WO2022172506A1 - 放射線画像処理方法、機械学習方法、学習済みモデル、機械学習の前処理方法、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム - Google Patents

放射線画像処理方法、機械学習方法、学習済みモデル、機械学習の前処理方法、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム Download PDF

Info

Publication number
WO2022172506A1
WO2022172506A1 PCT/JP2021/037173 JP2021037173W WO2022172506A1 WO 2022172506 A1 WO2022172506 A1 WO 2022172506A1 JP 2021037173 W JP2021037173 W JP 2021037173W WO 2022172506 A1 WO2022172506 A1 WO 2022172506A1
Authority
WO
WIPO (PCT)
Prior art keywords
noise
image
radiation
pixel
radiographic image
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2021/037173
Other languages
English (en)
French (fr)
Japanese (ja)
Inventor
智 土屋
達也 大西
敏康 須山
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hamamatsu Photonics KK
Original Assignee
Hamamatsu Photonics KK
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hamamatsu Photonics KK filed Critical Hamamatsu Photonics KK
Priority to KR1020237028293A priority Critical patent/KR20230145081A/ko
Priority to CN202180093696.5A priority patent/CN116916827A/zh
Priority to EP21925753.2A priority patent/EP4252658A4/en
Priority to US18/270,897 priority patent/US20240054617A1/en
Priority to JP2022581176A priority patent/JP7712967B2/ja
Publication of WO2022172506A1 publication Critical patent/WO2022172506A1/ja
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/60—Image enhancement or restoration using machine learning, e.g. neural networks
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/42—Arrangements for detecting radiation specially adapted for radiation diagnosis
    • A61B6/4291—Arrangements for detecting radiation specially adapted for radiation diagnosis the detector being combined with a grid or grating
    • A—HUMAN NECESSITIES
    • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
    • A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
    • A61B6/00—Apparatus or devices for radiation diagnosis; Apparatus or devices for radiation diagnosis combined with radiation therapy equipment
    • A61B6/52—Devices using data or image processing specially adapted for radiation diagnosis
    • A61B6/5258—Devices using data or image processing specially adapted for radiation diagnosis involving detection or reduction of artifacts or noise
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10116—X-ray image
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20081—Training; Learning
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20084—Artificial neural networks [ANN]
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/20—Special algorithmic details
    • G06T2207/20172—Image enhancement details
    • G06T2207/20182—Noise reduction or smoothing in the temporal domain; Spatio-temporal filtering

