EP-4167866-B1 - SYSTEMS AND METHODS FOR GENERATING COLOR DOPPLER IMAGES FROM SHORT AND UNDERSAMPLED ENSEMBLES

Inventors

• APOSTOLAKIS, Iason Zacharias
• MERAL, Faik Can
• SHIN, JUN SEOB
• VIGNON, FRANCOIS GUY GERARD MARIE
• WANG, SHIYING
• ROBERT, Jean-Luc Francois-Marie

Dates

Publication Date

20260506

Application Date

20210615

Claims (15)

1. An ultrasound imaging system comprising: a processor (260) configured to: receive ultrasound signals corresponding to a first radiofrequency. RF, ensemble comprising a first length, a first pulse repetition frequency, PRF, and a first sensitivity; use artificial intelligence to estimate a second RF-ensemble using the first RF-ensemble as input, wherein the estimate is based, at least in part, on reference ultrasound signals that correspond to at least one reference RF-ensemble comprising at least one of a second pulse repetition rate, a second length, or a second sensitivity compared to the first RF-ensemble, wherein the estimated second RF- ensemble comprises at least one of a higher pulse repetition rate, a longer length, or a higher sensitivity; characterized in that the processor is configured to use the artificial intelligence to output a color Doppler image generated from the first RF ensemble that mimics the appearance of a color Doppler image of the estimated second RF-ensemble.
2. The ultrasound imaging system of claim 1, wherein the processor is further configured to wall-filter the ultrasound signals corresponding to the first RF-ensemble.
3. The ultrasound imaging system of claim 1, wherein the processor implements a neural network to estimate the second RF-ensemble, wherein the neural network comprises a series of convolutional neural networks.
4. The ultrasound imaging system of claim 3, wherein the series of convolutional neural networks are uNets.
5. The ultrasound imaging system of claim 3, wherein, the neural network further generates the color Doppler image, and wherein a first neural network of the series of convolutional neural networks receives the ultrasound signals corresponding to the RF-ensemble and provides a mean Doppler phase image and a Doppler phase as a first output, wherein a second neural network of the series of convolutional neural networks receives the first output and provides the color Doppler image as a second output.
6. A method comprising: receiving (902) ultrasound signals from an external probe corresponding to a first radiofrequency, RF, ensemble comprising a first length, a first pulse repetition frequency, PRF, and a first sensitivity; using artificial intelligence to estimate a second RF-ensemble using the first RF-ensemble as input, wherein the estimate is based, at least in part, on reference ultrasound signals that correspond to at least one reference RF-ensemble comprising at least one of a second pulse repetition rate, a second length, or a second sensitivity compared to the first RF-ensemble, wherein the estimated second RF- ensemble comprises at least one of a higher pulse repetition rate, a longer length, or a higher sensitivity; and characterized in that the method comprises using the artificial intelligence to output a color Doppler image generated from the first RF ensemble that mimics the appearance of a color Doppler image of the estimated second RF-ensemble.
7. The method of claim 6, further comprising wall-filtering the ultrasound signals prior to the estimating.
8. The method of claim 6, wherein the estimating is performed by a neural network comprising a series of convolutional networks.
9. The method of claim 8, further comprising (906) training the neural network, wherein the training comprises: providing at least one RF-ensemble having at least one of a third length, a third PRF, or a third sensitivity that is the same as the first RF-ensemble; and providing a corresponding color Doppler image generated from the reference RF-ensemble as a desired output.
10. The method of claim 9, wherein the training further comprises applying a masked mean squared error (MSE) loss.
11. The method of claim 10, wherein the masked MSE loss is based, at least in part, on power Doppler data of the reference RF-ensemble.
12. The method of claim 9, wherein the training further comprises applying an adversarial loss.
13. The method of claim 12, wherein the adversarial loss is based on a binary cross-entropy loss generated by training a discriminator network to distinguish color Doppler images generated from the first RF-ensemble from color Doppler images generated from the reference RF-ensemble.
14. The method of claim 13, wherein the discriminator network has a conditional generative adversarial network architecture.
15. A non-transitory computer-readable medium containing instructions, that when executed, causes an imaging system to: receive ultrasound signals corresponding to a first radiofrequency, RF, ensemble comprising a first length, a first pulse repetition frequency, PRF, and a first sensitivity; use artificial intelligence to estimate a second RF-ensemble using the first RF-ensemble as input, wherein the estimate is based, at least in part, on reference ultrasound signals that correspond to at least one reference RF-ensemble comprising at least one of a second pulse repetition rate, a second length, or a second sensitivity compared to the first RF-ensemble, wherein the estimated second RF- ensemble comprises at least one of a higher pulse repetition rate, a longer length, or a higher sensitivity; characterized in that the instructions cause the imaging system to use the artificial intelligence to output a color Doppler image generated from the first RF ensemble that mimics the appearance of a color Doppler image of the estimated second RF-ensemble.

