【免费】长DeepNeuralNetworksforUltrasoundBeamforming1资源-CSDN文库

需积分: 0 64 浏览量 2022-08-03 12:38:01 上传评论收藏 2.74MB PDF 举报

在超声成像领域，传统的延迟和求和（Delay-and-Sum, DAS）波束形成算法是目前广泛采用的标准方法。然而，这种方法在抑制非轴向散射、提高图像对比度和信噪比（SNR）方面存在局限性。为了解决这些问题，近年来研究人员开始探索深度神经网络（Deep Neural Networks, DNNs）在超声波束形成中的应用潜力。本文"长Deep Neural Networks for Ultrasound Beamforming"由Adam C. Luchies和Brett C. Byram两位IEEE成员发表于2018年9月的《IEEE Transactions on Medical Imaging》第37卷第9期，研究了如何利用DNN技术来改善超声通道数据中的非轴向散射抑制。该研究的独特之处在于它在频率域中通过短时傅立叶变换（Short-Time Fourier Transform, STFT）操作，使得DNN能够处理复杂的数据结构。研究中，DNN的输入包括阵列孔径上观察到的实部和虚部（即同相和正交分量），针对单个频率和单个深度。不同频率对应训练不同的网络。网络输出保持与输入相同的结构，实部和虚部组合成复数数据后，通过逆短时傅立叶变换重建通道数据。通过模拟实验、物理幻影实验和人体肝脏的活体扫描，研究人员比较了DNN方法与标准DAS波束形成以及使用相干因子的自适应成像技术的性能。结果显示，在模拟的点目标中，DNN方法的旁瓣比标准DAS低约60dB；对于模拟的无回声囊肿，DNN方法提升了8.8dB的对比度比率（CR）和0.3dB的对比噪声比（CNR）；在物理幻影中的无回声囊肿，这些提升值分别为17.1dB和0.7dB；在两个活体扫描中，DNN方法分别提高了13.8dB的CR和9.7dB的CNR。此外，作者还探讨了分析网络功能的方法，以深入理解DNN在超声波束形成中的工作原理。这项工作为超声成像技术的改进提供了新的视角，通过DNN的应用，有望实现更准确、更清晰的图像，从而提高诊断的准确性和效率。关键词：超声成像，神经网络，波束形成，图像对比度增强，非轴向散射。

资源详情

资源评论

资源推荐

2010 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 37, NO. 9, SEPTEMBER 2018

Deep Neural Networks for Ultrasound

Beamforming

Adam C. Luchies ,

Member, IEEE

, and Brett C. Byram,

Member, IEEE

Abstract

— We investigate the use of deep neural

networks (DNNs) for suppressing off-axis scattering in

ultrasound channel data. Our implementation operates in

the frequency domain via the short-time Fourier transform.

The inputs to the DNN consisted of the separated real

and imaginary components (i.e. in-phase and quadrature

components) observed across the aperture of the array,

at a single frequency and for a single depth. Different

networks were trained for different frequencies. The out-

put had the same structure as the input and the real and

imaginary components were combined as complex data

before an inverse short-time Fourier transform was used

to reconstruct channel data. Using simulation, physical

phantom experiment, and

in vivo

scans from a human liver,

we compared this DNN approach to standard delay-and-

sum (DAS) beamforming and an adaptive imaging technique

that uses the coherence factor. For a simulated point target,

the side lobes when using the DNN approach were about

60 dB below those of standard DAS. For a simulated ane-

choic cyst, the DNN approach improved contrast ratio (CR)

and contrast-to-noise (CNR) ratio by 8.8 dB and 0.3 dB,

respectively, compared with DAS. For an anechoic cyst in

a physical phantom, the DNN approach improved CR and

CNR by 17.1 dB and 0.7 dB, respectively. For two

in vivo

scans, the DNN approach improved CR and CNR by 13.8 dB

and 9.7 dB, respectively. We also explored methods for

examining how the networks in this paper function.

Index Terms

— Ultrasound imaging, neural networks,

beamforming, image contrast enhancement, off-axis

scattering.