Definitions

  • One aspect of the embodiments relates to a radiation image processing method, a machine learning method, a trained model, a machine learning preprocessing method, a radiation image processing module, a radiation image processing program, and a radiation image processing system.
  • an object is to provide a radiographic image processing module, a radiographic image processing program, and a radiographic image processing system.
  • a radiation image processing method includes an image acquisition step of acquiring a radiation image in which an object is irradiated with radiation and the radiation transmitted through the object is captured, and evaluation of the spread of pixel values and noise values.
  • An evaluation value is derived from the pixel value of each pixel of the radiographic image based on relational data representing the relationship with the evaluation value obtained, and a noise map is generated as data in which the derived evaluation value is associated with each pixel of the radiographic image.
  • a radiographic image processing module includes an image acquisition unit that acquires a radiographic image obtained by irradiating an object with radiation and capturing the radiation that has passed through the object, and a pixel value and a noise value.
  • An evaluation value is derived from the pixel value of each pixel in the radiographic image based on relational data representing the relationship with the evaluation value that evaluates the spread, and noise is data in which the derived evaluation value is associated with each pixel in the radiographic image.
  • a noise map generation unit that generates a map, and a processing unit that inputs the radiographic image and the noise map to a trained model constructed in advance by machine learning and executes image processing to remove noise from the radiographic image.
  • a radiographic image processing program includes a processor, an image acquiring unit that acquires a radiographic image obtained by irradiating an object with radiation and capturing the radiation that has passed through the object, pixel values and noise Evaluation values are derived from the pixel values of each pixel in the radiographic image based on relationship data that expresses the relationship with the evaluation values that evaluate the spread of values, and the derived evaluation values are associated with each pixel in the radiographic image
  • a certain noise map generation unit and a processing unit that inputs a radiographic image and a noise map to a learned model constructed in advance by machine learning and performs image processing for removing noise from the radiographic image.
  • a radiographic image processing system includes the radiation module described above, a source that irradiates a target with radiation, and an imaging device that captures radiation transmitted through the target to obtain a radiation image. And prepare.
  • the evaluation value is derived from the pixel value of each image of the radiographic image based on the relationship data representing the relationship between the pixel value and the evaluation value that evaluates the spread of the noise value.
  • a noise map which is data in which the derived evaluation values are associated with each pixel of the radiographic image, is generated.
  • the radiographic image and the noise map are input to a learned model constructed in advance by machine learning, and image processing is performed to remove noise from the radiographic image.
  • noise in each pixel of the radiographic image is removed by machine learning, taking into consideration the spread of noise values evaluated from the pixel values of each pixel of the radiographic image. Accordingly, noise removal corresponding to the relationship between the pixel value and the spread of noise in the radiographic image can be realized using the learned model. As a result, noise in radiographic images can be effectively removed.
  • FIG. 1 is a schematic configuration diagram of an image acquisition device 1 according to an embodiment
  • FIG. 2 is a block diagram showing an example of a hardware configuration of a control device 20 of FIG. 1
  • FIG. 2 is a block diagram showing a functional configuration of a control device 20 of FIG. 1
  • FIG. 4 is a diagram showing an example of an X-ray image acquired by an image acquiring unit 203 in FIG. 3
  • FIG. 4 is a diagram showing an example of noise standard deviation map generation by the noise map generation unit 204 of FIG. 3.
  • FIG. FIG. 4 is a diagram showing an example of input/output data of a trained model 207 of FIG. 3
  • FIG. FIG. 10 is a diagram showing an example of a training image, which is one of training data used to build a trained model 207.
  • FIG. 4 is a flowchart showing a procedure of observation processing by the image acquisition device 1;
  • 3A and 3B are diagrams showing examples of X-ray images before and after noise removal processing acquired by the image acquisition device 1;
  • FIG. 3A and 3B are diagrams showing examples of X-ray images before and after noise removal processing acquired by the image acquisition device 1;
  • FIG. 11 is a block diagram showing a functional configuration of a control device 20A according to a modified example of the present disclosure;
  • 10 is a flowchart showing a procedure of observation processing by the image acquisition device 1 according to the modified example of the present disclosure;
  • It is a graph which shows an example of the simulation calculation result of the energy spectrum of the transmission X-ray by 202 A of calculation parts of FIG.
  • 12 is a chart showing an example of simulation calculation results of the relationship between the thickness of the target object and the average energy and transmittance by the calculation unit 202A of FIG. 11; 12 is a graph showing an example of a simulation calculation result of the relationship between the thickness of the object and the X-ray transmittance by the calculation unit 202A of FIG. 11; 12 is a graph showing an example of a simulation calculation result of the relationship between the thickness of the object and the average energy of transmitted X-rays by the calculation unit 202A of FIG. 11; 11. It is a graph which shows an example of the simulation calculation result of the relationship between the pixel value of an X-ray image, and average energy by 202 A of calculation parts of FIG.
  • FIG. 7 is a graph showing an example of a simulation calculation result of the relationship between the pixel value of an X-ray image and the standard deviation of the noise value
  • 12 is a graph showing an example of the relationship between the pixel value and the standard deviation of the noise value when the material of the object changes, which is derived by the calculation unit 202A of FIG. 11
  • FIG. 10 is a block diagram showing the functional configuration of a control device 20B according to the second embodiment of the present disclosure
  • FIG. 9 is a flowchart showing procedures of observation processing by the image acquisition device 1 according to the second embodiment of the present disclosure
  • 21 is a diagram showing an example of noise standard deviation map generation by the noise map generation unit 204B of FIG. 20;
  • FIG. 11 is a perspective view showing an example of the structure of a jig used for imaging in the image acquisition device 1 according to the second embodiment;
  • FIG. 24 is a diagram showing an example of a captured image of the jig in FIG. 23;
  • FIG. 11 is a schematic configuration diagram of an image acquisition device 1C according to a third embodiment;
  • FIG. 11 is a block diagram showing the functional configuration of a control device 20C according to a third embodiment;
  • FIG. 26 is a block diagram showing the functional configuration of the X-ray detection camera 10C of FIG. 25;
  • FIG. 4 is a diagram showing an example of an X-ray image generated by simulation calculation; 7 is a graph showing an example of noise distribution used for generating teacher data;
  • FIG. 1 is a configuration diagram of an image acquisition device 1, which is a radiation image processing system according to the first embodiment.
  • the image acquisition device 1 irradiates an object F transported in the transport direction TD with X-rays (radiation), and determines the object F based on the X-rays transmitted through the object F. It is a device that acquires a captured X-ray image (radiation image).
  • the image acquisition device 1 uses an X-ray image to perform foreign matter inspection, weight inspection, product inspection, etc. of the object F, and is used for food inspection, baggage inspection, circuit board inspection, battery inspection, and material inspection. etc.
  • the image acquisition device 1 includes a belt conveyor (conveying means) 60, an X-ray irradiator (radiation source) 50, an X-ray detection camera (imaging device) 10, a control device (radiation image processing module) 20, and a display. It comprises a device 30 and an input device 40 for performing various inputs.
  • the radiographic image in the embodiments of the present disclosure is not limited to an X-ray image, and includes an image obtained by radiation other than X-rays such as ⁇ -rays.
  • the belt conveyor 60 has a belt portion on which the object F is placed, and by moving the belt portion in the conveying direction TD, the object F is conveyed in the conveying direction TD at a predetermined conveying speed.
  • the conveying speed of the object F is, for example, 48 m/min.
  • the belt conveyor 60 can change the conveying speed to, for example, 24 m/min or 96 m/min, as required. Further, the belt conveyor 60 can change the height position of the belt portion as appropriate to change the distance between the X-ray irradiator 50 and the object F.
  • the X-ray irradiator 50 is a device that irradiates (outputs) the object F with X-rays as an X-ray source.
  • the X-ray irradiator 50 is a point light source, and radiates X-rays by diffusing them in a predetermined angular range in a predetermined irradiation direction.
  • the X-ray irradiator 50 is arranged so that the X-ray irradiation direction is directed toward the belt conveyor 60 and the diffusing X-rays cover the entire width direction of the object F (the direction intersecting the conveying direction TD). is arranged above the belt conveyor 60 at a predetermined distance from the .
  • the irradiation range of the X-ray irradiator 50 is a predetermined divided range in the length direction of the object F (the direction parallel to the transport direction TD), and the object F is transferred to the belt conveyor 60.
  • the object F is conveyed in the conveying direction TD, so that the entire length of the object F is irradiated with X-rays.
  • the tube voltage and tube current of the X-ray irradiator 50 are set by the controller 20 .
  • the X-ray irradiator 50 irradiates the belt conveyor 60 with X-rays having a predetermined energy and radiation dose according to the set tube voltage and tube current.
  • a filter 51 that transmits a predetermined wavelength range of X-rays is provided near the belt conveyor 60 side of the X-ray irradiator 50 .
  • the filter 51 is not necessarily required and may be omitted as appropriate.
  • the X-ray detection camera 10 detects X-rays that have passed through the object F among the X-rays that have been irradiated to the object F by the X-ray irradiator 50, and outputs a signal based on the X-rays.
  • the X-ray detection camera 10 is a dual-line X-ray camera in which two sets of components for detecting X-rays are arranged. In the image acquisition device 1 according to the first embodiment, X-ray images are generated based on X-rays detected in each line (first line and second line) of the dual-line X-ray camera.
  • a clearer image can be obtained with a smaller amount of X-rays than when an X-ray image is generated based on X-rays detected in one line. (high brightness) image can be acquired.
  • the X-ray detection camera 10 includes a filter 19, scintillators 11a and 11b, line scan cameras 12a and 12b, a sensor control section 13, amplifiers 14a and 14b, AD converters 15a and 15b, and correction circuits 16a and 16b. , output interfaces 17 a and 17 b, and an amplifier control section 18 .
  • the scintillator 11a, the line scan camera 12a, the amplifier 14a, the AD converter 15a, the correction circuit 16a, and the output interface 17a are electrically connected to each other and have a configuration related to the first line.
  • the scintillator 11b, the line scan camera 12b, the amplifier 14b, the AD converter 15b, the correction circuit 16b, and the output interface 17b are electrically connected to each other, and have a configuration related to the second line.
  • the line scan camera 12a for the first line and the line scan camera 12b for the second line are arranged side by side along the transport direction TD.
  • the structure common to a 1st line and a 2nd line is demonstrated as a representative of the structure of a 1st line.
  • the scintillator 11a is fixed on the line scan camera 12a by adhesion or the like, and converts the X-rays transmitted through the object F into scintillation light.
  • the scintillator 11a outputs scintillation light to the line scan camera 12a.
  • the filter 19 transmits a predetermined wavelength range of X-rays toward the scintillator 11a.
  • the filter 19 is not necessarily required and may be omitted as appropriate.
  • the line scan camera 12a detects scintillation light from the scintillator 11a, converts it into electric charge, and outputs it as a detection signal (electrical signal) to the amplifier 14a.
  • the line scan camera 12a has a plurality of line sensors arranged in parallel in a direction intersecting the transport direction TD.
  • the line sensor is, for example, a CCD (Charge Coupled Device) image sensor, a CMOS (Complementary Metal-Oxide Semiconductor) image sensor, or the like, and includes a plurality of photodiodes.
  • the sensor control unit 13 controls the line scan cameras 12a and 12b to repeatedly capture images at predetermined detection intervals so that the line scan cameras 12a and 12b can capture X-rays that have passed through the same area of the object F. .
  • the predetermined detection cycle is, for example, the distance between the line scan cameras 12a and 12b, the speed of the belt conveyor 60, the distance between the X-ray irradiator 50 and the object F on the belt conveyor 60 (FOD (Focus Object Distance: radiation source distance between objects)) and the distance between the X-ray irradiator 50 and the line scan cameras 12a and 12b (FDD (Focus Detector Distance: distance between source sensors)), the period common to the line scan cameras 12a and 12b may be set.
  • the predetermined cycle may be set individually based on the pixel width of the photodiode in the direction orthogonal to the pixel arrangement direction of the line sensor of each of the line scan cameras 12a and 12b.
  • the distance between line scan cameras 12a and 12b, the speed of conveyor belt 60, the distance (FOD) between X-ray irradiator 50 and object F on belt conveyor 60, and the distance between X-ray irradiator 50 and A difference (delay time) in the detection cycle between the line scan cameras 12a and 12b may be specified according to the distance (FDD) to the line scan cameras 12a and 12b, and individual cycles may be set for each.