Description

TECHNICAL FIELD The present disclosure pertains to imaging systems and methods for generating color Doppler images. In particular, imaging systems and methods for generating color Doppler images from short and undersampled radiofrequency (RF) ensembles. BACKGROUND Color Doppler (CD) has traditionally been used to diagnose atherosclerotic disease in the extracranial arteries and inspect the heart's hemodynamic properties. However, inspection of blood flows in the heart as well as in stenosed arteries often poses challenges due to high and turbulent blood velocities. CD images are generated by estimating the mean phase shift within an ensemble of ultrasound pulses coming from the same sample volume (radiofrequency (RF)-ensemble). The most widely used phase shift estimation technique is the lag-1 autocorrelation operating on the slow-time RF-ensemble pulses. Power Doppler (PD) is extracted from the lag-0 autocorrelation. CD provides information relating to velocity (e.g., speed and direction) of blood flow. PD provides more sensitive detection of blood flow than CD, but does not provide directional information. Moving blood echoes can be separated from tissue background by applying a high-pass filter along the slow time direction (e.g., wall filter). One determinant of CD signal quality is the number of pulses/slow-time observations in the ensemble (RF-ensemble size) used to generate a single CD image. A larger RF-ensemble size provides increased accuracy and sensitivity of CD blood flow velocity estimation compared to a smaller RF-ensemble size. Another CD quality factor is the pulse repetition frequency (PRF) of the ensemble pulses. The PRF determines the time-sampling rate of the Doppler signal and the highest blood flow velocity that can be recovered. WO2020/083679 discloses ultrasound systems configured to enhance flow imaging and analysis by adaptively adjusting one or more imaging parameters in response to acquired flow measurements. SUMMARY The invention is defined by the claims. As disclosed herein, artificial intelligence (e.g., deep learning) is leveraged to correlate short/decimated ensembles with longer ensembles and/or ensembles with higher PRF in color Doppler images. In this manner, high quality color Doppler images is generated from the short/decimated ensembles that mimic the appearance color Doppler images generated from longer ensembles and/or ensembles with higher PRF. An ultrasound imaging system according to an example of the present disclosure is provided in claim 1. A method according to an example of the present disclosure is provided in claim 6. In accordance with an example of the present disclosure, a non-transitory computer-readable medium containing instructions is provided in claim 15. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 shows example CD images for a long ensemble, a short, ensemble, and a decimated ensemble.FIG. 2 is a block diagram of an ultrasound system in accordance with principles of the present disclosure.FIG. 3 is a block diagram illustrating an example processor in accordance with principles of the present disclosure.FIG. 4 is a block diagram of a process for training and deployment of a neural network in accordance with the principles of the present disclosure.FIG. 5 shows an overview of a training phase and a deployment phase of a neural network in accordance with the principles of the present disclosure.FIG. 6 illustrates an example workflow for generating a training data set in accordance with the principles of the present disclosure.FIG. 7 is a diagram of a neural network in accordance with the principles of the present disclosure.FIG. 8 is a flowchart of a method that illustrates use of optional masked loss in accordance with the principles of the present disclosure.FIG. 9 is a flow chart of a method in accordance with the principles of the present disclosure.FIG. 10 shows an example decimated ensemble color Doppler image and an example enhanced color Doppler image in accordance with the principles of the present disclosure.FIG. 11 shows example short ensemble, long ensemble, and enhanced raw phase shift result color Doppler images in accordance with the principles of the present disclosure.FIG. 12 shows example short ensemble, long ensemble, and enhanced raw phase shift result color Doppler images in accordance with the principles of the present disclosure.FIG. 13 shows example short ensemble, long ensemble, and enhanced post-processed color Doppler images in accordance with the principles of the present disclosure.FIG. 14 shows example short ensemble, long ensemble, and enhanced post-processed color Doppler images in accordance with the principles of the present disclosure. DETAILED DESCRIPTION The following description of certain embodiments is merely exemplary in nature and is in no way intended to limit the invention or its applications or uses. In the following detailed description of embodiments of the present systems and methods, reference is made to