Intelligent Computer-Aided Diagnostic System for Magnifying Endoscopy Images of Superficial Esophageal Squamous Cell Carcinoma

Comment #1

1.     More related works should be cited and compared (e.g., compared with deep learning based methods, Everson, M., et al. "Artificial intelligence for the real-time classification of intrapapillary capillary loop patterns in the endoscopic diagnosis of early oesophageal squamous cell carcinoma: A proof-of-concept study." United European gastroenterology journal 7.2 (2019): 297-306).

Reply #1

We have cited two recent studies on artificial intelligence in the endoscopic diagnosis of SESCCs and have compared our results with those reported by these two studies. The most important difference is that the main outcome/objective of these two studies is to differentiate between the abnormal (i.e. SESCC) and normal, while out study seeks to classify the B1, B2, and B3 types in SESCCs. We have added the following paragraph describing the above issues in the Discussion section.

A recent study on artificial intelligence for the classification of IPCL patterns in the endoscopic diagnosis of SESCCs analyzed 7046 magnifying endoscopy images from 17 patients (10 SESCC, 7 normal) using a convolutional neural network (CNN) [19]. The accuracy, sensitivity and specificity for abnormal IPCL patterns were 93.7%, 89.3%, and 98%, respectively. Another recent study introduced a computer-aided diagnosis system for the real-time automated diagnosis of precancerous lesions and SESCCs in 6473 narrow-band imaging images [20]. The sensitivity and specificity for diagnosing precancerous lesions and SESCCs were 98.0% and 95.0%, respectively. These outcomes were also observed in video datasets. However, the main output/outcome of these two studies was to only differentiate between the normal and abnormal. Detailed classification into the B1, B2, and B3 types, as presented in this study, was not reported. In addition, the non-B1 type was sub-classified into the B2 and B3 types using the microvessel thickness measurement method developed in this study.

19.  Everson M.; Herrera L.; Li W.; Luengo IM.; Ahmad O.; Banks M.; Magee C.; Alzoubaidi D.; Hsu HM.; Graham D.; Vercauteren T.; Lovat L.; Ourselin S.; Kashin S.; Wang HP.; Wang WL.; Haidry RJ. Artificial intelligence for the real-time classification of intrapapillary capillary loop patterns in the endoscopic diagnosis of early oesophageal squamous cell carcinoma: A proof-of-concept study. United European Gastroenterol. J. 2019, 7, 297-306.

20.  Guo L.; Xiao X.; Wu C.; Zeng X.; Zhang Y.; Du J.; Bai S.; Xie J.; Zhang Z.; Li Y.; Wang X.; Cheung O.; Sharma M.; Liu J.; Hu B. Real-time automated diagnosis of precancerous lesions and early esophageal squamous cell carcinoma using a deep learning model (with videos). Gastrointest. Endosc. 2020, 91, 41-51.

Comment #2

2.     As SVM is a machine learning technique widely used in many fields, the author could shorten Section 2.2 and just give a short review of this method.

Reply #2

We have now shortened Section 2.2 as suggested (indicated by the test in red).

Comment #3

In Section 3, it would be better to make the contribution of the manuscript clearer.

Reply #3

We have now added the importance of fuzzy stretching and the ART2 procedure in the implemented software. The following sentences have now been included in the first part of section 3.  

Efficient and accurate noise removal is essential as the proposed method relies on pixel clustering to separate the target organ area. Fuzzy stretching and ART2-based quantization procedures are applied to separate the microvessel area accurately, as explained in sections 3.1 and 3.2.

Comment #4

The data set in experiments is limited (only 114 cases in total, especially B3 type). The proposed method should be tested on a larger data set as well as compared with other related works in order to demonstrate its robustness, effectiveness and superiority.

Replies #4

We completely agree with the above comment. As the incidence of esophageal cancers was 5.9 per 100,000 population in 2013 and the proportion of SESCCs in whole esophageal cancers is <10% in Korea, the dataset has a relatively small number of images. Future work will involve conducting a large-scale, multi-center study to validate the diagnostic ability of the proposed system to predict the invasion depth of the tumor in SESCCs. We have now added the below paragraphs to the Discussion section explaining the limitations of our study and highlighting the importance of our work.