  • the amplifier 14a amplifies the detection signal with a predetermined set gain to generate an amplified signal, and outputs the amplified signal to the AD converter 15a.
  • the set amplification factor is an amplification factor set by the amplifier controller 18 .
  • the amplifier control unit 18 sets the set amplification factors of the amplifiers 14a and 14b based on predetermined imaging conditions.
  • the AD converter 15a converts the amplified signal (voltage signal) output by the amplifier 14a into a digital signal and outputs it to the correction circuit 16a.
  • the correction circuit 16a performs predetermined correction such as signal amplification on the digital signal, and outputs the corrected digital signal to the output interface 17a.
  • the output interface 17 a outputs the digital signal to the outside of the X-ray detection camera 10 .
  • the AD converter, the correction circuit, and the output interface exist individually in FIG. 1, they may be integrated.
  • the control device 20 is, for example, a computer such as a PC (Personal Computer).
  • the control device 20 generates an X-ray image based on digital signals (amplified signals) output from the X-ray detection camera 10 (more specifically, the output interfaces 17a and 17b).
  • the control device 20 generates one X-ray image by averaging or adding two digital signals output from the output interfaces 17a and 17b.
  • the generated X-ray image is output to the display device 30 after undergoing noise removal processing, which will be described later, and displayed by the display device 30 .
  • the controller 20 also controls the X-ray irradiator 50 , the amplifier controller 18 , and the sensor controller 13 .
  • the control device 20 of the first embodiment is a device provided independently outside the X-ray detection camera 10 , but may be integrated inside the X-ray detection camera 10 .
  • FIG. 2 shows the hardware configuration of the control device 20.
  • the control device 20 physically includes a CPU (Central Processing Unit) 101 and a GPU 105 (Graphic Processing Unit) as processors, a RAM (Random Access Memory) 102 and a ROM (Read Only Memory) 103, a communication module 104, an input/output module 106, etc., which are electrically connected to each other.
  • the control device 20 may include a display, a keyboard, a mouse, a touch panel display, etc. as the input device 40 and the display device 30, and may include a data recording device such as a hard disk drive and a semiconductor memory. Also, the control device 20 may be configured by a plurality of computers.
  • FIG. 3 is a block diagram showing the functional configuration of the control device 20.
  • the control device 20 includes an input unit 201 , a calculation unit 202 , an image acquisition unit 203 , a noise map generation unit 204 , a processing unit 205 and a construction unit 206 .
  • Each functional unit of the control device 20 shown in FIG. 3 controls the CPU 101 and the GPU 105 by loading a program (radiation image processing program of the first embodiment) onto hardware such as the CPU 101, the GPU 105, and the RAM 102.
  • the communication module 104, the input/output module 106, etc. are operated, and data is read from and written to the RAM 102.
  • the CPU 101 and GPU 105 of the control device 20 cause the control device 20 to function as the functional units shown in FIG. 3, and sequentially execute processes corresponding to the radiographic image processing method described later.
  • the CPU 101 and the GPU 105 may be single hardware, or only one of them may be used.
  • the CPU 101 and GPU 105 may be implemented in programmable logic such as FPGA, like soft processors.
  • the RAM and ROM may be stand-alone hardware, or may be built in programmable logic such as FPGA.
  • Various data necessary for executing this computer program and various data generated by executing this computer program are all stored in built-in memories such as ROM 103 and RAM 102, or storage media such as hard disk drives.
  • a built-in memory or a storage medium in the control device 20 stores in advance a trained model 207 that is read by the CPU 101 and the GPU 105 and causes the CPU 101 and the GPU 105 to perform noise removal processing on an X-ray image. (described later).
  • the input unit 201 accepts input of condition information indicating either the condition of the radiation source or the imaging condition when the object F is imaged by irradiating radiation. Specifically, the input unit 201 inputs the operating conditions of the X-ray irradiator (radiation source) 50 when capturing an X-ray image of the object F, or the conditions indicating the imaging conditions of the X-ray detection camera 10, and the like. Input of information is accepted from the user of the image acquisition device 1 . Operating conditions include all or part of tube voltage, target angle, target material, and the like.
  • the condition information indicating the imaging conditions includes the material and thickness of the filters 51 and 19 arranged between the X-ray irradiator 50 and the X-ray detection camera 10, Distance (FDD), type of window material of the X-ray detection camera 10, information on materials and thickness of the scintillators 11a and 11b of the X-ray detection camera 10, X-ray detection camera information (for example, gain setting value, circuit noise value, saturation charge amount, conversion coefficient value (e-/count), camera line rate (Hz) or line speed (m/min)), information on the object F, and the like, all or part of them.
  • FDD Distance
  • type of window material of the X-ray detection camera 10 information on materials and thickness of the scintillators 11a and 11b of the X-ray detection camera 10
  • X-ray detection camera information for example, gain setting value, circuit noise value, saturation charge amount, conversion coefficient value (e-/count), camera line rate (Hz) or line speed (m/min)
  • the input unit 201 may receive the input of the condition information as a direct input of information such as numerical values, or as a selective input of information such as numerical values set in advance in the internal memory.
  • the input unit 201 accepts input of the above condition information from the user, but may acquire part of the condition information (tube voltage, etc.) according to the detection result of the control state by the control device 20 .
  • the calculation unit 202 calculates the average energy of X-rays (radiation) transmitted through the object F based on the condition information.
  • the condition information includes any of the tube voltage of the source, information about the object F, filter information of the camera used to image the object F, scintillator information of the camera, and filter information of the X-ray source. At least one is included.
  • the calculation unit 202 uses the image acquisition device 1 to transmit X-rays through the object F to be detected by the X-ray detection camera 10. Calculate the average energy value.
  • the calculation unit 202 calculates the tube voltage, the target angle, the material of the target, the material and thickness of the filters 51 and 19 and their presence/absence, the type of window material of the X-ray detection camera 10 and their presence/absence, which are included in the condition information. Based on information such as the material and thickness of the scintillators 11a and 11b of the X-ray detection camera 10, the X-ray spectrum detected by the X-ray detection camera 10 is calculated using, for example, a known Tucker approximation formula. do.
  • the calculation unit 202 further calculates the spectral intensity integral value and the photon number integral value from the X-ray spectrum, and divides the spectral intensity integral value by the photon number integral value to obtain the average energy value of the X-ray. calculate.
  • the calculation unit 202 when the calculation unit 202 specifies that the target is tungsten and the target angle is 25°, Em: kinetic energy at the time of electron target collision, T: electron kinetic energy in the target, A: proportionality constant determined by the atomic number of the target material , ⁇ : density of target, ⁇ (E): linear attenuation coefficient of target material, B: Z and T function that changes slowly, C: Thomson-Whiddington constant, ⁇ : target angle, c: speed of light in vacuum , can be determined. Furthermore, the calculation unit 202 can calculate the irradiation X-ray spectrum by calculating the following formula (1) based on them.
  • Em can be determined from information on the tube voltage
  • A, ⁇ , and ⁇ (E) can be determined from information on the material of the object F
  • ⁇ can be determined from information on the angle of the object F.
  • the calculation unit 202 can calculate the X-ray energy spectrum transmitted through the filter and the object F and absorbed by the scintillator using the following X-ray attenuation formula (2).
  • ⁇ is the attenuation coefficient of the object F, filter, scintillator, etc.
  • x is the thickness of the object F, filter, scintillator, etc.
  • FIG. ⁇ can be determined from information on the material of the object F, the filter, and the scintillator
  • x can be determined from information on the thickness of the object F, the filter, and the scintillator.
  • the X-ray photon number spectrum is obtained by dividing this X-ray energy spectrum by the energy of each X-ray.
  • the calculation unit 202 calculates the average energy of X-rays using the following formula (3) by dividing the integrated value of the energy intensity by the integrated value of the number of photons.
  • Average energy E spectral intensity integral value/photon number integral value (3)
  • the calculator 202 calculates the average energy of X-rays.
  • a known approximation formula by Kramers or Birch et al. may be used.
  • the image acquisition unit 203 acquires a radiographic image in which the object F is irradiated with radiation and the radiation transmitted through the object F is captured. Specifically, the image acquisition unit 203 generates an X-ray image based on digital signals (amplified signals) output from the X-ray detection camera 10 (more specifically, the output interfaces 17a and 17b). The image acquisition unit 203 generates one X-ray image by averaging or adding two digital signals output from the output interfaces 17a and 17b.
  • FIG. 4 is a diagram showing an example of an X-ray image acquired by the image acquiring unit 203. As shown in FIG.
  • the noise map generation unit 204 derives an evaluation value from the pixel value of each pixel of the radiographic image based on relational data representing the relationship between the pixel value and the evaluation value that evaluates the spread of the noise value, and calculates the evaluation value of each pixel of the radiographic image.
  • a noise map is generated as data in which the evaluation values derived from are associated with each other.
  • the noise map generator 204 derives an evaluation value from the average energy of the radiation transmitted through the object F and the pixel value of each pixel of the radiographic image.
  • the noise map generation unit 204 uses the relational expression (relationship data) between the pixel value and the standard deviation of the noise value (evaluation value obtained by evaluating the spread of the noise value).
  • a standard deviation of the noise value is derived from the average energy of X-rays and the pixel value of each pixel of the X-ray image (radiation image) acquired by the image acquisition unit 203 .
  • the noise map generator 204 generates a noise standard deviation map (noise map) by associating each pixel of the X-ray image with the derived standard deviation of the noise value.
  • the variable Noise is the standard deviation of the noise value
  • the variable Signal is the signal value of the pixel (pixel value)
  • the constant F is the noise factor
  • the constant M is the multiplication factor by the scintillator
  • the constant C is Coupling efficiency between the line scan camera 12a and the scintillator 11a, or between the line scan camera 12b and the scintillator 11b in the X-ray detection camera 10
  • the constant Q is the quantum efficiency of the line scan camera 12a or the line scan camera 12b.
  • the constant cf is a conversion coefficient for converting the signal value of a pixel into an electric charge in the line scan camera 12a or the line scan camera 12b
  • the variable Em is the average energy of X-rays
  • the constant D is the dark current generated by thermal noise in the image sensor.
  • the noise and the constant R are information representing the readout noise in the line scan camera 12a or the line scan camera 12b, respectively.
  • the numerical value of the average energy calculated by is substituted.
  • the variable Noise calculated using the above equation (4) is obtained by the noise map generation unit 204 as a numerical value of the standard deviation of the noise values.
  • other parameters including the average energy may be obtained by receiving inputs through the input unit 201, or may be set in advance.
  • FIG. 5 is a diagram showing an example of noise standard deviation map generation by the noise map generation unit 204.
  • the noise map generating unit 204 uses the relational expression (4) between the pixel values and the standard deviation of the noise values to substitute various pixel values for the variable Signal and obtain the correspondence relationship between the pixel values and the variable Noise.
  • a relationship graph G3 representing the correspondence relationship between the pixel value and the standard deviation of the noise value is derived.
  • the noise map generation unit 204 derives relationship data G2 representing the correspondence relationship between each pixel position and pixel value from the X-ray image G1 acquired by the image acquisition unit 203 .
  • the noise map generation unit 204 applies the correspondence shown in the relationship graph G3 to each pixel value in the relationship data G2, thereby deriving the standard deviation of the noise values corresponding to the pixels at each pixel position in the X-ray image. do.
  • the noise map generation unit 204 associates the derived noise standard deviation with each pixel position, and derives relationship data G4 indicating the correspondence relationship between each pixel position and the noise standard deviation.
  • the noise map generator 204 then generates a noise standard deviation map G5 based on the derived relational data G4.
  • the processing unit 205 inputs the radiographic image and the noise map to the learned model 207 constructed in advance by machine learning, and executes image processing to remove noise from the radiographic image. That is, as shown in FIG. 6, the processing unit 205 acquires a learned model 207 (described later) constructed by the construction unit 206 from the built-in memory or storage medium within the control device 20 . The processing unit 205 inputs the X-ray image G ⁇ b>1 acquired by the image acquisition unit 203 and the noise standard deviation map G ⁇ b>5 generated by the noise map generation unit 204 to the trained model 207 . Accordingly, the processing unit 205 uses the learned model 207 to perform image processing for removing noise from the X-ray image G1, thereby generating an output image G6. Then, the processing unit 205 outputs the generated output image G6 to the display device 30 or the like.
