Image generation device, image generation method, and learned model generation method
This image generation device is provided with: a learning image generation unit ( 1 ) for generating a training input image and a training output image based on three-dimensional data; a noise addition unit ( 2 ) for adding the same noise to the training input image and the training output image; a learning unit ( 3 ) for learning a learning model for extracting or removing a specific portion by performing machine learning based on the training input image to which the noise has been added and the training output image to which the noise has been added; and an image generation unit ( 4 ) for generating an image from which the specific portion has been extracted or removed by using a learned learning model.
1. An image generation device comprising:
a learning image generation unit configured to generate, based on three-dimensional data having three-dimensional pixel value data, a two-dimensional training input image in an area including a specific portion of a subject and a training output image that is an image showing the specific portion or an image excluding the specific portion at the same angle of view as the training input image;
a noise addition unit configured to add the same noise to the training input image and the training output image;
a learning unit configured to learn a learning model for extracting or removing the specific portion by performing machine learning based on the training input image to which the noise has been added and the training output image to which the noise has been added; and
an image generation unit configured to generate, from a captured image including the specific portion, an image from which the specific portion has been extracted or removed, using a learned learning model,
wherein the noise addition unit is configured to add noise adjusted based on noise of an imaging result that captured an image to the training input image and the training output image.
2. The image generation device as recited in claim 1 ,
wherein the learning image generation unit is configured to acquire the three-dimensional data based on a computed tomography image that captured the subject and generate the training input image and the training output image by a digital reconstruction simulation.
3. The image generation device as recited in claim 1 ,
wherein the noise addition unit is configured to add Gaussian noise and is configured to add Gaussian noise adjusted to a standard deviation corresponding to a standard deviation of noise of the imaging result that captured the image to the training input image and the training output image.
4. The image generation device as recited in claim 1 ,
wherein the specific portion is a bone,
wherein the learning image generation unit is configured to generate a two-dimensional training input image in an area including a bone of the subject and the training output image excluding the bone of the subject,
wherein the learning unit is configured to learn the learning model for removing the bone by performing machine learning based on the training input image to which the noise has been added and the training output image to which the noise has been added, and
wherein the image generation unit is configured to generate, from the image including the imaged bone, an image from which the bone has been removed, using the learned learning model.
5. The image generation device as recited in claim 1 ,
wherein the specific portion is a blood vessel,
wherein the learning image generation unit is configured to generate a two-dimensional training input image of an area including a blood vessel of the subject and the training output image indicating the blood vessel of the subject,
wherein the learning unit is configured to learn the learning model for extracting the blood vessel by performing machine learning, based on the training input image to which the noise has been added and the training output image to which the noise has been added, and
wherein the image generation unit is configured to generate, from an image including an imaged blood vessel, an image by extracting the blood vessel using the learned learning model.
6. An image generation method comprising:
generating a two-dimensional training input image in an area including a specific portion of a subject, based on three-dimensional data having three-dimensional pixel value data;
generating a training output image which is an image showing the specific portion or an image excluding the specific portion at the same angle of view as the training input image;
adding the same noise to the training input image and the training output image, the noise being noise adjusted based on noise of an imaging result that captured an image;
learning a learning model for extracting or removing the specific portion by performing machine learning based on the training input image to which the noise has been added and the training output image to which the noise has been added; and
generating an image from which the specific portion has been extracted or removed, from the captured image including the specific portion, using the learned learning model.
7. A method of producing a learned model, comprising:
generating a two-dimensional training input image in an area including the specific portion of a subject, based on three-dimensional data having three-dimensional pixel value data;
generating a training output image which is an image showing the specific portion or an image excluding the specific portion at the same angle of view as the training input image;
adding the same noise to the training input image and the training output image, the noise being noise adjusted based on noise of an imaging result that captured an image; and
learning a learning model for extracting or removing the specific portion by performing machine learning based on the training input image to which the noise has been added and the training output image to which the noise has been added.