Ultrasonic imaging device
An ultrasonic imaging device has a learning result as a relationship between the ultrasonic reception data and the shape data obtained by deep learning using the ultrasonic reception data of an imaging target obtained by transmitting and receiving ultrasonic signals by the plurality of element and the shape data of the pedestal (arrangement of multiple elements). Then the ultrasonic imaging device obtains the estimated shape data as the estimated shape data of the pedestal (estimated arrangement of multiple elements) by applying the learning result to the ultrasonic reception data, and constructs an image of the imaging target based on the estimated shape data and the ultrasonic reception data.
1 . An ultrasonic imaging device equipped with a probe having a plurality of elements capable of transmitting and receiving ultrasonic signals arranged in alignment on a pedestal formed from a deformable plate-like material:
wherein the ultrasonic imaging device has a learning result as a relationship between ultrasonic reception data and shape data obtained by deep learning using the ultrasonic reception data of an imaging target obtained by transmitting and receiving ultrasonic signals by the plurality of elements and the shape data of the pedestal; and
the ultrasonic imaging device is configured to:
obtain estimated shape data as the estimated shape data of the pedestal by applying the learning result to the ultrasonic reception data,
construct an image of the imaging target based on the estimated shape data and the ultrasonic reception data, and
output the constructed image to a display in real-time.
2 . The ultrasonic imaging device according to claim 1 , wherein
the estimated shape data is a sequence of coefficients for each basis function in a shape function represented by a linear combination of any sequence of basis functions.
3 . The ultrasonic imaging device according to claim 2 , wherein the shape function is P(1)sin(x)+P(2)sin(2x)+ . . . +P(n)sin(nx).
4 . The ultrasonic imaging device according to claim 2 , wherein
the deep learning process uses the random shape data of the pedestal and the simulation reception data as the shape data and the ultrasonic reception data to obtain the learning result,
the random shape data of the pedestal is the data when the plurality of elements are randomly placed on the processed image obtained by edge detection processing on the natural image, and
the simulation reception data is obtained by transmitting and receiving ultrasonic signals by the plurality of elements in the random shape data against the processed image by simulation.
5 . The ultrasonic imaging device according to claim 3 , wherein
the deep learning process uses the random shape data of the pedestal and the simulation reception data as the shape data and the ultrasonic reception data to obtain the learning result,
the random shape data of the pedestal is the data when the plurality of elements are randomly placed on the processed image obtained by edge detection processing on the natural image, and
the simulation reception data is obtained by transmitting and receiving ultrasonic signals by said plurality of elements in said random shape data against said processed image by simulation.
6 . The ultrasonic imaging device according to claim 4 , wherein the learning result is obtained by deep learning using the random shape data and the simulated reception data as the shape data and the ultrasonic reception data, and then performing additional deep learning on the results obtained by this process, using the measured actual shape data of the pedestal and the measured actual reception data by the probe when the probe is attached to the living body as the imaging target, as the shape data and the ultrasonic reception data.
7 . The ultrasonic imaging device according to claim 2 , wherein the deep learning obtains the learning result using the measured actual shape data of the pedestal and the measured actual reception data by the probe when the probe is attached to the living body as the imaging target, as the shape data and the ultrasonic reception data.
8 . The ultrasonic imaging device according to claim 3 , wherein the deep learning obtains the learning result using the measured actual shape data of the pedestal and the measured actual reception data by the probe when the probe is attached to the living body as the imaging target, as the shape data and the ultrasonic reception data.