  • the construction unit 206 generates a training image that is a radiation image, a noise map that is generated from the training image based on the relational expression between the pixel value and the standard deviation of the noise value, and noise that is data obtained by removing noise from the training image. Using the removed image data as training data, a trained model 207 that outputs noise-removed image data based on the training image and the noise map is constructed by machine learning.
  • the construction unit 206 stores the constructed learned model 207 in the built-in memory or storage medium in the control device 20 .
  • Machine learning includes supervised learning, unsupervised learning, and reinforcement learning, including deep learning, neural network learning, and the like.
  • a two-dimensional convolutional neural network described in Kai Zhang et al.'s paper "Beyond a Gaussian Denoiser: Residual Learning of Deep CNN for Image Denoising” is employed as an example of a deep learning algorithm.
  • the learned model 207 may be generated by an external computer or the like and downloaded to the control device 20 instead of being constructed by the construction unit 206 .
  • Radiographic images used for machine learning include radiographic images obtained by imaging known structures or images obtained by reproducing the radiographic images.
  • FIG. 7 is an example of a training image, which is one of the training data used to build the trained model 207.
  • X-ray images of patterns of various thicknesses, various materials, and various resolutions can be used as training images.
  • the example shown in FIG. 7 is a training image G7 generated for chicken.
  • the training image G7 may be an X-ray image actually generated for a plurality of types of known structures using the image acquisition device 1, or may be an image generated by simulation calculation. .
  • the X-ray image may be acquired using a device different from the image acquisition device 1 .
  • the constructing unit 206 As preprocessing for performing machine learning, the constructing unit 206 generates an evaluation value from the pixel value of each pixel of the radiographic image based on relational data representing the relationship between the pixel value and the evaluation value obtained by evaluating the spread of the noise value. A noise map is generated as data in which each pixel of the radiographic image is associated with the derived evaluation value.
  • the construction unit 206 acquires training images generated by actual imaging or simulation calculations from the image acquisition unit 203 or the like. Then, the construction unit 206 sets, for example, the operating conditions of the X-ray irradiator 50 of the image acquisition device 1 or the imaging conditions of the image acquisition device 1 .
  • the construction unit 206 sets operating conditions or imaging conditions of the X-ray irradiator 50 at the time of simulation calculation.
  • the constructing unit 206 uses the same technique as the calculating unit 202 to calculate the average energy of X-rays based on the above operating conditions or imaging conditions. Further, the constructing unit 206 generates a noise standard deviation map based on the average energy of X-rays and the training images using a method similar to the method by the noise map generating unit 204 as shown in FIG.
  • the preprocessing method of the machine learning method derives an evaluation value from the pixel value of each pixel of the radiographic image based on the relationship data representing the relationship between the pixel value and the evaluation value that evaluates the spread of the noise value
  • a noise map generation step is provided for generating a noise map, which is data in which the derived evaluation values are associated with each pixel of the image.
  • the construction unit 206 constructs a trained model 207 by machine learning using training images, a noise map generated from the training images, and noise-removed image data, which is data from which noise has been removed in advance from the training images, as training data. do. Specifically, the construction unit 206 acquires in advance noise-removed image data obtained by removing noise from the training image. When the training images are X-ray images generated by simulation calculation, the constructing unit 206 uses the images before noise is added in the process of generating the training images as the noise-removed image data.
  • the constructing unit 206 selects a mean value filter or a median value from the X-ray images.
  • An image from which noise has been removed using image processing such as a filter, bilateral filter, or NLM filter is used as noise-removed image data.
  • the constructing unit 206 constructs a trained model 207 that outputs noise-removed image data based on the training images and the noise standard deviation map by performing training by machine learning.
  • FIG. 8 is a flow chart showing a procedure of observation processing by the image acquisition device 1. As shown in FIG.
  • the construction unit 206 uses the training images, the noise standard deviation map generated from the training images based on the relational expression, and the noise-removed image data as training data.
  • a trained model 207 that outputs the removed image data is constructed by machine learning (step S100).
  • the input unit 201 receives input of condition information indicating operating conditions of the X-ray irradiator 50 or imaging conditions of the X-ray detection camera 10 from the operator (user) of the image acquisition apparatus 1 (step S101). .
  • the calculation unit 202 calculates the average energy value of the X-rays detected by the X-ray detection camera 10 (step S102).
  • the object F is set in the image acquisition device 1, the object F is imaged, and the X-ray image of the object F is acquired by the control device 20 (step S103). Furthermore, the control device 20 derives the standard deviation of the noise value from the average energy of the X-rays and the pixel value of each pixel of the X-ray image based on the relational expression between the pixel value and the standard deviation of the noise value. A noise standard deviation map is generated by associating the obtained noise standard deviation with each pixel value (step S104).
  • the processing unit 205 inputs the X-ray image and the noise standard deviation map of the object F to the trained model 207 constructed and stored in advance, and performs noise removal processing on the X-ray image (step S105). Furthermore, an output image, which is an X-ray image subjected to noise removal processing by the processing unit 205 , is output to the display device 30 . (Step S106).
  • the standard deviation of the noise value is derived from the pixel value of each image of the X-ray image using the relational expression of the standard deviation of the pixel value and the noise value.
  • a noise standard deviation map is generated, which is data in which the standard deviations of the derived noise values are associated with the pixels.
  • the X-ray image and the noise standard deviation map are input to the trained model 207 constructed in advance by machine learning, and image processing is performed to remove noise from the X-ray image.
  • the noise in each pixel of the X-ray image is removed by machine learning, taking into consideration the standard deviation of the noise value derived from the pixel value of each pixel of the X-ray image.
  • noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value in the X-ray image can be realized.
  • noise in the X-ray image can be effectively removed.
  • a trained model must be constructed for each noise condition such as the average energy of X-rays, the gain of the X-ray detection camera, and the type of X-ray camera, and a huge number of models must be constructed. It is necessary to generate a trained model of , and it may take a lot of time for construction. As an example, when there are 10 X-ray average energies, 8 X-ray detection camera gains, and 3 product types, 240 trained models are required. If it takes one day per model, it will take 240 days for machine learning.
  • the present embodiment by generating a noise map from an X-ray image and using the noise map as input data for machine learning, it is possible to reduce noise conditions that require generation of a trained model. , the learning time for building the trained model 207 is greatly reduced.
  • FIG. 9 and 10 respectively show examples of X-ray images before and after noise removal processing acquired by the image acquisition device 1.
  • FIG. FIG. 9 is an X-ray image acquired under the same operating conditions of the X-ray irradiator 50 and imaging conditions of the X-ray detection camera 10 as the training data used to construct the trained model 207.
  • G8 is a measured image
  • X-ray image G9 is an image subjected to noise removal processing with a trained model trained under the same imaging conditions as the comparative example
  • X-ray image G10 is an image obtained by the control device 20 in the first embodiment.
  • Each image is denoised using a noise standard deviation map.
  • the standard deviations of the noise values in X-ray images G8, G9 and G10 are 14.3, 3.4 and 3.7 respectively.
  • FIG. 10 is an X-ray image acquired when the operating conditions of the X-ray irradiator 50 or the imaging conditions of the X-ray detection camera 10 are different from the training data used to construct the trained model 207.
  • the X-ray image G11 is a measured image
  • the X-ray image G12 is an image subjected to noise removal processing with a trained model trained under conditions different from the imaging conditions of the comparative example
  • the X-ray image G13 is an image obtained in the first embodiment.
  • the X-ray images subjected to noise removal processing using the noise standard deviation map by the control device 20 are respectively shown.
  • the standard deviations of the noise values in X-ray images G11, G12 and G13 are 3.5, 2.0 and 0.9 respectively.
  • the training data used to build the learned model 207 when the operating conditions of the X-ray irradiator 50 or the imaging conditions of the X-ray detection camera 10 are the same as the training data used to build the learned model 207, X In the line image G9, the standard deviation of noise values is sufficiently reduced compared to the X-ray image G8 before noise removal processing.
  • the trained model in the comparative example can output an X-ray image from which noise is sufficiently removed.
  • FIG. 9 when the operating conditions of the X-ray irradiator 50 or the imaging conditions of the X-ray detection camera 10 are the same as the training data used to build the learned model 207, X In the line image G9, the standard deviation of noise values is sufficiently reduced compared to the X-ray image G8 before noise removal processing.
  • the trained model in the comparative example can output an X-ray image from which noise is sufficiently removed.
  • FIG. 9 when the operating conditions of the X-ray irradiator 50 or the imaging conditions of the X-
  • the trained model in the comparative example cannot output an X-ray image from which noise is sufficiently removed under different conditions at the time of training and at the time of imaging.
  • the learned model 207 is constructed in consideration of changes in the operating conditions of the X-ray irradiator 50 or the imaging conditions of the X-ray detection camera 10 during X-ray image measurement. be.
  • the standard deviation of the noise values is sufficiently reduced compared to the respective X-ray images G8 and 11 before noise removal processing. Therefore, according to the first embodiment, sufficient noise removal corresponding to changes in operating conditions of the X-ray irradiator 50 or imaging conditions of the X-ray detection camera 10 is realized. This allows a single trained model 207 to be used to effectively remove noise in X-ray images.
  • X-ray images contain noise derived from X-ray generation. It is conceivable to increase the X-ray dose in order to improve the SN ratio of the X-ray image. However, it is difficult to achieve both an improvement in the SN ratio and a long life. In addition, since the amount of heat generated increases as the dose of X-rays increases, it may be necessary to take measures to dissipate the increased heat. In the first embodiment, since it is not necessary to increase the X-ray dose, it is possible to achieve both an improvement in the SN ratio and a longer life, and to omit heat dissipation measures.
  • the control device 20 of the first embodiment also has a function of deriving the standard deviation of the noise value from the average energy of the X-rays that have passed through the object F and the pixel value of each pixel of the X-ray image.
  • the average energy of the X-rays transmitted through the object F is taken into consideration, and the standard deviation of the noise value in the pixel value of each pixel of the X-ray image is derived. Noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value can be realized.
  • the difference in average energy is reflected in the noise standard deviation map, and the noise standard deviation map is input to the trained model.
  • the number of models is one. This greatly reduces the learning time for constructing the trained model 207 .
  • control device 20 of the first embodiment receives input of condition information indicating either the operating condition of the X-ray irradiator 50 or the imaging condition of the X-ray detection camera 10, and calculates the average energy based on the condition information. It has a function of calculating.
  • condition information includes the tube voltage of the X-ray irradiator 50, information on the object F, information on filters provided in the X-ray irradiator 50, information on filters provided in the X-ray detection camera 10, and information on filters provided in the X-ray detection camera 10. information on the scintillator.
  • the average energy of the X-rays that pass through the object F can be calculated with high accuracy, so noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value can be achieved. As a result, noise in the X-ray image can be removed more effectively.
  • the spread of noise values is evaluated as the standard deviation of the noise values.
  • the spread of the noise value in the pixel value of each pixel of the X-ray image is evaluated more precisely, so noise removal corresponding to the relationship between the pixel value and the noise can be realized.
  • noise in the X-ray image can be removed more effectively.
  • control device 20 of the first embodiment includes a training image that is an X-ray image, a noise standard deviation map generated from the training image based on the relational expression between the pixel value and the standard deviation of the noise value, and the training image A function to build a trained model 207 that outputs noise-removed image data based on a training image and a noise standard deviation map by machine learning, using noise-removed image data, which is data from which noise has been removed, as training data.
  • the trained model 207 used for noise removal of X-ray images is constructed by machine learning using training data.
  • noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value can be realized. .
  • noise in the X-ray image can be removed more effectively.
  • noise standard deviation map that is training data for machine learning, based on the relational expression between the pixel value and the standard deviation of the noise value, A noise standard deviation is derived from the pixel value of each pixel, and a noise standard deviation map, which is data in which the derived noise standard deviation is associated with each pixel of the training image, is generated.
  • the noise standard deviation map which is training data for machine learning, corresponds to the relational expression between the pixel value and the standard deviation of the noise value.
  • FIG. 11 is a block diagram showing the functional configuration of the control device 20A in the modified example of the first embodiment.
  • the control device 20A has a function of deriving the average energy of X-rays from the pixel values of the X-ray image in the calculation unit 202A, and calculates the X-ray image in the noise map generation unit 204A. , and the average energy of X-rays derived from the X-ray image, it has a function of deriving a noise standard deviation map.
  • FIG. 12 is a flow chart showing the procedure of observation processing by the image acquisition device 1 including the control device 20A of FIG. As shown in FIG.
  • step S103 of the control device 20 according to the first embodiment shown in FIG. 8 is performed immediately after step S100. Then, in the control device 20A, the processes shown in S102A and S104A are replaced with the processes in steps S102 and S104 of the control device 20 and executed.
  • the calculation unit 202A calculates the average energy from the pixel value of each pixel of the radiographic image (step S102A). Specifically, the calculation unit 202A derives in advance the relationship between the pixel value and the average energy for each piece of condition information by simulation calculation of the X-ray spectrum or the like. The calculation unit 202A acquires condition information including at least the tube voltage acquired by the input unit 201 and information on the scintillator included in the X-ray detection camera 10 . Then, based on the condition information, the calculation unit 202A selects the relationship corresponding to the condition information from the previously derived relationships between the pixel values and the average energies. Further, the calculation unit 202A derives the average energy of each pixel from the pixel value of each pixel of the X-ray image acquired by the image acquisition unit 203 based on the selected relationship.
  • the calculator 202A creates a graph G18 representing the relationship between the thickness of the object F and the X-ray transmittance, and the relationship between the thickness of the object F and the average energy of X-rays.
  • Graph G19 is derived.
  • the calculation unit 202A calculates the target Energy spectra G14 to G17 of X-rays transmitted when the thickness of the object F is varied are calculated by simulation calculation.
  • FIG. 13 is a graph showing an example of a simulation calculation result of the energy spectrum of X-rays transmitted through the object F by the calculator 202A.
  • energy spectra G14 to G17 of transmitted X-rays are illustrated when the simulation calculation is performed by gradually increasing the thickness of the object F made of water. Furthermore, based on the calculated energy spectra G14 to G17, the calculation unit 202A calculates the average energy of X-rays transmitted when the thickness of the object F is varied. In addition to the simulation calculation, the calculation unit 202A obtains the relationship between the thickness of the object F and the average energy based on an X-ray image obtained by imaging a structure with a known thickness. good too.
  • the calculation unit 202A also derives the relationship between the thickness of the object F and the X-ray transmittance based on the above simulation results.
  • FIG. 14 is a chart showing an example of the relationship between the thickness of the object F and the average energy and transmittance derived by the calculator 202A. As shown in FIG. 14, the average energy of transmitted X-rays and the X-ray transmittance are derived corresponding to each of the energy spectra G14 to G17 calculated for each thickness of the object F.
  • the calculation unit 202A derives a graph G18 showing the relationship between the thickness of the object F and the X-ray transmittance from the X-ray transmittance derived for the object F having various thicknesses.
  • FIG. 15 is a graph showing the relationship between the thickness of the object F and the X-ray transmittance of the object F derived by the calculator 202A.
  • the calculation unit 202A derives a graph G19 showing the relationship between the thickness of the object F and the average energy of X-rays from the average energy of X-rays derived for objects F of various thicknesses.
  • FIG. 16 is a graph showing the relationship between the thickness of the object F and the average energy of X-rays that pass through the object F, derived by the calculator 202A.
  • the calculation unit 202A creates a graph G20 showing the relationship between the pixel values of the X-ray image and the average energy as shown in FIG. It is derived for each of various condition information.
  • FIG. 17 is a graph showing the relationship between the pixel value of the X-ray image and the average energy derived by the calculator 202A.
  • the calculation unit 202A derives the pixel value I0 of the X-ray transmission image when the object F does not exist based on the condition information. Then, the calculation unit 202A sets the pixel value I of the X-ray image when the object F exists, and calculates I/ I0 which is the X-ray transmittance.
  • the calculation unit 202A calculates the thickness of the object F from the calculated X-ray transmittance I/I 0 based on the graph G18 of the thickness of the object F and the X-ray transmittance of the object F. to derive Finally, the calculation unit 202A calculates the average energy of transmitted X-rays corresponding to the thickness based on the derived thickness of the object F and the graph G19 of the thickness of the object F and the average energy of transmitted X-rays.
  • the calculation unit 202A performs the above derivation for each of various condition information while varying the pixel value I of the X-ray image, thereby obtaining the pixel value of the X-ray image and the average energy of transmitted X-rays.
  • a graph G20 showing the relationship between is derived for each condition information.
  • the calculation unit 202A determines that the thickness corresponding to the X-ray transmittance of 0.1 is 30 mm based on the graph G18 showing the relationship between the thickness of the object F and the X-ray transmittance of the object F.
  • the calculation unit 202A derives that the average energy corresponding to the pixel value 500 is 27 keV based on the graph G19 showing the relationship between the thickness of the object F and the average energy of transmitted X-rays. Finally, the calculation unit 202A repeats the derivation of the average energy of X-rays for each pixel value, and derives a graph G20 showing the relationship between the pixel value of the X-ray image and the average energy.
  • the calculation unit 202A selects the graph G20 corresponding to the condition information acquired by the input unit 201 from among the plurality of graphs G20 derived in advance by the above procedure.
  • the calculation unit 202A derives the average energy of transmitted X-rays corresponding to the pixel value of each pixel of the X-ray image acquired by the image acquisition unit 203 based on the selected graph G20.
  • the calculation unit 202A does not derive the relationship between the pixel value and the average energy of X-rays for each condition information in advance, but from the condition information acquired by the input unit 201 and the pixel value of each pixel of the X-ray image. , graphs G18 and G19 may be referred to derive the average energy of X-rays. Specifically, the calculation unit 202A derives the pixel value I0 of the X-ray image when the object does not exist based on the condition information. Then, the calculation unit 202A calculates the transmittance by obtaining the ratio of the pixel value I of each pixel of the X-ray image acquired by the image acquisition unit 203 to the pixel value I0 .
  • the calculator 202A derives the thickness based on the graph G18 showing the relationship between the thickness and the X-ray transmittance and the calculated transmittance. Then, the calculation unit 202A derives the average energy for each pixel value of each pixel of the X-ray image by deriving the average energy based on the graph G19 showing the relationship between the thickness and the average energy and the derived thickness. do.
  • the noise map generation unit 204A generates a noise standard deviation map from the X-ray image acquired by the image acquisition unit 203 and the average energy of X-rays corresponding to each pixel of the X-ray image derived by the calculation unit 202A. (step S104A). Specifically, the noise map generation unit 204A uses the pixel value of each pixel of the X-ray image acquired by the image acquisition unit 203 and the average energy derived for each pixel by the calculation unit 202A as the relational expression (4 ) to derive the standard deviation of the noise value for each pixel considering the thickness of the object. The noise map generation unit 204A generates the standard deviation of the noise values corresponding to each pixel of the X-ray image as a noise standard deviation map.
  • FIG. 18 is a graph showing an example of the relationship between pixel values and standard deviations of noise values.
  • This graph shows the relationship between the standard deviation of the noise values derived from the pixel values of the X-ray image and the pixel values of the X-ray image by the calculation unit 202A and the noise map generation unit 204A according to this modification.
  • the standard deviation of the noise value is derived in consideration of the thickness of the object. Therefore, as the pixel value increases, the thickness of the object becomes smaller and the average energy in the pixel decreases. Therefore, as can be deduced from the relational expression (4), the change in the standard deviation of the noise value when the pixel value increases differs between the first embodiment and this modified example.
  • the graph G22 of this modified example has a smaller degree of increase in the standard deviation of the noise value when the pixel value increases than the graph G21 of the first embodiment.
  • the control device 20A of the modified example of the first embodiment calculates the average energy from the pixel value of each pixel of the X-ray image.
  • the average energy differs greatly for each object, and noise cannot be sufficiently removed from the X-ray image.
  • the average energy of the X-rays that pass through the object F is calculated for each pixel value of each pixel of the X-ray image. Noise removal corresponding to the relationship between the pixel value of each pixel and noise can be realized. As a result, noise in the X-ray image can be effectively removed.
  • the control device 20A derives the average energy from the pixel values of the X-ray image using the graph G20 derived for each of various condition information.
  • the average energy may be derived from the pixel values while ignoring the difference in the material of the object F.
  • FIG. 19 is a graph showing the relationship between the pixel value of the X-ray image and the standard deviation of the noise value derived by the calculator 202A.
  • the relationship is derived taking into consideration the change in the material of the object F as condition information. shows an example of derivation when the material is copper.
  • the control device 20A can accurately derive the average energy from the pixel values of the X-ray image even if the difference in the material of the object F as the condition information is ignored. Even in such a case, according to the control device 20A of this modified example, noise removal corresponding to the relationship between the pixel value and the standard deviation of noise can be realized. As a result, noise in the X-ray image can be removed more effectively.
  • FIG. 20 is a block diagram showing the functional configuration of the control device 20B in the second embodiment.
  • the control device 20B has a function of acquiring an X-ray image of the jig in the image acquisition unit 203B and an X-ray image of the jig in the noise map generation unit 204B. The difference is that it has a function of deriving a graph showing the relationship between the pixel value and the standard deviation of the noise value.
  • FIG. 21 is a flow chart showing the procedure of observation processing by the image acquisition device 1 including the control device 20B of FIG. As shown in FIG. 21, in the control device 20B according to the second embodiment, the processes shown in steps S201 and S202 are performed by the control device 20 according to the first embodiment shown in FIG. Executed instead of processing.
  • the image acquisition unit 203B acquires a radiographic image of the jig in which the jig is irradiated with radiation and the radiation transmitted through the jig is imaged (step S201). Specifically, the image acquisition unit 203B acquires an X-ray image captured by irradiating the jig and the object F with X-rays using the image acquisition device 1 . As the jig, a plate member or the like having a known thickness and material is used. That is, the image acquisition unit 203B acquires an X-ray image of the jig captured using the image acquisition device 1 prior to the observation processing of the object F. FIG.
  • the image acquisition unit 203B acquires an X-ray image of the object F imaged using the image acquisition device 1 .
  • the acquisition timing of the X-ray images of the jig and object F is not limited to the above, and may be simultaneous or reverse timing (step S103).
  • the image acquiring unit 203B acquires an X-ray image obtained by irradiating the object F with X-rays and capturing the X-rays transmitted through the object F in the same manner as the image acquiring unit 203B.
  • a jig is set in the image acquisition device 1 and an image of the jig is captured, and the noise map generation unit 204B calculates the relationship between the pixel value and the evaluation value obtained by evaluating the spread of the noise value from the radiographic image of the jig obtained as a result. (step S202). Specifically, the noise map generator 204B derives a noise standard deviation map representing the relationship between the pixel value and the standard deviation of the noise value from the X-ray image of the jig.
  • FIG. 22 is a diagram showing an example of noise standard deviation map generation by the noise map generation unit 204B.