This study has several limitations. First, the evaluated dataset had a relatively small number of images (114), and all magnifying endoscopy images were retrospectively obtained from a single center. The age-standardized incidence rate was 5.9 per 100,000 population in 2013, and the proportion of SESCCs in whole esophageal cancers was less than 10% in Korea [21]. Furthermore, only high-quality magnifying endoscopy images were employed. The proposed system might produce erroneous results for images that are blurred or out-of-focus. Therefore, only a small number of SESCC images were included in this study. Therefore, the ability of the proposed automatic classification method to evaluate poor-quality endoscopy images should be investigated after validating its performance on a large dataset of high-quality images.

Despite these limitations, the results demonstrated the ability of the proposed machine learning based computer-aided diagnostic system to obtain objective data by analyzing the pattern and caliber of the microvessels with acceptable performance. Future work will focus on a large-scale, multi-center study to validate the diagnostic ability of the proposed system to predict the invasion depth of the tumor in SESCCs.

Comment #1

The abstract need to be structured and to be shorter. The document is well structured, the methods carefully described, the results are clear and support the conclusions. The authors should refrain from using personal pronouns such as “we” or “our”, it is inelegant in scientific writing.

Replies #1

Thank you for the comment. We have now revised the abstract and avoided the use of personal pronouns in the main text.

Comment #2

There is only one reference of 2018 and none after that date, we are in 2020, that might be some similar studies over the last two years which are not cited.

Replies #2

We have cited two recent studies on artificial intelligence based endoscopic diagnosis of SESCCs and have compared our results with those reported by these two studies.

A recent study on artificial intelligence for the classification of IPCL patterns in the endoscopic diagnosis of SESCCs analyzed 7046 magnifying endoscopy images from 17 patients (10 SESCC, 7 normal) using a convolutional neural network (CNN) [19]. The accuracy, sensitivity and specificity for abnormal IPCL patterns were 93.7%, 89.3%, and 98%, respectively. Another recent study introduced a computer-aided diagnosis system for the real-time automated diagnosis of precancerous lesions and SESCCs in 6473 narrow-band imaging images [20]. The sensitivity and specificity for diagnosing precancerous lesions and SESCCs were 98.0% and 95.0%, respectively. These outcomes were also observed in video datasets. However, the main output/outcome of these two studies was to only differentiate between the normal and abnormal. Detailed classification into the B1, B2, and B3 types, as presented in this study, was not reported. In addition, the non-B1 type was sub-classified into the B2 and B3 types using the microvessel thickness measurement method developed in this study.

20.  Everson M.; Herrera L.; Li W.; Luengo IM.; Ahmad O.; Banks M.; Magee C.; Alzoubaidi D.; Hsu HM.; Graham D.; Vercauteren T.; Lovat L.; Ourselin S.; Kashin S.; Wang HP.; Wang WL.; Haidry RJ. Artificial intelligence for the real-time classification of intrapapillary capillary loop patterns in the endoscopic diagnosis of early oesophageal squamous cell carcinoma: A proof-of-concept study. United European Gastroenterol. J. 2019, 7, 297-306.

21.  Guo L.; Xiao X.; Wu C.; Zeng X.; Zhang Y.; Du J.; Bai S.; Xie J.; Zhang Z.; Li Y.; Wang X.; Cheung O.; Sharma M.; Liu J.; Hu B. Real-time automated diagnosis of precancerous lesions and early esophageal squamous cell carcinoma using a deep learning model (with videos). Gastrointest. Endosc. 2020, 91, 41-51.

Comment #1

1.     The labels of the equations should be aligned

Reply #1

2.     We have checked and correctly aligned the labels of the equations.

Comment #2

2.     The equation itself should also be aligned (e.g., Equ 19)

Reply #2

3.     We have checked and correctly aligned the equations themselves.

Comment #3

The subtitles should be in the same page as the subfigures (e.g., Figure 7)

Reply #3

4.       We have rearranged all the figures and their subtitles as recommended by the reviewer.

Comment #4

In Figure 9, the ‘No’ label near ‘All Pattern’ should be on the right side

Replies #4

We have changed Figure 9 as follows:

Comment #5

The algorithms described in Table 1 and Table 2 should be standardized

Replies #5

In Table 1, we briefly describe the contour tracing algorithm, as depicted in Figure 12. Generally, in papers related to imaging processing, it is common to describe the mask tracing direction technique as depicted in Figure 12. In Table 2, we show how we measured the thickness in our study. Because the method of thickness measurement is not an algorithm, we think it is appropriate to express it as Table 2 rather than as a pseudo code.

Comment #6

Please use the same font for the same parameter (e.g., distance D in line 367-369)

Replies #6

We have used the same font for the abstract and main text as indicated by the reviewer.