Medical image processing apparatus and medical image processing system
A medical image processing apparatus according to an embodiment is a medical image processing apparatus that generates image data to be used in a trained model trained with image data acquired by a second apparatus, from image data acquired by a first apparatus and that includes a first obtaining unit and a first generating unit. The first obtaining unit is configured to obtain first patient image data obtained by the first apparatus by imaging a patient. The first generating unit is configured to generate, by using a first generative model and from the first patient image data, intermediate image data in which a characteristic of first image data acquired by the first apparatus has been deleted and is configured to generate, by using a second generative model and from the intermediate image data, second patient image data having a characteristic of second image data acquired by the second apparatus.
1. A medical image processing apparatus that generates second subject image data to be used in a trained model trained with second image data acquired by a second apparatus, from first subject image data acquired by a first apparatus, the medical image processing apparatus comprising:
processing circuitry configured to:
obtain the first subject image data acquired by the first apparatus by imaging a subject;
use a first trained model to generate intermediate image data by deleting, from the first subject image data, a characteristic of first image data acquired by the first apparatus; and
use a second trained model to generate, from the intermediate image data, the second subject image data having a characteristic of the second image data acquired by the second apparatus.
2. The medical image processing apparatus according to claim 1 , wherein the processing circuitry is further configured to generate the first trained model by using a generative adversarial network, based on the first image data and reference image data having neither the characteristic of the first image data nor the characteristic of the second image data.
3. The medical image processing apparatus according to claim 1 , wherein the processing circuitry is further configured to generate the second trained model by using a generative adversarial network, based on reference image data having neither the characteristic of the first image data nor the characteristic of the second image data, and the second image data.
4. The medical image processing apparatus according to claim 1 , wherein the processing circuitry is further configured to cause the trained model to analyze the second subject image data.
5. The medical image processing apparatus according to claim 4 , wherein the processing circuitry is further configured to
obtain the trained model, and
cause the obtained trained model to analyze the second subject patient image data.
6. The medical image processing apparatus according to claim 1 , wherein the processing circuitry is further configured to use the first trained model to generate the intermediate image data by deleting, from the first subject image data, the characteristic of the first image data acquired by the first apparatus by imaging an evaluation-purpose phantom used for evaluating a capability.
7. The medical image processing apparatus according to claim 6 , wherein the processing circuitry is further configured to use the second trained model to generate, from the intermediate image data, the second subject image data having the characteristic of the second image data acquired by the second apparatus by imaging the evaluation-purpose phantom.
8. The medical image processing apparatus according to claim 1 , wherein the processing circuitry is further configured to use the first trained model to generate the intermediate image data by deleting, from the first subject image data, the characteristic of the first image data acquired by the first apparatus by imaging a human body.
9. The medical image processing apparatus according to claim 8 , wherein, as the second image data, the processing circuitry is further configured to use image data acquired by the second apparatus by imaging the human body.
10. A medical image processing system, comprising:
a model generating apparatus that generates trained models; and
a medical image processing apparatus that generates image data by using the trained models, wherein
the model generating apparatus includes first processing circuitry configured to:
generate a first trained model by using a generative adversarial network, based on first image data acquired by a first apparatus and reference image data having neither a characteristic of the first image data nor a characteristic of second image data acquired by a second apparatus; and
generate a second trained model by using a generative adversarial network based on reference image data having neither the characteristic of the first image data nor the characteristic of the second image data, and the second image data, and
the medical image processing apparatus includes second processing circuitry configured to:
obtain first subject image data acquired by the first apparatus by imaging a subject; and
use the first trained model to generate intermediate image data by deleting the characteristic of the first image data acquired by the first apparatus from the first subject image data; and
use the second trained model to generate, from the intermediate image data, second subject image data having the characteristic of the second image data acquired by the second apparatus.