IP Library Granted Patent US 12667340
Granted Patent B2
US 12667340 · App. 17/464,161 · Granted Jun 30, 2026

Hypercomplex-number operation device and medical image diagnostic apparatus

Inventors: Hidenori Takeshima (Kawasaki, JP); Ryota Osumi (Nasushiobara, JP)
Assignee: Canon Kabushiki Kaisha
A61B8/5223A61B5/055A61B8/5207G06N3/045G06N3/08G06T7/0012G06T2207/10088G06T2207/10136G06T2207/20081G06T2207/20084
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12667340
App. No.
17/464,161
Granted
Jun 30, 2026
Kind
B2
Abstract

A hypercomplex operation device according to an embodiment includes processing circuitry. The processing circuitry acquires data including a hypercomplex number, inputs a parametric function in which a function form for a first component and a function form for a second component that is different from the first component differ from each other, inputs the data including a hypercomplex number, to apply to the parametric function, and thereby outputs output data.

Claims (38)

1 . A hypercomplex operation device comprising:

processing circuitry configured to acquire data including a hypercomplex number, the data being data acquired from an ultrasound probe or magnetic resonance imaging apparatus, configured to input a parametric function in which a function form for a first component and a function form for a second component are different, the second component being different from the first component, and configured to input data including a hypercomplex number to apply to the parametric function, thereby outputting output data, the output data being an ultrasound diagnostic image or a magnetic resonance image, wherein

the parametric function is a neural network,

the neural network comprises a plurality of subnetworks in which the hypercomplex numbers are represented by respective different coordinate systems and includes a subnetwork that connects at least two subnetworks out of the plurality of subnetworks,

the neural network comprises a plurality of layers,

each of the layers includes a first subnetwork in which the hypercomplex number is represented by a first coordinate system, a second subnetwork in which the hypercomplex number is represented by a second coordinate system different from the first coordinate system, a third subnetwork that connects the first subnetwork and the second subnetwork, and that accepts an input from the first subnetwork and the second subnetwork, a fourth subnetwork that accepts an input from the first subnetwork and the third subnetwork, and a fifth subnetwork that accepts an input from the second subnetwork and the third subnetwork,

the first coordinate system is a Cartesian coordinate system,

the second coordinate system is a polar coordinate system, and

the fourth subnetwork outputs data represented by the first coordinate system and the fifth subnetwork outputs data represented by the second coordinate system.

2 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number includes a real number as a proper subset.

3 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number is a quaternion.

4 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number constitutes a non-commutative algebra.

5 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number is a complex number, and

the neural network includes a function to convert a complex number represented by Cartesian coordinates into a complex number represented by polar coordinates, and a function to convert a complex number represented by polar coordinates into a complex number represented by Cartesian coordinates.

6 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number is a complex number, and

the neural network includes a parametric function relating to an argument in polar coordinate representation of a complex number.

7 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number is a quaternion, and

the neural network includes a function to convert a three-dimensional position into a quaternion, and a function to convert a quaternion into a three-dimensional position.

8 . The hypercomplex operation device according to claim 1 , wherein

the parametric function is a function in which number of parameters relating to a function form for the first component and number parameters relating to a function form for the second component are different.

9 . The hypercomplex operation device according to claim 1 , wherein

the hypercomplex number is a complex number.

10 . A medical image diagnostic apparatus comprising:

a scanning unit; and

the hypercomplex operation device according to claim 1 , wherein

the processing circuitry acquires data that is acquired by the scanning unit as the data including the hypercomplex number.

11 . The medical image diagnostic apparatus according to claim 10 , wherein

the scanning unit is an ultrasound probe, and

the processing circuitry outputs an ultrasound diagnostic image as the output data.

12 . The medical image diagnostic apparatus according to claim 10 , wherein

the scanning unit collects a magnetic resonance signal as sequence control circuitry performing pulse sequence, and

the processing circuitry outputs a magnetic resonance image as the output data.