  • the noise map generator 204B derives a relationship graph G27 representing the correspondence relationship between the pixel value and the standard deviation of the noise value from the X-ray image G26 of the jig. Then, similarly to the first embodiment, the noise map generation unit 204B derives relational data G2 representing the correspondence between each pixel position and pixel value from the X-ray image G1 acquired by the image acquisition unit 203B. do. Furthermore, the noise map generation unit 204 applies the correspondence shown in the relationship graph G27 to each pixel in the relationship data G2 to derive the standard deviation of the noise values corresponding to the pixels at each pixel position in the X-ray image. .
  • the noise map generation unit 204 associates the derived noise standard deviation with each pixel position, and derives relationship data G4 indicating the correspondence relationship between each pixel position and the noise standard deviation.
  • the noise map generator 204 then generates a noise standard deviation map G5 based on the derived relational data G4.
  • FIG. 23 shows an example of the structure of a jig used for imaging in the second embodiment.
  • a member P1 whose thickness changes stepwise in one direction can be used.
  • FIG. 24 shows an example of an X-ray image of the jig of FIG.
  • the noise map generation unit 204B derives pixel values (hereinafter referred to as true pixel values) when there is no noise for each step of the jig in the X-ray image G26 of the jig.
  • the noise map generator 204B derives the average value of the pixel values at a certain step of the jig. Then, the noise map generation unit 204B takes the derived average value of the pixel values as the true pixel value at that step. In that step, the noise map generator 204B derives the difference between each pixel value and the true pixel value as a noise value. The noise map generator 204B derives the standard deviation of the noise values from the derived noise values for each pixel value.
  • the noise map generation unit 204B derives the relationship between the true pixel value and the standard deviation of the noise value as a relationship graph G27 between the pixel value and the standard deviation of the noise value. Specifically, the noise map generator 204B derives the true pixel value and the standard deviation of the noise value for each step of the jig. The noise map generation unit 204B plots the derived relationship between the true pixel value and the standard deviation of the noise value on a graph, and draws an approximate curve to obtain the relationship representing the relationship between the pixel value and the standard deviation of the noise value. Graph G27 is derived. Note that exponential approximation, linear approximation, logarithmic approximation, polynomial approximation, exponential approximation, or the like is used for the approximation curve.
  • the control device 20B of the second embodiment generates relational data based on radiographic images obtained by imaging actual jigs. As a result, the optimum relational data for removing noise from the radiation image of the object F can be obtained. As a result, noise in radiographic images can be removed more effectively.
  • the noise map generation unit 204B derives the relationship between the pixel value and the standard deviation of the noise value from the captured image when the tube current or the exposure time is changed in the absence of the object without using a jig. may According to this configuration, since the relational data is generated based on the radiographic image obtained by actually imaging and the noise map is generated, noise removal corresponding to the relationship between the pixel value and the spread of noise can be realized. As a result, noise in radiographic images can be removed more effectively.
  • the image acquisition unit 203B acquires a plurality of radiographic images captured without an object (step S201), and the noise map generation unit 204B generates a , the relationship between the pixel value and the standard deviation of the noise value may be derived (step S202).
  • the plurality of radiographic images are a plurality of images that differ from each other in at least one of the radiation source conditions and the imaging conditions.
  • the image acquisition unit 203B acquires a plurality of X-rays captured using the image acquisition device 1 without the object F prior to observation processing of the object F while changing the tube current or the exposure time. Get an image.
  • the noise map generator 204B derives the true pixel value for each X-ray image, and derives the noise standard deviation based on the true pixel value in the same manner as in the second embodiment. Furthermore, the noise map generator 204B plots the relationship between the true pixel value and the noise standard deviation on a graph and draws an approximation curve to obtain the standard deviation of the pixel value and the noise value in the same manner as in the second embodiment. Derive a relationship graph representing the relationship with the deviation. Finally, the noise map generation unit 204B generates a noise standard deviation map from the X-ray image acquired by the image acquisition unit 203B, based on the derived relationship graph, in the same manner as in the first embodiment. [Third Embodiment]
  • FIG. 25 is a configuration diagram of an image acquisition device 1C, which is a radiation image processing system according to the third embodiment.
  • FIG. 26 is a block diagram showing an example of the functional configuration of a control device 20C according to the third embodiment.
  • An image acquisition apparatus 1C according to the third embodiment has an X-ray detection camera 10C (imaging device) having a two-dimensional sensor 12C and the like, a construction unit 206C and a It differs in that it has a control device 20C that has a trained model 207C and that it does not have a belt conveyor 60 .
  • the image acquisition device 1C uses an X-ray transmission image to perform a foreign matter inspection, a weight inspection, an inspection inspection, etc. of the object F, and is used for food inspection, baggage inspection, board inspection, battery inspection, material inspection, inspection and the like. Furthermore, the use of the image acquisition device 1C includes medical use, dental use, industrial use, and the like. Medical applications include, for example, chest X-ray, mammography, CT (computed tomography), dual energy CT, tomosynsense, and the like. Dental applications include transmissive, panoramic and CT. Industrial applications include non-destructive testing, security and battery testing.
  • the image acquisition device 1C outputs an X-ray image obtained by picking up an X-ray transmission image based on X-rays that pass through an object F in a stationary state.
  • the image acquisition device 1C may have a belt conveyor 60 like the image acquisition device 1 described above, and may be configured to pick up an image of the conveyed object F.
  • FIG. 27 is a block diagram showing the configuration of the X-ray detection camera 10C.
  • the X-ray detection camera 10C has, as shown in FIGS. 25 and 27, a filter 19, a scintillator layer 11C, a two-dimensional sensor 12C, a sensor control section 13C, and an output section 14C.
  • the sensor control section 13C is electrically connected to the two-dimensional sensor 12C, the output section 14C and the control device 20C.
  • the output section 14C is also electrically connected to the two-dimensional sensor 12C and the control device 20C.
  • the scintillator layer 11C is fixed on the two-dimensional sensor 12C by adhesion or the like, and converts the X-rays transmitted through the object F into scintillation light (detailed configuration will be described later).
  • the scintillator layer 11C outputs scintillation light to the two-dimensional sensor 12C.
  • the filter 19 transmits a predetermined wavelength range of X-rays toward the scintillator layer 11C.
  • the two-dimensional sensor 12C detects the scintillation light from the scintillator layer 11C, converts it into electric charge, and outputs it as a detection signal (electrical signal) to the output section 14C.
  • the two-dimensional sensor 12C is, for example, a line sensor or flat panel sensor, and is arranged on the substrate 15C.
  • the two-dimensional sensor 12C has M ⁇ N pixels P 1,1 to P M,N two-dimensionally arranged in M rows and N columns.
  • the M ⁇ N pixels P 1,1 to P M,N are arranged at a constant pitch both in the row direction and the column direction.
  • the pixel P m,n is located at the m-th row and the n-th column.
  • Each of the N pixels P m,1 to P m,N in the m-th row is connected to the sensor control section 13C by the m-th row selection line L V,m .
  • the output terminals of the M pixels P 1,n to P M,n in the n-th column are connected to the output section 14C by the n-th column readout line L 0 ,n .
  • M and N are integers of 2 or more, m is an integer of 1 or more and M or less, and n is an integer of 1 or more and N or less.
  • the output unit 14C outputs a digital value generated based on the amount of charge input via the readout line LO ,n .
  • the output section 14C includes N integration circuits 41(1) to 41(N), N hold circuits 42(1) to 42(N), an AD conversion section 43 and a storage section .
  • Each integrating circuit 41(n) has a common configuration.
  • Each hold circuit 42(n) has a common configuration.
  • Each integration circuit 41(n) accumulates charges input to the input end via any one of the column readout lines LO ,n .
  • Each integration circuit 41(n) outputs a voltage value corresponding to the accumulated charge amount from the output terminal to the hold circuit 42(n).
  • Each of the N integration circuits 41(1) to 41(N) is connected to the sensor control section 13C by a reset wiring LR.
  • Each hold circuit 42(n) has an input connected to the output of the integration circuit 41(n). Each hold circuit 42 (n) holds the voltage value input to the input terminal and outputs the held voltage value to the AD converter 43 from the output terminal.
  • Each of the N hold circuits 42(1) to 42(N) is connected to the sensor control section 13C by hold wiring LH .
  • Each hold circuit 42(n) is connected to the sensor control section 13C by the n-th column selection wiring LH ,n .
  • the AD converter 43 receives the voltage values output from the N hold circuits 42(1) to 42(N), and performs AD conversion processing on the input voltage values (analog values).
  • the AD conversion section 43 outputs a digital value corresponding to the input voltage value to the storage section 44 .
  • the storage unit 44 receives and stores the digital values output from the AD conversion unit 43, and sequentially outputs the stored digital values.
  • the sensor control unit 13C supplies the m-th row selection control signal Vsel(m) to each of the N pixels P m,1 to P m,N in the m-th row via the m-th row selection wiring L V,m . Output.
  • the sensor control section 13C outputs a reset control signal Reset to each of the N integration circuits 41(1) to 41(N) through the reset wiring LR .
  • the sensor control section 13C outputs the hold control signal Hold to each of the N hold circuits 42(1) to 42(N) via the hold wiring LH .
  • the sensor control unit 13C outputs the n-th column selection control signal Hsel(n) to the hold circuit 42(n) via the n-th column selection wiring LH ,n .
  • the sensor control unit 13C also controls AD conversion processing in the AD conversion unit 43 and also controls writing and reading of digital values in the storage unit 44 .
  • the scintillator layer 11C has K ⁇ L pieces (K and L are rectangular scintillator portions Q 1,1 to Q K,L (integers equal to or greater than 1) and separation portions R located between the scintillator portions Q 1,1 to Q K,L are formed.
  • the number L may be 1 or more and N or less, and the number K may be 1 or more and M or less. Furthermore, the number L may be an integer of 1 or more and may be an integer obtained by dividing N by an integer, and the number K may be an integer of 1 or more and may be an integer obtained by dividing M by an integer. good. In this case, blurring due to spread of light can be suppressed according to the interval of the separating portions R of the scintillator layer 11C. Also, the number L may be an integer larger than N, and the number K may be an integer larger than M.
  • the interval between the separating portions R of the scintillator layer 11C is smaller than the interval between the plurality of pixels P 1,1 to P M,N , but the distance between the scintillator layer 11C and the plurality of pixels P 1,1 to P M,N Alignment becomes easier.
  • the K ⁇ L scintillator sections Q 1,1 to Q K,L are made of a scintillator material capable of converting incident X-rays into scintillation light, and these make up the entire pixels P 1,1 to P M,N . placed to cover.
  • M ⁇ N scintillator sections Q 1,1 to Q M,N are arranged so as to entirely cover the corresponding pixels P 1,1 to P M,N .
  • the separating portion R is formed in a mesh shape so as to separate the K ⁇ L scintillator portions Q 1,1 to Q K,L and is made of a material capable of shielding scintillation light.
  • the separating portion R may contain a material that reflects the scintillation light.
  • the separating section R may be made of a material that can shield radiation.
  • a material constituting the scintillator layer 11C and a method for manufacturing the scintillator layer 11C for example, materials and manufacturing methods described in Japanese Patent Application Laid-Open No. 2001-99941 or Japanese Patent Application Laid-Open No. 2003-167060 can be used. can.
  • the material and manufacturing method of the scintillator layer 11C are not limited to those described in the above document.
  • the control device 20C generates an X-ray image based on the digital signal output from the X-ray detection camera 10C (more specifically, the storage section 44 of the output section 14C).
  • the generated X-ray image is output to the display device 30 after undergoing noise removal processing, which will be described later, and displayed by the display device 30 .
  • the controller 20C also controls the X-ray irradiator 50 and the sensor controller 13C.
  • the control device 20C of the third embodiment is a device provided independently outside the X-ray detection camera 10C, but may be integrated inside the X-ray detection camera 10C.
  • FIG. 29 is a flowchart showing a procedure for creating image data, which is teacher data (training images in the first and second embodiments) used for building the trained model 207C by the building unit 206C.
  • the image data (also referred to as teacher image data), which is training data, is created by a computer in the following steps.
  • an image of a structure having a predetermined structure (structure image) is created (step S301).
  • an image of a structure (for example, jig) having a predetermined structure may be created by simulation calculation.
  • a structure image may be created by acquiring an X-ray image of a structure such as a chart having a predetermined structure.
  • a sigma value which is the standard deviation of pixel values, is calculated for one pixel selected from among the plurality of pixels forming the structure image (step S302).
  • the noise distribution is set based on the sigma value obtained in step S302 (step S303).
  • this noise distribution is set so that the probability that the pixel values to which noise is added is greater than the original pixel value is high. is more than 1.2 times the original pixel value (details will be described later).
