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

1 . A radiographic image processing method comprising:

acquiring a radiographic image obtained by irradiating a subject with radiation and capturing an image of the radiation passing through the subject;

converting a pixel value of each pixel in the radiographic image into a standard deviation of a noise value on the basis of relationship data indicating a relationship between the pixel value and the standard deviation, and generating a noise standard deviation map that is data in which the converted standard deviation is associated with each pixel in the radiographic image; and

inputting the radiographic image and the noise standard deviation map in parallel to a trained model constructed in advance through machine learning and executing image processing of removing noise from the radiographic image.

2 . The radiographic image processing method according to claim 1 , wherein the acquiring of the radiographic image includes converting average energy related to the radiation passing through the subject and the pixel value of each pixel in the radiographic image into the standard deviation.

3 . The radiographic image processing method according to claim 2 , further comprising:

accepting an input of condition information indicating either conditions of a source of radiation or imaging conditions when the radiation is radiated to capture an image of the subject; and

calculating the average energy on the basis of the condition information,

wherein the condition information includes at least any one of a tube voltage of the source, information relating to the subject, information on a filter included in a camera used to capture an image of the subject, information on a filter included in the source, and information on a scintillator included in the camera.

4 . The radiographic image processing method according to claim 2 , further comprising calculating the average energy from the pixel value of each pixel in the radiographic image.

5 . The radiographic image processing method according to claim 1 , wherein the acquiring of the radiographic image includes acquiring a radiographic image of a jig obtained by irradiating the jig with radiation and capturing an image of the radiation passing through the jig, and

the deriving of the evaluation value includes deriving the relationship data from the radiographic image of the jig.

6 . The radiographic image processing method according to claim 1 , wherein the acquiring of the radiographic image includes acquiring a plurality of radiographic images without the subject,

the deriving of the evaluation value includes deriving the relationship data from the plurality of radiographic images, and

the plurality of radiographic images are a plurality of images in which conditions of at least one out of conditions of a source of radiation and imaging conditions differ from each other.

7 . A machine-learning method comprising:

converting a pixel value of each pixel in a radiographic image into a standard deviation of a noise value on the basis of relationship data indicating a relationship between the pixel value and the standard deviation, and generating a noise standard deviation map that is data in which the converted standard deviation is associated with each pixel in the radiographic image; and

using the radiographic image as a training image and using the noise standard deviation map generated from the training image on the basis of the relationship data indicating the relationship between the pixel value and the standard deviation, and noise-removed image data which is data obtained by removing noise from the training image, as training data, to construct a trained model that outputs the noise-removed image data on the basis of the training image and the noise standard deviation map through machine learning.

8 . A trained model constructed using the machine-learning method according to claim 7 , the trained model causing a processor to execute image processing of removing noise from a radiographic image of a subject.

9 . A radiographic image processing module comprising a processor configured to:

acquire a radiographic image obtained by irradiating a subject with radiation and capturing an image of the radiation passing through the subject;

convert a pixel value of each pixel in the radiographic image into a standard deviation of a noise value on the basis of relationship data indicating a relationship between the pixel value and the standard deviation, and generate a noise map that is data in which the converted standard deviation is associated with each pixel in the radiographic image; and

input the radiographic image and the noise standard deviation map in parallel to a trained model constructed in advance through machine learning and execute image processing of removing noise from the radiographic image.

10 . The radiographic image processing module according to claim 9 , wherein the processor converts average energy related to the radiation passing through the subject and the pixel value of each pixel in the radiographic image into the standard deviation.

11 . The radiographic image processing module according to claim 10 , wherein

the processor accept an input of condition information indicating either conditions of a source of radiation or imaging conditions when the radiation is radiated to capture an image of the subject, and

the processor calculate the average energy on the basis of the condition information,

wherein the condition information includes at least any one of a tube voltage of the source, information relating to the subject, information on a filter included in a camera used to capture an image of the subject, information on a filter included in the source, and information on a scintillator included in the camera used to capture an image of the subject.

12 . The radiographic image processing module according to claim 10 , wherein the processor calculate the average energy from the pixel value of each pixel in the radiographic image.

13 . The radiographic image processing module according to claim 9 , wherein the processor acquires a radiographic image of a jig obtained by irradiating the jig with radiation and capturing an image of the radiation passing through the jig, and derives the relationship data from the radiographic image of the jig.

14 . The radiographic image processing module according to claim 9 , wherein the processor acquires a plurality of radiographic images without the subject, and

derives the relationship data from the plurality of radiographic images, and

the plurality of radiographic images are a plurality of images in which conditions of at least one out of conditions of a source of radiation and imaging conditions differ from each other.

15 . The radiographic image processing module according to claim 9 , wherein the processor uses a training image which is a radiographic image, the noise map generated from the image on the basis of the relationship data, and noise-removed image data which is data obtained by removing noise from the training image, as training data, to construct a trained model that outputs the noise-removed image data on the basis of the training image and the noise map through machine learning.

16 . A radiographic image processing program causing a processor to function as:

acquiring a radiographic image obtained by irradiating a subject with radiation and capturing an image of the radiation passing through the subject;

converting a pixel value of each pixel in the radiographic image into a standard deviation of a noise value on the basis of relationship data indicating a relationship between the pixel value and the standard deviation, and generating a noise map that is data in which the converted standard deviation is associated with each pixel in the radiographic image; and

inputting the radiographic image and the noise standard deviation map in parallel to a trained model constructed in advance through machine learning and executing image processing of removing noise from the radiographic image.

17 . A radiographic image processing system comprising:

the radiographic image processing module according to claim 9 ;

a source configured to irradiate the subject with radiation; and

an imaging device configured to capture an image of the radiation passing through the subject and acquire the radiographic image.

18 . The radiographic image processing system according to claim 17 , wherein the imaging device has a line sensor.

19 . The radiographic image processing system according to claim 17 , wherein the imaging device has a two-dimensional sensor.