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 article is accepted for publication.
[# PeerJ Staff Note - this decision was reviewed and approved by Xiangjie Kong, a PeerJ Section Editor covering this Section #]
The issues have been addressed.
The issues have been addressed.
The issues have been addressed.
Kindly revise the manuscript as per the reviewer suggestions and resubmit it.
The authors have dealed with my concerns. I recommend accept this article.
no comment
no comment
no comment
1. The resolution of the words in Figure 2 is still low.
2. The author still does not print the title of the x-axis and y-axis for Figures 3, 4, 5, and 6.
3. The author explains that the number 138437456055600 in Figure 7 is the number of nodes in the pipeline. What are the nodes? The 138437456055600 nodes as input of neural network? That's quite weird. Usually, the input of a neural network is the shape of the data rather than the "node".
Based on the code provided by author, the CNN-Resnet does not connect the CNN and Resnet:
https://www.kaggle.com/code/anubhavde/skinsentry-mark-v
"""
model = ResNet50(weights='imagenet', include_top=False, input_shape=(75, 100, 3))
x = model.output
x = GlobalAveragePooling2D()(x)
x = Dense(1024)(x)
x = LeakyReLU()(x)
x = Dense(1024)(x)
x = LeakyReLU()(x)
predictions = Dense(num_classes, activation='softmax')(x)
model = Model(inputs=model.input, outputs=predictions)
"""
The author only added some Dense layers for ResNet50 rather than CNNs. I suggest the author check the code to see whether there are errors here.
The issues have been addressed.
Based on the referee reports, I recommend a major revision of the manuscript. The author should improve the manuscript, taking carefully into account the comments of the reviewers in the reports, and resubmit the paper.
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
**PeerJ Staff Note:** Please ensure that all review, editorial, and staff comments are addressed in a response letter and that any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.
**Language Note:** The review process has identified that the English language must be improved. PeerJ can provide language editing services - please contact us at [email protected] for pricing (be sure to provide your manuscript number and title). Alternatively, you should make your own arrangements to improve the language quality and provide details in your response letter. – PeerJ Staff
This article is written in good English and do a good job in dermatological classification of histopathological skin images. It addresses challenges like overfitting, which is an important issue in deep learning. Some discussions of the related works could be strengthened. For example, please discuss the differences between your work and:
1. "K Chen, et.al. PCCT: Progressive class-center triplet loss for imbalanced medical image classification" 2. "S Zhao, et. al. Reasoning discriminative dictionary-embedded network for fully automatic vertebrae tumor diagnosis"
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
The authors carry out sufficient experiments.
The proposed method is validated by several datasets, which, in my opinion, meets the standards of PeerJ.
no comment
1. There are a few grammar or verbal problems that should be addressed:
(a) In the article, the authors sometimes use the third person to describe their work (e.g. the researcher/they in session 1, 3.2, and 3.3) but sometimes use the first person (e.g. we/our in the abstract and session 1 and 7). The authors also use the third person to describe previously published works. It makes readers hard to distinguish the object that the author is talking about. So I suggest the author use "we/our" to discuss their own work and use "they" to describe others' work.
(b) line 232, "To obtain the ground truth labels for the dataset the dermatologists individually reviewed the dermatoscopic images" A comma should be added to the sentences (...dataset, the...)
2. Some figures can be further improved:
(a) The words in Figure 2 are too small to be seen.
(b) The author does not print the title of the x-axis and y-axis for Figures 3, 4, 5, and 6.
(c) What is the meaning of 138437456055600 in Figure 7? The author does not clarify the meaning of this number.
(d) I suggest the author provide the model plot of CNN-RestNet and CNN-DenseNet, which can help reader understand the hybrid model better.
1. The author does not provide the version of tensorflow/keras and other dependent packages.
2. I cannot find the early stop criterion for CNN-DenseNet
3. The structures of CNN-DenseNet and CNN-ResNet are not clearly defined. From the paragraph of line 333 (session 4.2) and session 4.3, I can hardly understand how CNN and DenseNet/ResNet are connected. As the authors said, the densenet-121 is append by a global average pooling layer, two fully connected layers with 1024 units, a ReLU activation function, and a dense layer with the number (lines 333-337) but there is no description of the CNN here. There is also no description of the structure of CNN-ResNet in session 4.3. So I suggest authors to clarify these details. If authors can provide code files, that will be much better for reproducibility.
4. I cannot find the fine-tune detail of CNN-DenseNet/CNN-ResNet. The author just mention the word "fine-tune" in line 273 and 328 but they do not mention how they apply to CNN-DenseNet/CNN-ResNet
The author has compared the accuracy of their CNN-DenseNet model and others' models. However, these models are not trained by the same dataset or the data are not under the same preprocessing. So, the performance of these models is not comparable. The author should use their own dataset to train and test those models to make the accuracy comparable.
Predicting skin diseases from dermatoscopic images is a meaningful topic for human health. The author proposed a CNN-DenseNet model to accurately predict skin diseases. Compared to the existing models, it shows great improvement in this task. However, the authors did not describe the model clearly and the evaluation of their model is not sufficiently reliable. I suggest author address these issues to improve their works.
1. The abstract is very long. Please shorten it.
2. The abstract must summarize these points: what is the problem and challenge?
3. The paper needs to be revised by an English expert.
4. The paper needs to be revised by a dermatologist expert.
5. The author needs to shine on the novelty of the proposed method to be readable and more precise.
6. I advise authors to merge the subsection “Convolutional Neural Networks” to the main section “Deep Neural Network Model Construction.
7. The literature in the “Hybridization Models and Evaluations” section must be moved to related work, and I must talk about these limitations.
8. To enrich the literature, I encourage the authors to discuss the pros and cons of recent papers such as:
- Skin-Net: a novel deep residual network for skin lesions classification using multilevel feature extraction and cross-channel correlation with detection of outlier
Journal of Big Data, vol. 10 (1), Article No. 105, 2023.
- Refined Residual Deep Convolutional Network for Skin Lesion Classification
Journal of Digital Imaging, vol. 35 (2), pp. 258-280, 2022.
- Classification of skin lesions into seven classes using transfer learning with AlexNet
Journal of digital imaging, vol. 33, pp. 1325-1334, 2020.
- Skin melanoma classification using ROI and data augmentation with deep convolutional neural networks
Multimedia Tools and Applications, vol. 79, pp. 24029-24055, 2020.
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
9. The section “Data Visualization and Processing” is unclear and misunderstood.
10. What is meant by procure dataset?
11. Section 5 is divided into several subsections. I think this division is not meaningful.
12. The output of the pre-processing step must be clarified by skin image before and after each stage.
13. Used equations to measure the performance of the proposed method need to be added.
14. The symbols of all equations must be defended accurately.
15. What are the strengths and weaknesses of the proposed method?
16. A discussion of the results should be added.
17. A confusion matrix must added.
18. ROC must be added.
19. The conclusion is very long and needs to be summarized.
No comments
No comments
No comments
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.