pressure image and energy subtraction image, and (iv) energy subtraction image only (iv) to (iv) ) And an input radiation image formed by any of the above and a radiographic image of each subject, the image quality degradation is less than that of the input radiation image of the subject, and a specific part in the subject is emphasized. The input radiographic image representing each subject is input using the teacher radiographic image for each subject represented. A step of obtaining a teacher trained filter to learn the teacher radiation image corresponding to the object as a teacher,
Creating a radiation image of the same type as the input radiation image for a given subject of the same type as the subject;
A procedure of inputting the radiographic image to the teacher-learned filter to form a radiographic image of the subject in which image quality degradation is compensated and the same part as the specific part in the given subject is emphasized A program for causing a computer to execute a radiographic image processing method.