I. INTRODUCTION

HE delay-and-sum (DAS) beamformer is the standard

method for combining ultrasound array channel signals

into the ultrasound radio-frequency (RF) signals used to create

a B-mode image. The algorithm consists of applying phase

delays and weights to each channel before channel signal

summation. The phase delays are used to focus the beam on

receive and the channel weights are used to control the beam

characteristics, including main lobe width and side lobe levels.

Sources of ultrasound image degradation such as off-axis

scattering, multi-path reverberation, and phase aberration limit

Manuscript received January 5, 2018; accepted February 15, 2018.

Date of publication February 27, 2018; date of current version August 30,

2018. This work was supported by the National Institutes of Health under

Grant R01EB020040.

(Corresponding author: Adam C. Luchies.)

The authors are with the Department of Biomedical Engi-

neering, Vanderbilt University, Nashville, TN 37212 USA (e-mail:

adam.c.luchies@vanderbilt.edu; brett.c.byram@vanderbilt.edu).

Color versions of one or more of the ﬁgures in this paper are available

online at http://ieeexplore.ieee.org.

Digital Object Identiﬁer 10.1109/TMI.2018.2809641

the clinical use of ultrasound imaging. For this reason, numer-

ous techniques have been proposed to improve ultrasound

image quality [1]–[4]. Few of these techniques have transferred

to the clinic. One notable exception is harmonic imaging,

which discards all of the useful information at the fundamen-

tal frequency and forms images using the second harmonic

waveform generated from non-linear propagation.

Byram et al. developed a model based beamforming

method called aperture domain model image reconstruction

(ADMIRE) [4], [5]. They tuned ADMIRE to improve ultra-

sound image quality by suppressing sources of image degrada-

tion. In addition, the development of ADMIRE demonstrated

that beamforming could be posed as a nonlinear regression

problem, which suggests that a deep neural network (DNN)

might be used to accomplish the same task. This is important

because ADMIRE is relatively inefﬁcient. In contrast, DNNs

take a long time to train but once trained can be implemented

efﬁciently.

Recently, DNNs have been used to improve the state of

the art in applications such as image classiﬁcation [6], speech

recognition [7], natural language processing [8], object detec-

tion [9], and others. DNNs consist of cascaded layers of

artiﬁcial neurons. Although individual artiﬁcial neurons are

simplistic processing units, when combined in a single layer

network, they can be used to approximate any continuous

function [10]. The potential contributions of DNNs to medical

image classiﬁcation have been and continue to be investi-

gated [11]. In contrast, the impact of DNNs on medical image

reconstruction are just starting to be examined [12], [13].

The goal of this paper was to signiﬁcantly expand and report

on our previous work to integrate DNNs into an ultrasound

beamformer and to train them to improve the quality of the

resulting ultrasound images [14].

II. M

ETHODS

A. Frequency Domain Sub-Band Processing

Several methods have relied on processing ultrasound chan-

nel data in the frequency domain to offer improvements

to ultrasound image quality. For example, Holfort et al.

applied the minimum variance (MV) beamforming tech-

nique in the frequency domain to improve ultrasound image

resolution [15]. Shin and Huang [16] applied frequency-

space prediction ﬁltering in the frequency domain to sup-

press acoustic clutter and noise. ADMIRE also operates in

the frequency domain to suppress off-axis scattering and

reverberation [4], [5].

See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

LUCHIES AND BYRAM: DNNS FOR ULTRASOUND BEAMFORMING 2011

Fig. 1. Flow chart for the proposed DNN approach.

Fig. 2. Frequency domain processing, where the inputs were a single

axially gated section of channel data. Each channel was transformed

into the frequency domain using the discrete Fourier transform (DFT).

DNNs speciﬁc to each DFT frequency and trained to suppress off-axis

scattering removed clutter. The data was transformed back into the time

domain using an inverse discrete Fourier transform (IDFT). Note that the

IDFT and summation operations could be interchanged.