  • step S302 The processing from step S302 to step S305 is performed for each of a plurality of pixels forming the structure image (step S306), and teacher image data serving as teacher data is generated (step S307). If more teacher image data is required, it is determined that the processing from steps S301 to S307 should be performed for another structure image (step S308), and another teacher image data to be teacher data is prepared. Generate.
  • the different structure image may be an image of a structure having the same structure or an image of a structure having a different structure.
  • FIGS. 30 and 31 are diagrams showing examples of the noise distribution set in step S303 described above.
  • the horizontal axes of FIGS. 30 and 31 indicate pixel values to which noise values have been added (hereinafter referred to as noise pixel values), with the pixel values before noise values being added being 100.
  • FIG. The vertical axis in FIGS. 30 and 31 is the relative frequency of noise pixel values.
  • the relative frequency of noise pixel values is a value that indicates the relative frequency of pixel values after adding noise to the pixel values.
  • the noise distributions G28 and G29 have a higher probability that the noise pixel value exceeds the original pixel value compared to the normal distribution (Poisson distribution). noise pixel value corresponding to the case where white spots appear in the image).
  • the case where the X-rays detected by the sensor appear as white dots in the X-ray image is the case where the X-rays are not absorbed by the scintillator but pass through the scintillator and are directly converted into electrons by the sensor. is.
  • the sensor detects visible light when the X-rays are absorbed by the scintillator and converted to visible light. On the other hand, if the x-rays are not absorbed by the scintillator, they are converted directly to electrons in the sensor. That is, a sensor that detects visible light detects not only scintillation light (visible light) generated in the scintillator layer 11C but also X-rays that have passed through the scintillator layer 11C. At this time, the number of electrons generated from the X-rays incident on the sensor is greater than that in the case where the X-rays are converted into visible light by the scintillator. It appears as a white spot in the image.
  • this white spot causes noise in the X-ray image. Therefore, by constructing the trained model 207C using the noise distribution described above and performing noise removal using the constructed trained model 207C, the white spots appearing in the X-ray image can be removed as noise. .
  • a case where white spots are likely to occur is, for example, a case where the tube voltage of the X-ray irradiator is high.
  • the structure image is preferably an image with little noise, ideally an image with no noise. Therefore, generating structure images by simulation calculation can generate many noise-free images, so generating structure images by simulation calculation is effective.
  • An image acquisition device 1C of the third embodiment includes a two-dimensional sensor 12C as a flat panel sensor. Further, scintillator portions Q 1,1 to Q M,N and separation portions R of the scintillator layer 11C are provided for each of the pixels P 1,1 to P M,N of the two-dimensional sensor 12C. This reduces blurring of the X-ray image acquired by the image acquisition device 1C. The result is higher contrast and higher noise intensity in the X-ray image.
  • noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value in the X-ray image is performed using the learned model 207C constructed in advance by machine learning. executed. This reduces only the intensity of noise in the X-ray image. As described above, the image acquisition apparatus 1C can acquire an X-ray image with reduced noise intensity and enhanced contrast.
  • FIGS. 32(a), 32(b), and 32(c) respectively show examples of simulation results of X-ray images acquired by the image acquisition device 1C.
  • An X-ray image G30 shown in FIG. 32(a) is an X-ray image generated by simulation calculation under the condition that the scintillator layer 11C is made of CsI (cesium iodide).
  • a scintillator made of CsI has, for example, a sheet-like shape extending along the pixels of the two-dimensional sensor 12C. The thickness of the scintillator made of CsI is set to 450 ⁇ m.
  • FIG. 32(b) is an X-ray image generated by simulation calculation under the condition that the pixel scintillator having the structure shown in FIG. 28 was used as the scintillator layer 11C.
  • the X-ray image G32 shown in FIG. 32(c) was captured using a pixel scintillator as the scintillator layer 11C, and was subjected to noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value (first 3 is an X-ray image generated by simulation calculation based on the condition that the same noise removal as in the image acquisition apparatus 1C according to the third embodiment is performed.
  • a pixel scintillator is provided for each pixel.
  • the thickness of the pixel scintillator is set at 200 ⁇ m, for example.
  • the thickness of the partition wall (separation portion) of the pixel scintillator along the direction in which the pixels are arranged is set to 40 ⁇ m.
  • the X-ray image G32 is an X-ray image after noise removal corresponding to the relationship between the pixel value and the standard deviation of the noise value has been performed on the X-ray image G31. In each simulation, the pixels of the two-dimensional sensor are set to have a rectangular shape with a side of 120 ⁇ m.
  • the value indicating the magnitude of noise is the value of the standard deviation of the intensity in the background portion (the portion where black dots are not captured).
  • the value indicating the contrast is the difference between the average intensity value in the background portion and the minimum intensity value in the portion where the black dots appear.
  • the CN ratio (CNR: Contrast to Noise Ratio) in the X-ray images G30, G31 and G32 is a value obtained by dividing a value indicating contrast by a value indicating the magnitude of noise.
  • the noise magnitude values are 301.8, 1420.0 and 37.9, respectively, and the contrast values are 3808.1, 9670.9 and 8844 respectively. .3.
  • the CN ratios of X-ray images G30, G31 and G32 are 12.62, 6.81 and 233.16, respectively.
  • the contrast is high and the noise is also large compared to the X-ray image G30.
  • the CN ratio in the X-ray image G31 is half the CN ratio in the X-ray image G30.
  • an X-ray image from which noise is sufficiently removed cannot be acquired only by using the pixel scintillator as the scintillator layer 11C.
  • the X-ray image obtained by using the pixel scintillator as the scintillator layer 11C is subjected to the X-ray image using the learned model 207C constructed in advance by machine learning.
  • the X-ray image G32 according to the third embodiment has higher contrast and less noise than the X-ray image G30.
  • the CN ratio of the X-ray image G32 is 20 times the CN ratio of the X-ray image G30.
  • the image acquisition apparatus 1C according to the third embodiment has the same conditions as the simulation conditions for the X-ray image G32, so it can acquire an X-ray image from which noise is sufficiently removed.
  • the noise distribution has a higher probability that the noise-applied pixel value exceeds the original pixel value compared to the normal distribution. Then, pixel values to which noise is added along the noise distribution are calculated to generate teacher image data.
  • a learned model 207C is constructed using the generated teacher image data.
  • An X-ray image and a noise standard deviation map are input to the built trained model 207C, and image processing is performed to remove noise from the X-ray image.
  • image processing for removing noise from the X-ray image is performed in consideration of the fact that X-rays detected by the sensor in imaging using a scintillator appear as white spots in the X-ray image.
  • the image acquisition device 1C using the scintillator layer 11C can acquire an X-ray image from which noise is more effectively removed.
  • FIG. 33 is a diagram showing a normal distribution G33 used for generating teacher data.
  • the horizontal axis of FIG. 33 indicates the pixel value to which the noise value is added, with the pixel value before the noise value being added being 100.
  • the vertical axis in FIG. 33 is a relative value representing the frequency of noise pixel values.
  • the X-ray detection camera 10 is not limited to a dual-line X-ray camera, a single-line X-ray camera, a dual energy X-ray camera, a TDI (Time Delay Integration) scan X-ray camera, or two or more lines.
  • the image acquisition device 1 is not limited to the above embodiment, and may be a radiation image processing system such as a CT (Computed Tomography) device that captures an image of the object F while it is stationary. Furthermore, a radiation image processing system that captures an image while rotating the object F may be used.
  • CT Computer Planar Tomography
  • the noise map generation step derives the evaluation value from the average energy of the radiation transmitted through the object and the pixel value of each pixel of the radiographic image.
  • the noise map generator preferably derives the evaluation value from the average energy of radiation transmitted through the object and the pixel value of each pixel of the radiographic image.
  • the average energy of the radiation transmitted through the object is taken into account, and the spread of the noise value in the pixel value of each pixel of the radiographic image is evaluated. Noise removal corresponding to the relationship with the spread of noise can be realized. As a result, noise in radiographic images can be removed more effectively.
  • the average energy is calculated based on the input step of accepting the input of the condition information indicating either the condition of the radiation source or the imaging condition when irradiating the radiation to image the object, and the condition information. It is also preferable to further comprise a calculating step of calculating.
  • the input unit receives input of condition information indicating either the condition of the radiation source or the imaging condition when irradiating radiation to image an object, and the average energy is calculated based on the condition information. It is also preferable to further include a calculation unit for calculating.
  • condition information includes the tube voltage of the source, information on the object, information on the filter provided in the camera used to image the object, information on the filter provided in the source, and information on the scintillator provided in the camera used for imaging the object. information.
  • the average energy of the radiation that passes through the object is calculated with high accuracy, so noise removal corresponding to the relationship between the pixel value and the spread of noise can be realized. As a result, noise in radiographic images can be removed more effectively.
  • the jig is irradiated with radiation, and a radiographic image of the jig is obtained by imaging the radiation transmitted through the jig, and in the noise map generation step, relational data is derived from the radiographic image of the jig.
  • the image acquisition unit acquires a radiographic image of the jig in which the jig is irradiated with radiation and the radiation transmitted through the jig is imaged, and the noise map generation unit derives relational data from the radiographic image of the jig. It is preferred that According to this configuration, the relationship data is generated based on the radiographic image obtained by actually imaging the jig, and the noise map is generated. realizable. As a result, noise in radiographic images can be removed more effectively.
  • the image acquisition step a plurality of radiographic images captured without an object are acquired; in the noise map generation step, relational data is derived from the plurality of radiographic images; Preferably, the plurality of images differ from each other in at least one of the source conditions and the imaging conditions. Further, the image acquisition unit acquires a plurality of radiographic images captured without an object, the noise map generation unit derives relational data from the plurality of radiographic images, and the plurality of radiographic images is obtained from the generation of radiation. Preferably, the plurality of images differ from each other in at least one of the source conditions and the imaging conditions.
  • the evaluation value is the standard deviation of the noise value.
  • the spread of the noise value in the pixel value of each pixel of the radiographic image is evaluated more precisely, so that noise removal corresponding to the relationship between the pixel value and the noise can be realized. As a result, noise in radiographic images can be removed more effectively.
  • the machine learning method uses a radiation image as a training image, and a noise map generated from the training image based on relational data representing the relationship between the evaluation value obtained by evaluating the spread of the pixel value and the noise value, and , Noise-removed image data, which is data obtained by removing noise from training images, as training data, and building a trained model that outputs noise-removed image data based on training images and noise maps by machine learning. step.
  • training images that are radiation images, a noise map generated from the training images based on the relational data, and denoised image data that is data obtained by removing noise from the training images are used as training data.
  • a construction unit that constructs a trained model that outputs noise-removed image data based on the training image and the noise map by machine learning.
  • a trained model used for noise removal of radiographic images is constructed by machine learning using training data. Accordingly, when a radiographic image and a noise map generated from the radiographic image are input to the trained model, noise removal corresponding to the relationship between the pixel value and the spread of noise can be realized. As a result, noise in the radiographic image of the object can be removed more effectively.
  • the trained model according to the above embodiment is a trained model constructed by the construction step described above, and causes the processor to perform image processing for removing noise from the radiographic image of the target.
  • noise is removed from the radiographic image by machine learning, taking into consideration the spread of noise values evaluated from the pixel values of each pixel of the radiographic image. Accordingly, noise removal corresponding to the relationship between the pixel value and the spread of noise in the radiographic image can be realized using the learned model. As a result, noise in radiographic images can be effectively removed.
  • the preprocessing method for the machine learning method is a method for generating a noise map, which is training data for the above machine learning method.
  • the noise map which is training data for the machine learning method, corresponds to the relationship between the pixel values, the pixel values, and the evaluation values obtained by evaluating the spread of the noise values.

