All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
The authors correctly addressed all the points raised by me and by the reviewers and therefore I can recommend this article for acceptance.
[# PeerJ Staff Note - this decision was reviewed and approved by Xiangjie Kong, a PeerJ Section Editor covering this Section #]
The authors addressed the points raised by the reviewers, but some minor changes are still needed.
The binary classification results must be reported not only through accuracy, F1 score, and average precision, but by employing the Matthews correlation coefficient (MCC), sensitivity, specificity, and negative predictive value.
Links for the availability of data and of software code should be included as well.
All revision suggestions have been addressed appropriately.
All revision suggestions have been addressed appropriately.
All revision suggestions have been addressed appropriately.
Authors may consider changing the title to something more specific (Ex: Can mention the Framework name in the title)
The article cannot be accepted in its current form. The reviewers raised a number of issues that need to be addressed by the authors. Please prepare a new version addressing these issues.
1. Authors should pay attention to lines 40-46, where these statements are unclear. They can rewrite the sentences to make them flow better and ensure that the reader can understand them well.
2. In the Introduction Section, the problem statement lacks clarity. While the explanation of CNN’s approach in lines 64-69 is well-crafted, the authors should delve deeper into the shortcomings that necessitate this research.
3. The explanation about ANN is too lengthy and lacks clarity in conveying its intended message. This can be observed in lines 55-62. It’s important to note that in addition to the CNN-Based Model, Transformer-Based Models are also widely used for managing images, including breast cancer. The authors should focus on providing more concise and relevant background information. The explanation should be more precise and to the point.
4. Authors need to write BiFPN background well as a short introduction. Authors are expected to briefly discuss what the urgency and benefits of using this method.
5. In lines 85-88, the authors presented their rationale using CNN-BiFPN, but they failed to provide sufficient context or background information to support their claim. It is essential for the authors to revise their rationale and ensure that it clearly connects to the underlying problems and research gaps. Additionally, the authors should incorporate the most recent references to enhance the credibility of their research justification. The primary objective is to effectively convey the research findings and clearly outline the research gaps that need to be addressed.
6. In the related work section, the authors cited numerous Machine Learning and Convolutional Neural Network (CNN)-based models. However, they failed to identify any significant insights that can be drawn from this research. It is expected that authors can critically evaluate previous research and ultimately claim novelty in their own work.
7. For all figures, please replace it with better image quality
1. In the Proposed Framework section, Authors are requested to provide a more detailed explanation of the data collection process. This includes specifying the amount of data collected and describing the Dataset class. Additionally, it is expected that the preprocessing process is demonstrated. The significance of this stage should be explained. Authors are also expected to describe the dataset splits that were performed.
2. In section 3.3, the authors are asked to provide a more detailed explanation of whether the raw input image or the image that has undergone image enhancement is used in the process.
3. Please ensure that all formulas and metrics are explained clearly.
4. Since BiFPN was processed on a Multiscale Image, this paper has not discussed this aspect in detail. The authors are expected to provide a more comprehensive explanation of BiFPN.
5. In section 4.2, please use a table to describe the parameter settings.
6. Authors are expected to prepare step by step algorithms for existing processes
1. In the results section, please consider the significance of specificity and sensitivity as important metrics in medical imaging. Authors are expected to provide a more detailed explanation of each result displayed and to explain its relevance to real-world applications in the medical field.
2. The results of this experiment have yet to be compared with the state-of-the-art method. Please provide an explanation for this. Additionally, the authors are expected to discuss the findings in relation to the advantages and disadvantages of the proposed method.
3. Authors are expected to provide the dataset and code for this research online.
4. Conclusions are not well stated, not linked to original research question & limited to supporting results. Authors are expected to rewrite and adjust based on the suggestions above.
The author is expected to be able to explain in more detail each process in this research and claim novelty and show a comparison of the findings in this research with other research.
The manuscript presents a deep learning framework (CNBiFPN) for breast cancer detection in CT scans. The structure aligns with PeerJ standards, and the language is generally clear. However, several issues need addressing:
1. Clarity & Language: Some sections (e.g., Sections 3.3.1 and 3.3.2) contain repetitive or ambiguous phrasing (e.g., "the model adapts feature mappings from input image to the right segmentation map"). Minor grammatical errors (e.g., "themodel" in line 353) and inconsistent tense usage (e.g., shifting between past and present tense in methodology) should be revised.
2. Figures & Tables:
2.1 Figures 2–6 and 9–10 lack axis labels, captions, or legends, making interpretation difficult. For example, Figure 7’s x-axis is labeled "Epochs" but lacks units or context.
2.2 Table 1 lists baseline comparisons (CNN, CNN-SL, CNN-HD) but fails to cite or describe these methods in the text, leaving readers unfamiliar with their specifics.
1. Dataset Limitations:
The dataset (50 diseased + 50 normal images) is extremely small for deep learning, raising concerns about overfitting. No justification for the sample size or discussion of class imbalance (if any) is provided. Cross-validation or external validation on independent datasets is absent.
2. Preprocessing Details:
Steps like morphological closing and CLAHE are described superficially. Critical parameters (e.g., structuring element size for morphological operations, CLAHE clip limits) are omitted, hindering reproducibility.
1.Metrics & Statistical Analysis:
1.1 While precision, recall, F1, and accuracy are reported, confidence intervals or p-values for comparisons are absent, weakening statistical validity.
1.2 The claim of 97% accuracy based on radiologist validation lacks methodology (e.g., inter-rater agreement metrics like Cohen’s kappa).
2.Overfitting Risk: The validation loss being lower than training loss (Figure 9) is atypical and may indicate issues in training/validation split or leakage.
no comment
Correct the manuscript with a native english speaker
Update the references
Do the literature survey again
Update the paper with some mathematical equations and add some good quality figures throughout the article
The author did some good works, but failed to explain them in a better way
Comments to the author:
1. Revise the title in a better way
2. What is the novelty of the proposed work? For the last one decade, many articles were published in the relevant areas using the same kind of methodologies
3. Revise the abstract section
4. Add how much % improvement has been achieved by the proposed method over the existing methods
5. Revise the keywords section
6. State of the art is missing in the introduction section
7. Also cite all the articles given in the reference section and also align it alphabetically in order or arrange the references in such a way that how it’s being cited in the paper
8. Explain the drawbacks of the existing methods and also add the workflow of the paper, just before explaining the proposed work
9. Is it conclusion or conclusions?
10. Revise the conclusion section
11. Update the references and add some recent references and cite them accordingly
12. Most of the figures are blurry. Replace all the figures with good quality figures and with high resolution
13. In fig 7, define X and Y axes properly
14. Fig 8 is incomplete. Use some software tools for drawing figures
15. In fig 9, define the axes
16. Fig 10 remains incomplete
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.