Review Reports

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This work proposes a novel method for neural action potential detection by adopting a high-order nonlinear energy operator (HONEO) and an analog CMOS implementation, aiming to address the challenges of low signal-to-noise ratio in noisy neural recording environments. The paper's core contributions include: the development of a HONEO-based detection algorithm that enhances spike features while preserving waveform characteristics; the design of a dual-threshold detection mechanism capable of accurately identifying both positive and negative peaks to improve detection robustness; and the implementation of a low-power analog APD circuit, including differentiators, multipliers, and adaptive threshold generators, suitable for real-time, implantable neural interface systems. Befire publication, several questions should be resovled by authors to further enhance the manuscript:

（1）The manuscript claims that low power consumption is a key advantage of the proposed method. However, the experimental evaluation only includes comparisons in terms of accuracy, while lacking quantitative analysis of latency and power consumption across different methods. Including such metrics would significantly strengthen the validity of this claim.

（2）The primary distinction between the proposed HONEO method and NEO appears to lie in the selection of the parameter (K), indicating that this parameter plays a critical role. However, the manuscript only briefly states that (K=3) is chosen based on implementation probability and power consumption, without providing sufficient justification. A more comprehensive analysis is needed, such as presenting the trade-offs in power consumption and accuracy for different values of (K), to better support this design choice.

（3）In Table 1, the performance differences between HONEO and NEO are not sufficiently significant, which makes the advantages of the proposed method less convincing. Additional analysis or clearer presentation of the improvements would help highlight the contribution.

（4）The manuscript contains several formatting issues. For instance, some figures are not properly centered. In addition, in Section V (Conclusion), the final occurrence of “HONEO”is incorrectly formatted, with the letter“O”not capitalized.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript presents an action potential detector (APD) based on a high-order nonlinear energy operator (HONEO). The proposed architecture consists of a HONEO block, two adaptive threshold generators (for positive and negative peaks), and an XOR-based decision stage. This approach enables the detection of both polarities of the action potential and allows estimation of the spike half-width. The use of adaptive thresholds also contributes to improving the signal-to-noise ratio (SNR).

Overall, the manuscript is clearly written and technically sound. The reported performance — including a 14–20 dB SNR improvement, 97% detection accuracy, and 62 µW power consumption in a 0.18 µm CMOS process — is relevant for low-power neural interface applications. The paper can be recommended for publication after minor revisions.

 
Major Comments
1) OTA topology (Fig. 4)
The OTA shown in Fig. 4 appears to adopt a mirrored operational amplifier architecture; however, the motivation for this choice is not sufficiently discussed. Such a topology is generally not optimal in terms of noise and power efficiency compared to more conventional solutions.

The authors are encouraged to justify this design decision more clearly, possibly by comparing it with alternative implementations such as a differential pair with active load or a folded cascode OTA. Given the importance of noise performance in spike detection, a quantitative discussion of the input-referred noise and associated trade-offs would significantly strengthen the manuscript.

2) Comparison with prior work
While the manuscript demonstrates improvements over the conventional NEO approach, the comparison with existing spike detection methods is limited.

The authors are encouraged to expand the discussion and include a more comprehensive comparison with prior work, including:

SNEO-based detection approaches for implantable arrays (Electronics, 2021)
Noise-based adaptive thresholding methods (Rizk and Wolf, 2009)
In particular, it would be beneficial to clarify:

the robustness of the proposed HONEO approach to noise,
the advantages of the adopted adaptive thresholding with respect to noise-scaled thresholds,
and the trade-offs with respect to more advanced detection and sorting methods.
 
Minor Comments
The sentence “To be fabricated with a 0.18 µm CMOS process…” should be rephrased for clarity (e.g., “The circuit is designed in a 0.18 µm CMOS process…”).
Please clarify whether the reported ±100 mV output stability is based on post-layout simulations and specify the considered PVT conditions.
The use of a 1 kHz test signal should be justified with respect to typical neural spike characteristics.
Reporting the input-referred noise density would improve the completeness of the results.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This paper proposes an action potential detector (APD) based on a high-order non-linear energy operator (HONEO).

The topic is interesting, and the manuscript is quite well written.

However, the proposed system was designed and simulated at the schematic level, using a standard 180 nm CMOS process. Unfortunately, no post-layout  simulations or measurements were provided that could have further validated and strengthened the proposed system.

Furthermore, the authors did not consider how statistical device variations (device mismatch) could affect the performance of the proposed design.

Finally, in Table 2, Performance comparison with reported APDs, authors should indicate whether the APDs were simulated at the schematic level, at the post-layout level, or measured.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have addressed enough of my comments and suggestions.

In addition, the authors further improved this manuscript by answering other reviewers’ questions and revising the manuscript.

As a result, this manuscript now can be accepted for publication in its present form.