A 48-Channel High-Resolution Ultrasound Beamforming System for Ultrasound Endoscopy Applications
Luigi Fortuna
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThe paper is appropriate for the journal. It regards a system for signal processing in Endoscopy advanced application.
The paper is well introduced from the medical point of view with more details.
The architecture of the system is presented and also the electronic implementation on chip is included.
The merit of the authors is to presents a systems with some improvement respect the existing ones. The results are well presented.
The paper is a good design project implementation and in my opinion even is scientific innovative aspects respect the electronics are not widely emphasized it represents a very useful application example.
Moreover the authors must stress the electronic scientific items and also to include more items respect a possible fully innovative implementation of image signal processing. I suggest to look for this aspect to the Cellular Neural Network approach that is truly suitably coupled to the presente hardware. The authors could look at the following contribution that must be included in the references as a key point paper:
Nuclear Instruments and Methods in Physics Research, Section A: Accelerators, Spectrometers, Detectors and Associated EquipmentVolume 497, Issue 1, Pages 174 - 17821 January 2003 Breats Imaging18 April 2001through 21 April 2001Code 60651 Document type Conference Paper Source type Journal ISSN 01689002 DOI 10.1016/S0168-9002(02)01908-3 HGImage processing for medical diagnosis using CNN
- Arena, Paolo;
- Basile, Adriano;
- Bucolo, Maide;
- Fortuna, Luigi
Globally my opinion is positive.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
Comments and Suggestions for AuthorsReview: A 48-Channel High-Resolution Ultrasound Beamforming System for Ultrasound Endoscopy Applications
This paper presents a 48 channel ultrasound beam forming system in combination with SCF algorithm for high resolution imaging.
The article is well presented, literature self contained to the problem itself, and well written.
However there are some miss in this work, namely the evaluation of the impact of the number of channel into the required quality of the imaging process using the current beam forming proposal.
The concept of SCF is not new, its appear in some recents publications, for example
https://www.sciencedirect.com/science/article/abs/pii/S0301562922006378
How the proposed SCF compares against those in the state of the art?
Comments on the Quality of English Languagemore to the point sentences, paper needs some polishing, more direct sentences.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for AuthorsThe paper describes an ASIC that realizes a full ultrasound front-end for endoscopic applications. The paper is, in general, well organized and complete. The English is good. The paper includes a good introduction, description of the work, and experiments support the authors’ claims. Some part needs improvement. Below my comments.
- The ASIC is designed for PMUT transducers. Please add a description of the transducer employed in experiments.
- In Fig.1 the transducer seems a “D array (like PMUT transducers often are). How are the elements arranged in the connections to the 48 ASIC channels?
- Several picture are included without a clear explanation of what it is represented. For example Fig.2, but also the others. The blocks present in the figure should be explained in the text.
- Is RX dynamic focus possible in the beamformer ?
Other points:
L.55. We read “However, … generate dynamic power dissipation from the large parasitic capacitance from the transducers…” Capacitance does not dissipate power. Only resistance does.
L53 The charge redistribution method for transmission should be briefly explained. Reference only is not sufficient
Fig. 2 should be explained. Is the right section of Fig.2 part of a standard transmitter?
Fig. 4. Internal Labels should be in English
L.86 Usually digital beamformer is calculated on FGPAs or GPUs
L. 170-172. Not clear to me. Please explain what the additional calculation block does.
L.234-242 Is the procedure here described a method for approximating a division, or totally e different method. Pleas clarify.
L.250 What the “power loss of capacitance” is? An ideal capacitance does not loose power.
L256-257. Axial resolution is determined by several factors, the TX frequency is just one of these. Very important as well are, for example, the transducer bandwidth, the RX channel bandwidth, etc…
Fig 15 and 16. I suggest to add the grayscale with the dB scale.
L264. What does “while viewing the sidelobes” mean?
Figure 16 shows more than 30dB of dynamic difference between DAS and proposed method, but the Bmode images seem not show such a difference.
L292. What does”..by producing the echo signal as a matrix based on the reference image” mean? Is that an image acquired from the ASIC?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for AuthorsAuthors have addressed the questions i raised.
Comments on the Quality of English LanguageMinor typos
Author Response
There are no comments.
Reviewer 3 Report
Comments and Suggestions for AuthorsI have this last comments:
L. 57 “power by recycling the charge by discharging by shorting both …” please rephrase
Fig. 2 caption is : “Conventional TX Pulsers”. But this seems not the content of the figure, where “redistribution method” is still represented.
L.181-182 not clear to me. “The TX delay is calculated based on a focus of half the scan depth per scan line”. This is OK for fixed depth TX focus (fixed at half length of the line). “…for RX dynamic focusing” This is unclear. RX dynamic focusing mean that the focus depth is adjusted per receiving depth: i.e. during reception the focus is dynamically changed to focus EACH point of the scanline along depth. Alternatively, in an easier approach, the RX focus is fixed at a depth (like in TX). Dynamic focusing is what a standard (and complex) echograph does. This needs a continuous recalculation of RX focusing delays, that changes for every RX pixel along the scanline. My impression is that your architecture, without memory, cannot perform dynamic RX focusing. Can your ASIC produced dynamic focusing ?
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
