Ultrasonic diagnostic system and ultrasound image processing method

An ultrasound image processing method comprising:

• generating first and second derived images by applying a trained model to a target ultrasound image on a subject, the first derived image comprising a first image component for smoothing, the second derived image comprising a second image component for emphasizing a tissue structure, the trained model being a neural network trained to output, in response to input of an ultrasound image, an image comprising an image component for smoothing and an image comprising an image component for emphasizing a tissue structure;
• inputting a variable first adjustment parameter value and a second adjustment parameter value corresponding to respective of the first and second derived images;
• generating first and second adjusted derived images, the first adjusted derived image being generated by applying the variable first adjustment parameter value for adjusting a smoothing strength to the first derived, the second adjusted derived image being generated by applying the second adjustment parameter value for adjusting a strength of tissue boundary emphasis to the second derived image;
• and
• generating a synthesized image by synthesizing the target ultrasound image and the first and second adjusted derived images;
• the synthesized image comprising an image component for smoothing and an image component for emphasizing a tissue structure are generated based on the ultrasound image and first and second output images, the first and second output images being generated through a nonlinear image process, including,
• performing in at least one layer of multiresolution decomposition/reconstruction and including image filtering in which an image filter which is set with a filter parameter for tissue smoothing and tissue boundary emphasis,
• generating the first output image by applying the image filter which is set with the filter parameter calculated based on the ultrasound image in the nonlinear image process to the ultrasound image,
• generating the second output image by applying the image filter which is set with the filter parameter having an arbitrary value to the ultrasound image,
• the image comprising an image component for smoothing is a subtraction image obtained from the ultrasound image and the second output image, and the image comprising an image component for emphasizing a tissue structure is a subtraction image obtained from the first output image and the second output image;
• the method further including,
• defining input components each having a coordinate space of a number of dimensions corresponding to the number of the two derived images, and
• assigning an adjustment parameter value to each axis of the coordinate space.

An ultrasonic diagnostic system comprising:

• processing circuitry configured to:
• generate first and second derived images by applying a trained model to a target ultrasound image on a subject, the first derived image comprising a first image component for smoothing, the second derived image comprising a second image component for emphasizing a tissue structure, the trained model being a neural network trained to output, in response to input of an ultrasound image, an image comprising an image component for smoothing and an image comprising an image component for emphasizing a tissue structure;
• input first and second adjustment parameter value corresponding to respective of the first and second derived images;
• generate first and second adjusted derived images, the first adjusted derived image being generated by applying a variable first adjustment parameter value for adjusting a smoothing strength to the first derived image, the second adjusted derived image being generated by applying a second adjustment parameter value for adjusting a strength of tissue boundary emphasis to the second derived image;
• generate a synthesized image by synthesizing the target ultrasound image and the first and second adjusted derived images;
• the synthesized image comprising an image component for smoothing and an image component for emphasizing a tissue structure generated based on the ultrasound image and first and second output images, the first and second output images being generated through a nonlinear image process, wherein
• the nonlinear image process is performed in at least one layer of multiresolution decomposition/reconstruction and includes an image filter which is set with a filter parameter for tissue smoothing and tissue boundary emphasis,
• the first output image is generated by applying the image filter which is set with the filter parameter calculated based on the ultrasound image in the nonlinear image process to the ultrasound image,
• the second output image is generated by applying the image filter which is set with the filter parameter having an arbitrary value to the ultrasound image,
• the image comprising an image component for smoothing is a subtraction image obtained from the ultrasound image and the second output image, and
• the image comprising an image component for emphasizing a tissue structure is a subtraction image obtained from the first output image and the second output image;
• the ultrasonic diagnostic system further including input components each having a coordinate space of a number of dimensions corresponding to the number of the two derived images, wherein
• an adjustment parameter value is assigned to each axis of the coordinate space,
• the nonlinear image process includes a nonlinear image filter using a diffusion equation, and
• the parameter is a size of an edge relating to a diffusion tensor of the diffusion equation,
• the second output image is an output image of the nonlinear image process when the edge is zero, and
• the processing circuitry generates, as the two derived images, a first derived image which is a subtraction image of the target ultrasound image and the second output image, and a second derived image which is a subtraction image of the first output image and the second output image.

An ultrasonic diagnostic method comprising:

• generating first and second derived images by applying a trained model to a target ultrasound image on a subject, the first derived image comprising a first image component for smoothing, the second derived image comprising a second image component for emphasizing a tissue structure, the trained model being a neural network trained to output, in response to input of an ultrasound image, an image comprising an image component for smoothing and an image comprising an image component for emphasizing a tissue structure;
• inputting first and second adjustment parameter values corresponding to respective of the first and second derived images;
• generating first and second adjusted derived images, the first adjusted derived image being generated by applying a first adjustment parameter value, which is variable, for adjusting a smoothing strength to the first derived image, the second adjusted derived image being generated by applying the second adjustment parameter value for adjusting a strength of tissue boundary emphasis to the second derived image;
• generate a synthesized image by synthesizing the target ultrasound image and the first and second adjusted derived images;
• the synthesized image comprising an image component for smoothing and an image component for emphasizing a tissue structure generated based on the ultrasound image and first and second output images, the first and second output images being generated through a nonlinear image process, wherein
• the nonlinear image process is performed in at least one layer of multiresolution decomposition/reconstruction and includes an image filter which is set with a filter parameter for tissue smoothing and tissue boundary emphasis,
• the first output image is generated by applying the image filter which is set with the filter parameter calculated based on the ultrasound image in the nonlinear image process to the ultrasound image,
• the second output image is generated by applying the image filter which is set with the filter parameter having an arbitrary value to the ultrasound image,
• the image comprising an image component for smoothing is a subtraction image obtained from the ultrasound image and the second output image, and
• the image comprising an image component for emphasizing a tissue structure is a subtraction image obtained from the first output image and the second output image;
• wherein the diagnostic method further comprises,
• defining input components each having a coordinate space of a number of dimensions corresponding to the number of the two derived images,
• assigning an adjustment parameter value to each axis of the coordinate space,
• the nonlinear image process includes a nonlinear image filter using a diffusion equation, and
• the parameter is a size of an edge relating to a diffusion tensor of the diffusion equation,
• the second output image is an output image of the nonlinear image process when the edge is zero, and
• generating, as the two derived images, a first derived image which is a subtraction image of the target ultrasound image and the second output image, and a second derived image which is a subtraction image of the first output image and the second output image.