Landscapes

  • Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Animal Behavior & Ethology (AREA)
  • Biophysics (AREA)
  • Optics & Photonics (AREA)
  • Pathology (AREA)
  • Radiology & Medical Imaging (AREA)
  • Biomedical Technology (AREA)
  • Heart & Thoracic Surgery (AREA)
  • Molecular Biology (AREA)
  • Surgery (AREA)
  • High Energy & Nuclear Physics (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Veterinary Medicine (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Artificial Intelligence (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)
  • Apparatus For Radiation Diagnosis (AREA)
  • Analysing Materials By The Use Of Radiation (AREA)
PCT/JP2021/037173 2021-02-15 2021-10-07 放射線画像処理方法、機械学習方法、学習済みモデル、機械学習の前処理方法、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム Ceased WO2022172506A1 (ja)

Priority Applications (5)

Application Number Priority Date Filing Date Title
KR1020237028293A KR20230145081A (ko) 2021-02-15 2021-10-07 방사선 화상 처리 방법, 기계 학습 방법, 학습 완료 모델, 기계 학습의 전처리 방법, 방사선 화상 처리 모듈, 방사선 화상 처리 프로그램, 및 방사선 화상 처리 시스템
CN202180093696.5A CN116916827A (zh) 2021-02-15 2021-10-07 放射线图像处理方法、机械学习方法、学习完成模型、机械学习的前处理方法、放射线图像处理模块、放射线图像处理程序及放射线图像处理系统
EP21925753.2A EP4252658A4 (en) 2021-02-15 2021-10-07 X-RAY IMAGE PROCESSING METHOD, MACHINE LEARNING METHOD, TRAINED MODEL, MACHINE LEARNING PREPROCESSING METHOD, X-RAY IMAGE PROCESSING MODULE, X-RAY IMAGE PROCESSING PROGRAM, AND X-RAY IMAGE PROCESSING SYSTEM
US18/270,897 US20240054617A1 (en) 2021-02-15 2021-10-07 Radiographic image processing method, machine-learning method, trained model, machine-learning preprocessing method, radiographic image processing module, radiographic image processing program, and radiographic image processing system
JP2022581176A JP7712967B2 (ja) 2021-02-15 2021-10-07 放射線画像処理方法、機械学習方法、学習済みモデル、機械学習の前処理方法、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2021-021673 2021-02-15
JP2021021673 2021-02-15

Publications (1)

Publication Number Publication Date
WO2022172506A1 true WO2022172506A1 (ja) 2022-08-18

Family

ID=82838581

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/037173 Ceased WO2022172506A1 (ja) 2021-02-15 2021-10-07 放射線画像処理方法、機械学習方法、学習済みモデル、機械学習の前処理方法、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム

Country Status (7)

Country Link
US (1) US20240054617A1 (https=)
EP (1) EP4252658A4 (https=)
JP (1) JP7712967B2 (https=)
KR (1) KR20230145081A (https=)
CN (1) CN116916827A (https=)
TW (2) TWI890836B (https=)
WO (1) WO2022172506A1 (https=)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024157532A1 (ja) 2023-01-23 2024-08-02 浜松ホトニクス株式会社 画像処理方法、訓練方法、訓練済みモデル、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム
WO2024176503A1 (ja) 2023-02-20 2024-08-29 浜松ホトニクス株式会社 モデル利用評価方法、モデル利用評価システム及びモデル利用評価プログラム

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7626676B2 (ja) * 2021-07-05 2025-02-04 浜松ホトニクス株式会社 放射線画像取得装置、放射線画像取得システム、及び放射線画像取得方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001099941A (ja) 1999-09-30 2001-04-13 Hitachi Metals Ltd 放射線遮蔽板、放射線検出器及び放射線遮蔽板の製造方法
JP2003167060A (ja) 2001-11-30 2003-06-13 Toshiba Corp Ｘ線平面検出器
WO2013005805A1 (ja) * 2011-07-07 2013-01-10 株式会社 東芝 画像処理装置および方法、ｘ線診断装置
WO2019097796A1 (ja) * 2017-11-17 2019-05-23 株式会社日立製作所 医用画像処理装置および医用画像処理方法
JP2019091393A (ja) 2017-11-10 2019-06-13 アジア航測株式会社 赤色立体地図画像を用いた点群種類推定装置及び赤色立体地図画像を用いた点群種類推定プログラム
JP2019535451A (ja) * 2016-12-06 2019-12-12 コーニンクレッカ フィリップス エヌ ヴェＫｏｎｉｎｋｌｉｊｋｅ Ｐｈｉｌｉｐｓ Ｎ．Ｖ． 交互打ち消しを使用する画像ノイズ推定
JP2020141908A (ja) * 2019-03-07 2020-09-10 株式会社日立製作所 医用撮像装置、医用画像処理装置、医用画像処理プログラム、及び、機械学習モデル

Family Cites Families (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2011172847A (ja) * 2010-02-25 2011-09-08 Fujifilm Corp 放射線画像生成装置及び放射線画像生成方法
US20130051516A1 (en) * 2011-08-31 2013-02-28 Carestream Health, Inc. Noise suppression for low x-ray dose cone-beam image reconstruction
EA201200255A1 (ru) * 2012-02-22 2013-08-30 Закрытое Акционерное Общество "Импульс" Способ подавления шума цифровых рентгенограмм
JP6070939B2 (ja) * 2013-03-07 2017-02-01 富士フイルム株式会社 放射線画像撮影装置および方法
JP6071144B2 (ja) * 2013-07-31 2017-02-01 富士フイルム株式会社 放射線画像解析装置および方法並びにプログラム
JP2020096646A (ja) * 2016-12-05 2020-06-25 キヤノン株式会社 放射線撮影装置、放射線撮影システム、放射線撮影方法、及びプログラム
JP6454820B1 (ja) * 2017-10-24 2019-01-16 株式会社 システムスクエア 非破壊検査装置
JP7282487B2 (ja) * 2018-06-07 2023-05-29 キヤノンメディカルシステムズ株式会社 医用画像診断装置
KR102917833B1 (ko) * 2018-08-15 2026-01-28 에이에스엠엘 네델란즈 비.브이. 원시 이미지들로부터 고품질 평균 sem 이미지들의 자동 선택 시 기계 학습 활용
US10949951B2 (en) * 2018-08-23 2021-03-16 General Electric Company Patient-specific deep learning image denoising methods and systems
US11972559B2 (en) * 2020-02-13 2024-04-30 Fujifilm Corporation Radiographic image processing device, radiographic image processing method, and radiographic image processing program
EP4123297B1 (en) * 2020-04-16 2025-03-05 Hamamatsu Photonics K.K. Radiographic image processing method, trained model, radiographic image processing module, radiographic image processing program, and radiographic image processing system
EP4123296A4 (en) * 2020-04-16 2024-04-10 Hamamatsu Photonics K.K. Radiography method, trained model, radiography module, radiography program, radiography system, and machine learning method
JP7395523B2 (ja) * 2021-02-02 2023-12-11 富士フイルムヘルスケア株式会社 医用画像処理装置および医用画像処理方法

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001099941A (ja) 1999-09-30 2001-04-13 Hitachi Metals Ltd 放射線遮蔽板、放射線検出器及び放射線遮蔽板の製造方法
JP2003167060A (ja) 2001-11-30 2003-06-13 Toshiba Corp Ｘ線平面検出器
WO2013005805A1 (ja) * 2011-07-07 2013-01-10 株式会社 東芝 画像処理装置および方法、ｘ線診断装置
JP2019535451A (ja) * 2016-12-06 2019-12-12 コーニンクレッカ フィリップス エヌ ヴェＫｏｎｉｎｋｌｉｊｋｅ Ｐｈｉｌｉｐｓ Ｎ．Ｖ． 交互打ち消しを使用する画像ノイズ推定
JP2019091393A (ja) 2017-11-10 2019-06-13 アジア航測株式会社 赤色立体地図画像を用いた点群種類推定装置及び赤色立体地図画像を用いた点群種類推定プログラム
WO2019097796A1 (ja) * 2017-11-17 2019-05-23 株式会社日立製作所 医用画像処理装置および医用画像処理方法
JP2020141908A (ja) * 2019-03-07 2020-09-10 株式会社日立製作所 医用撮像装置、医用画像処理装置、医用画像処理プログラム、及び、機械学習モデル

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP4252658A4

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2024157532A1 (ja) 2023-01-23 2024-08-02 浜松ホトニクス株式会社 画像処理方法、訓練方法、訓練済みモデル、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム
KR20250137636A (ko) 2023-01-23 2025-09-18 하마마츠 포토닉스 가부시키가이샤 화상 처리 방법, 훈련 방법, 훈련 완료 모델, 방사선 화상 처리 모듈, 방사선 화상 처리 프로그램, 및 방사선 화상 처리 시스템
EP4623827A1 (en) 2023-01-23 2025-10-01 Hamamatsu Photonics K.K. Image processing method, training method, trained model, radiological image processing module, radiological image processing program, and radiological image processing system
WO2024176503A1 (ja) 2023-02-20 2024-08-29 浜松ホトニクス株式会社 モデル利用評価方法、モデル利用評価システム及びモデル利用評価プログラム

Also Published As

Publication number Publication date
EP4252658A1 (en) 2023-10-04
US20240054617A1 (en) 2024-02-15
TWI898064B (zh) 2025-09-21
TWI890836B (zh) 2025-07-21
JPWO2022172506A1 (https=) 2022-08-18
TW202238460A (zh) 2022-10-01
CN116916827A (zh) 2023-10-20
JP7712967B2 (ja) 2025-07-24
KR20230145081A (ko) 2023-10-17
TW202234049A (zh) 2022-09-01
EP4252658A4 (en) 2024-12-18

Similar Documents

Publication Publication Date Title
JP7712967B2 (ja) 放射線画像処理方法、機械学習方法、学習済みモデル、機械学習の前処理方法、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム
JP7546047B2 (ja) 放射線画像処理方法、学習済みモデル、放射線画像処理モジュール、放射線画像処理プログラム、放射線画像処理システム、及び機械学習方法
JP7546048B2 (ja) 放射線画像処理方法、学習済みモデル、放射線画像処理モジュール、放射線画像処理プログラム、及び放射線画像処理システム
CN115427795B (zh) 放射线图像取得装置、放射线图像取得系统及放射线图像取得方法
JP7626676B2 (ja) 放射線画像取得装置、放射線画像取得システム、及び放射線画像取得方法
KR20250137636A (ko) 화상 처리 방법, 훈련 방법, 훈련 완료 모델, 방사선 화상 처리 모듈, 방사선 화상 처리 프로그램, 및 방사선 화상 처리 시스템
US11185303B2 (en) Image processing apparatus, radiography system, image processing method, and image processing program
TWI885189B (zh) 放射線圖像處理方法、學習完成模型、放射線圖像處理模組、放射線圖像處理程式、放射線圖像處理系統及機器學習方法
TW202307785A (zh) 放射線圖像處理方法、學習完成模型、放射線圖像處理模組、放射線圖像處理程式及放射線圖像處理系統
Kotwaliwale et al. Calibration of a soft X-ray digital imaging system for biological materials
CN120937040A (zh) 图像处理方法、图像处理装置及图像处理程序
US10709402B2 (en) Radiography system, radiography method, radiography program, and derivation apparatus
JP2021056138A (ja) 放射線撮像装置、放射線撮像方法及びプログラム

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21925753

Country of ref document: EP

Kind code of ref document: A1

ENP Entry into the national phase

Ref document number: 2022581176

Country of ref document: JP

Kind code of ref document: A

WWE Wipo information: entry into national phase

Ref document number: 18270897

Country of ref document: US

ENP Entry into the national phase

Ref document number: 2021925753

Country of ref document: EP

Effective date: 20230629

WWE Wipo information: entry into national phase

Ref document number: 202180093696.5

Country of ref document: CN

ENP Entry into the national phase

Ref document number: 20237028293

Country of ref document: KR

Kind code of ref document: A

NENP Non-entry into the national phase

Ref country code: DE