Broadband Generalized Sidelobe Canceler Beamforming Applied to Ultrasonic Imaging

Round 1

Reviewer 1 Report

This paper presents broadband generalized sidelobe canceler for ultrasound imaging using simulational data. Splitting the broadband into subband using wavelet is a good approach and it shows promissing results. However, the paper is not written clearly so it needs extensive modification for clarification.

 

Background: equations are hard to follow. Some variables are not defined but used (e.g. wa, wq). Please check the equations. Proposed method: Variables in Fig. 2 do not match with the manuscript. k, K, M are mixed. Fig. 3 the three points in first row: aren't they supposed to be at the same depth? Or are they at the different depth? It seems the delays for the beamforming have errors. Can you explain why? Fig. 3, we can see the lateral resolution is improved. What about axial resolution? You need to address this although it is not improved. At least show if the axial resolution changes or not. Fig. 5, speckle noise is not noise but signal from real scatterers. Thus, you don't expect to have lower speckles in this case. Fig.6 should be comparison of Anechoic region. Please plot the cross section of the anechoic region (32 mm? depth). A sentence does not start with “And”. There are so many typos. Please check them carefully before resubmitting the paper.

Author Response

Dear editors and reviewers:

         Please see the attachment. Thank you for your advice. We have carefully modified the manuscript following your comments. The detail explanations of the revisions in the manuscript are presented point-by-point, thanks. 

Reviewer 1：

Background: equations are hard to follow. Some variables are not defined but used (e.g. wa, wq). Please check the equations.

Your suggestion is very meaningful. The explanation has been modified following your advice. The highlight equation only shows the way to achieve final weight vector . The equations for  and  are shown in following explanation, and we also indicates the definition of the  and .

                            (5)

   and  are temporal non-adaptive weight vector and adaptive weight vector, respectively.  is determined by all received echo signals, . And  is determined by the interference and noise signal, . is a  dimensional blocking matrix, which can block off the desire signal with the interference and noise signal[18]. The blocking matrix  must satisfy .

 

Proposed method: Variables in Fig. 2 do not match with the manuscript. k, K, M are mixed.

Thank you for your advice. We have modified the manuscript. We modified the Fig. 2 to show our new beamforming more accurately. And the mixing of k also has been modified, we use the variable Q to represent the number of Narrowbands, Thanks.

Fig.2 Block diagram of the broadband generalized sidelobe canceler

 

3 the three points in first row: aren't they supposed to be at the same depth? Or are they at the different depth? It seems the delays for the beamforming have errors. Can you explain why?

Dear reviewer, the points’ coordinates we set are as following. As you can see, the three points in first row are not at the same depth, they have the same distance to the center of the phased array.

 

3, we can see the lateral resolution is improved. What about axial resolution? You need to address this although it is not improved. At least show if the axial resolution changes or not.

Dear reviewer, thank you for your advice. The reason we are not show axial resolution here is that the adaptive beamforming can not improve the axial resolution. The axial resolution is determined by the center frequency of the ultrasonic , ultrasonic velocity , and the emitted ultrasonic numbers. The pictures below show the point at z = 41 mm after DS, SA, GSC, Broadband-GSC, respectively. As you can see, the beamforming algorithms can not improve the axial resolution. We can improve the axial resolution using algorithm like coded excitation.

 

5, speckle noise is not noise but signal from real scatterers. Thus, you don't expect to have lower speckles in this case. Fig.6 should be comparison of Anechoic region. Please plot the cross section of the anechoic region (32 mm? depth).

Dear reviewer, you are right that the speckle noise is not noise but signal from real scatterers. The points have bigger reflection coefficient than scatters, so the echo imaging seems get lower speckles, it is collimation error. The lateral variation of each beamforming at z = 32 mm is shown below. Broadband-GSC has almost the same results as DS, because the scatters do not have obvious higher signal-to-noise ratio (SNR) than background. For system's SNR also influences the performance of the adaptive beamformings. Reference [30] indicates that the performance of the adaptive beamforming is dependent on the system's SNR. As system's SNR decreases, imaging quality after adaptive beamformings will rapidly decrease.

Consequently, we discuss the advantages of combining plane wave emission with Broadband-GSC to improve the system's SNR.

