IP Library Granted Patent US 12676229
Granted Patent B2
US 12676229 · App. 17/730,954 · Granted Jul 7, 2026

Adaptive ultrasound deep convolution neural network denoising using noise characteristic information

Inventors: Liang Cai (Vernon Hills, IL); Jian Zhou (Vernon Hills, IL); Ting Xia (Vernon Hills, IL); Zhou Yu (Vernon Hills, IL); Tomohisa Imamura (Otawara, JP); Ryosuke Iwasaki (Otawara, JP); Hiroki Takahashi (Nasushiobara, JP)
Assignee: CANON KABUSHIKI KAISHA
G16H30/40A61B8/5269G06N3/08G06T5/70G06T2207/10132G06T2207/20081G06T2207/20084
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 12676229
App. No.
17/730,954
Granted
Jul 7, 2026
Kind
B2
Abstract

A method and system enable to-be-processed medical image data and its corresponding noise characteristic information to be normalized to resemble noise characteristic information of training data used to train at least one neural network for at least one ultrasound data acquisition mode. After normalizing, this processed medical image data is input into the trained neural network for producing output data used for generating cleaner images. Noise characteristic information can be used directly in training a neural network, generating a trained neural network that can handle medical image data with various noise characteristics.

Claims (31)

1 . A medical image processing apparatus, comprising:

a memory storing at least one trained model for denoising medical image data;

processing circuitry configured to (1) obtain processed medical image data by normalizing noise characteristic information of the medical image data to resemble noise characteristic information of training data used for training a corresponding trained model of the at least one trained model, and (2) input the processed medical image data into the corresponding trained model to obtain output data; and

display control circuitry configured to cause a display to display a medical image based on the obtained output data.

2 . The apparatus of claim 1 , wherein the at least one trained model for denoising medical image data stored in the memory comprises plural trained models, and

the processing circuitry is further configured to choose the corresponding trained model from the plural trained models using system parameters of a system used to collect the medical image data.

3 . The apparatus of claim 1 , wherein the at least one trained model for denoising medical image data stored in the memory comprises plural trained models, and

the processing circuitry is further configured to choose the corresponding trained model from the plural trained models using a pre-scanning of air by the apparatus.

4 . The apparatus of claim 1 , wherein the medical image processing apparatus is an ultrasound scanner.

5 . The apparatus of claim 4 , wherein the noise characteristic information is depth-specific noise information related to the ultrasound scanner.

6 . The apparatus of claim 5 , wherein the ultrasound scanner performs at least one of ultrasound B-mode imaging and Doppler ultrasound.

7 . The apparatus of claim 1 , wherein the medical image data is at least one of IQ data, RF data, and image data.

8 . A medical image processing apparatus, comprising:

a memory storing a trained model generated by a machine-learning process based on first medical image data, noise characteristic information of the first medical image data, and second medical image data based on the first medical image data, the second medical image data having less noise than the first medical image data, the noise characteristic information indicating noise at each depth;

processing circuitry configured to input medical image data and the noise characteristic information of the medical image data into the trained model to obtain output data; and

display control circuitry configured to cause a display to display a medical image based on the obtained output data.

9 . The apparatus of claim 8 , wherein the processing circuitry is further configured to obtain the noise characteristic information of the first medical image data by analyzing system parameters of a system used to collect the first medical image data.

10 . The apparatus of claim 8 , wherein the processing circuitry is further configured to obtain the noise characteristic information of the first medical image data by performing a pre-scanning scheme in a system used to collect the first medical image data.

11 . A method, comprising:

obtaining processed medical image data by normalizing noise characteristic information of medical image data to resemble noise characteristic information of training data used for training a corresponding trained model from at least one trained model;

inputting the processed medical image data into the corresponding trained model to obtain output data; and

displaying a medical image based on the obtained output data.

12 . The method of claim 11 , wherein the at least one trained model comprises plural trained models, and

the method further comprises choosing the corresponding trained model from the plural trained models using system parameters of a system used to collect the medical image data.

13 . The method of claim 11 , wherein the at least one trained model comprises plural trained models, and

the method further comprises choosing the corresponding trained model from the plural trained models by performing a pre-scanning of air.

14 . The method of claim 11 , wherein the noise characteristic information is depth-specific noise information related to an ultrasound scanner.

15 . The method of claim 11 , wherein the medical image data is obtained from an ultrasound scanner.

16 . The method of claim 15 , wherein the noise characteristic information is depth-specific noise information related to the ultrasound scanner.

17 . The method of claim 15 , wherein the ultrasound scanner performs at least one of Doppler ultrasound and ultrasound B-mode imaging.

18 . The method of claim 11 , wherein the medical image data is at least one of IQ data, RF data, and image data.