A ﬂow chart for the proposed DNN approach is in

Fig. 1. Frequency domain minimum variance beamforming,

frequency-space prediction ﬁltering, and ADMIRE have sim-

ilar ﬂow charts, except that the middle step is unique to

each method. In each method, a short-time Fourier transform

(STFT) is used to produce sub-band data in the frequency

domain, which is processed in some way to suppress acoustic

clutter and improve ultrasound image quality. Next, an inverse

short-time Fourier transform (ISTFT) is used to convert back

to the time-domain. In this work, DNNs were trained to

suppress off-axis scattering.

We used an STFT window length of one pulse length,

the STFT window overlap was 90%, the window was rec-

tangular, and no padding was used. The least squares ISTFT

described by Yang was used to transform back to the time

domain [17]. A diagram showing the frequency domain

processing for a single depth of channel data (i.e., axially gated

channel data from one depth) is in Fig. 2.

B. Neural Networks

We trained a different network for each frequency and the

real and imaginary components of the signals were treated as

separate inputs to the network as in Fig. 3 (a). Therefore, if the

array had N elements, the number of network inputs was 2N.

The conﬁguration of the network output was the same as that

of the network input, i.e., separated real and imaginary data.

We used feed-forward multi-layer fully connected networks

in this work. In this type of network topology, each neuron in

Fig. 3. (a) Example feed-forward multi-layer fully connected network

setup for a two element array. The inputs to the network were separated

into real and imaginary components. Information only moves from left to

right. (b) Diagram for a single neuron.

TABLE I

YPERPARAMETER SEARCH SPACE

one layer has directed connections to the neurons in the next

layer as shown in Fig. 3 (a). Depending on the number of hid-

den layers and the width of each hidden layer, a network may

have hundreds or even thousands of neurons. A single neuron

is simplistic processing unit that consists of a summation

operation over the inputs followed by an activation function

that nonlinearly maps to the output as shown in Fig. 3(b).

We performed a hyperparameter search with parameters in

the ranges indicated in Table I. A variant of stochastic gradient

descent (SGD), called Adam, was used to train the neural

networks in this work with mean-squared error (MSE) as the

loss function [18]. An early stopping criterion was employed,

where training was terminated if network performance did not

improve based on the validation loss after 20 epochs. For each

training epoch, the network was saved only if the validation

loss was the best obtained up to that point in the training.

In this way, the network with the best validation loss was

produced at the end of training.

To create and train the networks in this work, we used

Keras, a high-level neural network API written in Python [19],

with TensorFlow as the backend [20]. Training was performed

on a GPU computing cluster maintained by the Advanced

Computing Center for Research and Education at Vanderbilt

University.

Selection of the network activation function is an important

design choice and neural network activation functions are still

an active area of research [21], [22]. An activation function

剩余11页未读，继续阅读

评论收藏

内容反馈

挽挽深铃

粉丝: 14
资源: 274

长Deep Neural Networks for Ultrasound Beamforming1

评论0

最新资源

长Deep Neural Networks for Ultrasound Beamforming1

评论0

Deep Learning in Neural Networks

Learn Keras for Deep Neural Networks

Neural Networks and Deep Learning.zip

长Deep Neural Networks for Ultrasound Beamforming2

YouTube推荐系统Paper[2016]-Deep Neural Networks for YouTube Recommendations.pdf

Deep Neural Networks for YouTube Recommendations论文翻译.pdf

Deep Neural Networks for YouTube Recommendations.zip

Neural Networks and Deep Learning中文版

Deep Neural Networks for YouTube Recommendations

Neural Networks and Deep Learning

Deep Neural Networks are Easily Fooled

1-(开始与深度学习结合）Deep neural networks for acoustic modeling in speec

An Overview of Multi-Task Learning in Deep Neural Networks.pdf

Improved feature processing for Deep Neural Networks

DeepPose: Human Pose Estimation via Deep Neural Networks

Deep Neural Networks Hyperparameter tuning, Regularization and Optimization.zip

人工智能--Deep Neural Networks

学习摘要：Methods for interpreting and understanding deep neural networks

DeepPose-Human Pose Estimation via Deep Neural Networks

最新资源