Lateral variation of each beamforming at z = 32 mm

 

A sentence does not start with “And”. There are so many typos. Please check them carefully before resubmitting the paper.

Thank you for your advice. We have tried our best to modify the manuscript. And we will use a professional English editing service to modify our manuscript, if you can approve our research.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper aim is to improve the image quality of the medical diagnostic technologies adopting ultrasonic system. In particular, the Authors are proposing a method suppressing the interference and noise signal to improve the lateral resolution of ultrasound image adopting a broadband generalized sidelobe canceler technique. The technique was tested only on simulated data and then compared with more conventional image reconstruction approaches as the Delay-and-sum (DS) and the synthetic aperture (SA).

 

Major issue:

 

The paper aim is clear but seems an incremental work in respect to other work of the same author (i.e. Jiake Li, Xiaodong Chen, Yi Wang, Xiaoshuai Chen, Daoyin Yu. Forward-backward generalized sidelobe 257 canceler beamforming applied to medical ultrasound imaging. Aip Advances, 2017, 7: 015201

and  

Jiake Li, Xiaodong Chen, Yi Wang, Wei Li, Daoyin Yu. Eigenspace-based generalized sidelobe canceler 263 beamforming applied to medical ultrasound imaging. Sensors, 2016, 16(8): 1192).

The authors approach seems always the same.

The author must better highlight the novelty implemented in this work.

 

Minor issue:

The Broadband-GSC is applied only on simulated data obtained with the software Field II, a comparison with real acquired data is recommendable.

Author Response

Dear editors and reviewers:

         Please see the attachment. Thank you for your advice. We have carefully modified the manuscript following your comments. The detail explanations of the revisions in the manuscript are presented point-by-point, thanks.

Reviewer 2：

The paper aim is clear but seems an incremental work in respect to other work of the same author (i.e. Jiake Li, Xiaodong Chen, Yi Wang, Xiaoshuai Chen, Daoyin Yu. Forward-backward generalized sidelobe 257 canceler beamforming applied to medical ultrasound imaging. Aip Advances, 2017, 7: 015201

Dear reviewer, there are obvious improvement compared with our previous work. “Forward-backward generalized sidelobe canceler beamforming” is proposed to improve the robustness of the adaptive beamforming, we proposed a preprocessing algorithm named forward and backward subaperture averaging to achieve better sample covariance matrix .

In this manuscript we propose a new beamforming in the wavelet domain, which is more suitable for broadband echo data. And Broadband-GSC can improve the lateral resolution compared with DS and SA. Furthermore, previous works are basing on synthetic aperture (SA), which need N times motivation of the ultrasonic phased array. Proposed method in this manuscript only needs emit phased array once. Therefore, the new beamforming introduces the possibility for high frame-rate imaging and higher investigation depth with increased lateral resolution.

Following your advice, we have modified the Fig. 2 to highlight the improvement of our manuscript. Hoping you can understand our work much more obviously.

Fig.2 Block diagram of the broadband generalized sidelobe canceler

 

The authors approach seems always the same. The author must better highlight the novelty implemented in this work.

The same authors are highlight below:

Broadband Generalized Sidelobe Canceler Beamforming Applied to ultrasonic Imaging, Jiake Li, Zhe Ma, Lei Mao, Zhengjun Wang, Yi Wang, Huaiyu Cai, Xiaodong Chen, applied science

Forward-backward generalized sidelobe canceler beamforming applied to medical ultrasound imaging, Jiake Li, Xiaodong Chen, Yi Wang, Xiaoshuai Chen, and Daoyin Yu, AIP Advances

Eigenspace-Based Generalized Sidelobe Canceler Beamforming Applied to Medical Ultrasound Imaging, Jiake Li, Xiaodong Chen, Yi Wang, Wei Li and Daoyin Yu, sensors

Dear reviewer, my name is Jiake Li. Xiaodong Chen and Yi Wang are mine teachers, they help me accomplish this manuscript, Thank you.

 

Minor issue: The Broadband-GSC is applied only on simulated data obtained with the software Field II, a comparison with real acquired data is recommendable.

