Focus optimization for prediction in multi-frequency ultrasound imaging

Patent metadata

Field	Value
Publication number	US-2023148382-A1
Application number	US-202117916103-A
Country	US
Kind code	A1
Filing date	Mar 25, 2021
Priority date	Apr 1, 2020
Publication date	May 11, 2023
Grant date	—

Abstract

An imaging system (IS), comprising an image acquisition unit (AQ) for acquisition of image data (I1) of an object (OB). The image acquisition is based on an imaging signal imitable by the unit (AQ) to interact with the object. The image acquisition unit (AQ) is adjustable to operate at different acquisition parameters that determine a property of the imaging signal. A predictor component (PC) predicts, based at least on the acquired image data (I1), one or more properties of the object. An acquisition parameter adjuster (PA) adjusts, based on the predicted object properties, the acquisition parameter at which the image acquisition unit (AQ) is to acquire follow-up image data (I2).

First claim

1 . An imaging system (IS), comprising:
an image acquisition unit (AQ) for acquisition of image data (I 1 ) of at least a part of an object (OB), the said acquisition being based on an imaging signal imitable by the unit (AQ) to interact with the object, the image acquisition unit (AQ) adjustable to operate at different acquisition parameters that determine at least in part a property of the imaging signal; a predictor component (PC) configured to predict, based at least on the acquired image data (I 1 ), one or more properties of the object; and an acquisition parameter adjuster (PA) configured to adjust, based on the predicted one or more properties, the acquisition parameter at which the image acquisition unit (AQ) is to acquire follow-up image data ( 12 ).

2 . System of claim 1 , wherein the acquisition parameter is adjusted automatically by the acquisition parameter adjuster (PA) so as to increase image quality.

3 . System of claim 1 , wherein the predictor component (PC) is to predict plural object properties associated with a respective uncertainty value, wherein the acquisition parameter adjuster (PA) is to adjust the acquisition parameter based on the uncertainty value or on a gradient thereof.

4 . System of claim 3 , wherein the acquisition parameter adjuster (PA) is to adjust the acquisition parameter so as to decrease the uncertainty value of the predicted one or more properties.

5 . System of claim 1 , wherein the predictor component (PC) is to predict the one or more object properties based on a current acquisition parameter.

6 . System of claim 3 , wherein the uncertainty value is provided by a user via a user interface (UI).

7 . System of claim 1 , wherein the predictor component (PC) includes a pre-trained machine learning model.

8 . System of claim 7 , wherein the pre-trained machine learning model includes a neural network.

9 . System of claim 1 , wherein the image acquisition unit (AQ) includes a multi-frequency ultrasound imaging device, wherein the imaging signal is an ultrasound signal.

10 . System of claim 9 , wherein the acquisition parameter includes a frequency of the ultrasound signal.

11 . System of claim 9 , wherein the ultrasound imaging device is an intravascular ultrasound, IVUS, imaging device.

12 . An imaging method, comprising the steps of:
acquiring (S 610 ) image data (I 1 ) of at least a part of an object (OB), the said acquiring being based on an imaging signal imitable by an image acquisition unit (AQ) to interact with the object, the image acquisition unit (AQ) adjustable to operate at different acquisition parameters that determine at least in part a property of the imaging signal; predicting (S 620 ), based at least on the acquired image data (I 1 ), one or more properties of the object; and adjusting (S 630 ), based on the predicted one or more properties, the acquisition parameter at which the image acquisition unit (AQ) is to acquire follow-up image data ( 12 ).

13 . A training method configured to train the predictor component (PC) as per claim 7 .

14 . A computer program element, which, when being executed by at least one processing unit (PR,TS), is adapted to cause the at least one processing unit (PR,TS) to perform the method as per claim 12 .

15 . A computer readable medium having stored thereon the program element of claim 14 .

Assignees

• Koninklijke Philips Nv

Inventors

• Schnellbächer Nikolas David
• Wissel Tobias
• Nickisch Hannes
• Grass Michael

Classifications

• G06N3/09

  Supervised learning · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• A61B8/585 Primary

  Automatic set-up of the device · CPC title

• G01S7/52036

  using analysis of echo signal for target characterisation · CPC title

• A61B8/0891

  for diagnosis of blood vessels · CPC title