Dear reviewer, we also want to follow your advice. However, our laboratory can not obtain real acquired data because we don’t have a real ultrasonic imaging system. Field II is a simulation tool widely used in the medical ultrasonic imaging field, and previous proposed work in this field have indicated that the simulation echo data obtained by Field II can be used as original data to verify the imaging performance of the algorithm. So we only use the simulated data to discuss our work, hoping you can understand our difficulties, thank you very much.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Although authors answered to each comment, the answers were not fully reflected in the manuscript. The manuscript was not improved.

Author Response

Dear editors and reviewers:

       Thank you for your advice. We have carefully modified the manuscript following your comments. The detail modifications are presented point-by-point, thanks.

Modification 1：equations are hard to follow. Some variables are not defined but used (e.g. wa, wq). Please check the equations.

The explanation has been modified following your advice. The eq. 5 only shows the way to achieve final weight vector . The equations for  and  are shown in following explanation, and we also indicates the definition of the  and .

                            (5)

   and  are temporal non-adaptive weight vector and adaptive weight vector, respectively.  is determined by all received echo signals, .  is determined by the interference and noise signal, . is a  dimensional blocking matrix, which can block off the desire signal with the interference and noise signal[18]. The blocking matrix  must satisfy .

Modification 2：Variables in Fig. 2 do not match with the manuscript. k, K, M are mixed.

Thank you for your advice. We have modified the manuscript. We modified the Fig. 2 to show our new beamforming more accurately. The mixing of k has also been modified. We use the variable Q to represent the number of narrow bands, Thanks.

Fig.2 Block diagram of the broadband generalized sidelobe canceler

Modification 3： 3, we can see the lateral resolution is improved. What about axial resolution? You need to address this although it is not improved. At least show if the axial resolution changes or not.

Thank you for your advice, we have added explanation in “5. Discussions and conclusions” to explain the axial resolution will not change.

Meanwhile, beamforming algorithms will not improve the axial resolution. The axial resolution of the ultrasonic imaging system is determined by the center frequency of the ultrasonic , ultrasonic velocity , and the emitted ultrasonic numbers. The axial resolution of system can be improved by algorithm like coded excitation.

Modification 4：6 should be comparison of Anechoic region. Please plot the cross section of the anechoic region (32 mm? depth).

Dear reviewer, we have added the lateral variation of each beamforming at z = 32 mm in our new manuscript, thanks.

                    (a)                                      (b)

Fig.6 Lateral variation of each beamforming at z = 32 mm and 46.7 mm.

       Comparing the lateral variation after each beamforming at z = 32 mm, Broadband-GSC has almost the same result as DS, meanwhile, GSC generates the similar result as SA. The reason is that scatters do not have obvious higher signal-to-noise ratio (SNR) than background. System's SNR also influences the performance of the adaptive beamforming. Reference [28] indicates that the performance of the adaptive beamforming is dependent on the system's SNR. As system's SNR decreases, imaging quality after adaptive beamforming will rapidly decrease. Consequently, the advantages of combining plane wave emission with Broadband-GSC to improve both the system's SNR and lateral resolution will be discussed.

Modification 5：A sentence does not start with “And”. There are so many typos. Please check them carefully before resubmitting the paper.

Thank you for your advice. We have tried our best to modify the manuscript. Our manuscript also has been checked by a native English speaking colleague, thank you.

Author Response File: Author Response.pdf

Reviewer 2 Report

The work is now ready for pubblication.

Author Response

Dear editors and reviewers:

         Thank you for your advice. We have carefully modified the English following your comments. Manuscript also has been checked by a native English speaking colleague, thank you.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

All the comments were addressed accordingly.

Author Response

Dear editors and reviewers:

       Thank you for your advice. We have carefully modified the manuscript following your comments. The detail modifications are presented point-by-point, thanks.

“DS” is not the common acronym for “Delay-And-Sum”, it should be replaced by “DAS”.

Thank you for your advice, we have modified the manuscript. The DS is replaced by DAS.

In Eq.(12) “Broadband_Im” should be replaced by a variable, like “Bi”. Note that “Broadband_Im” it looks programming language, not a variable in an Equation. Same for “iwavelet”, it will be better to define a function to be applied on “NB(r_q)”

Your suggestion is very meaningful. The explanation has been modified following your advice. Modifications are as flowing:

Changing wavelet to DWT, iwavelet to IDWT Changing Broadband_Im to Bi

Author Response File: Author Response.pdf