Atlas-Based Shared-Boundary Deformable Multi-Surface Models through Multi-Material and Two-Manifold Dual Contouring

Round 1

Reviewer 1 Report

In this paper, a modified Dual Contouring has been introduced that is able to produce error free surface meshes. These meshes are multi-material as well as 2-manifold in the sense that the sub-meshes of individual materials are, by themselves, watertight and 2-manifold, even though the whole mesh can contain non-manifold elements along material interfaces. The sub-meshes have consistent shared boundaries. Each triangle of the mesh is identified using pairwise material indices. The proposed method is effective in generating geometrically correct, as well as accurate representation of anatomical structures. In general, this paper is well written and easy to follow. I would like to accept this paper if my following concerns are carefully addressed.

(1) The authors need to emphasise their contributions/novelties in the revision. In the current version, the authors did not discuss their contributions in detail.
(2) The proposed algorithm still can be improved if the ideas in the following papers are explored, i.e., "Making Sense of Spatio-Temporal Preserving Representations for EEG-Based Human Intention Recognition", "An Adaptive Semisupervised Feature Analysis for Video Semantic Recognition", and "A Semisupervised Recurrent Convolutional Attention Model for Human Activity Recognition". The authors are encouraged to discuss them in the revision.
(3) The authors should carefully proofread this paper and correct all the typos in the revision. In the current version, there are still some typos/grammar errors.
(4) Could the authors report the running time of the proposed algorithm? In this way, we can justify whether this algorithm can be applied to large-scale dataset.
Based on the above comments, I would like to accept this paper with minor revision.

Author Response

Reviewer 1:

(1) The authors need to emphasise their contributions/novelties inthe revision. In the current version, the authors did not discuss theircontributions in detail.

A new section is added, 2.1 Overview of Method and Summary of Contributions, which addresses this comment. Other subsections are renumbered accordingly.

(2) The proposed algorithm still can be improved if the ideas in the following papers are explored, i.e., "Making Sense of Spatio-Temporal Preserving Representations for EEG-Based Human Intention Recognition", "An Adaptive Semisupervised FeatureAnalysis for Video Semantic Recognition", and "A Semisupervised Recurrent Convolutional Attention Model for Human Activity Recognition". The authors are encouraged to discuss them in the revision.

The first two papers are cited in the newly added last paragraph in section 4. Discussion along with a similarly relevant publication of the senior author’s. 

(3) The authors should carefully proofread this paper and correct all the typos in the revision. In the current version, there are still some typos/grammar errors.

We thank Reviewer 2 for the suggestion. Another parsing exercised turned up a few mistakes, which we strived to correct.

(4) Could the authors report the running time of the proposed algorithm? In this way, we can justify whether this algorithm can be applied to large-scale dataset.

The running time depends on the size of the image and the choice of grid size. We added a line on complexity at lines 269-70.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper considers the surgical navigation of an intra-operative MRI-compatible surgical robot and proposes a multi-material Dual “Contouring” (DC) method that is able to transform a digital 3D voxel-based atlas of basal ganglia into a deformable discrete multi-surface model, which can improve the computational efficiency for intra-operative deformation. The proposed method is utilized for the initialization of a deformable multi-surface Simplex model for multi-surface atlas-based segmentation. The effectiveness of the proposed method on synthetic and deep brain atlas data is demonstrated and compared with that of the traditional DC. The proposed approach is original and may have its own merits in clinical practice. However, the paper is not written very well and the following issues need to be addressed before it can be accepted for publication.

1.     The authors may need to clearly define the problem they would like to solve in Section 2 before the DC algorithm for contouring is presented.

2.     It would be better if a figure can be created to sketch the steps of the proposed approach before the technical details of the approach are presented.

3.     Many equations in the text need much better explanations. For example, Equations (1) and (2) (in page 6) suddenly appear in the text and the meanings of symbols x, pi, Ni, A in the equations need to be clearly explained.

4.        Section 2.1 needs to be made more concise since the majority of the text should be dedicated to the approach the authors would like to propose in the paper.

5.     In Equation (4) (in page 15), how can one determine the values of Fin and Fext? Are they related to the j’s in Equations (9) and (10)?

6.  In the abstract, the authors claim that a comparison between the proposed approach and the traditional DC was made. However, I could not find the comparison in the text. I believe the authors may need to create a table to clearly show the result of the comparison and clearly describe the performance indices they have used to evaluate the effectiveness of both methods in the text.  

Author Response

Reviewer 2.

1. The authors may need to clearly define the problem they would like to solve in Section 2 before the DC algorithm for contouring is presented.

A new section is added, 2.1 Overview of Method and Summary of Contributions, which addresses this

comment. Other subsections are renumbered accordingly. Also, the geometry involved is made explicit with a new Figure 3.

2. It would be better if a figure can be created to sketch the steps of the proposed approach before the technical details of the approach are presented.

A new Figure 2 is added which lays out the workflow of the algorithm.

3. Many equations in the text need much better explanations. For example, Equations (1) and (2) (in page 6) suddenly appear in the text and the meanings of symbols x, pi, Ni, A in the equations need to be clearly explained.

We sought to heed this comment by referring that text to Figure 3, as well as making explicit every symbol in every expression. This has resulted in adding a few equations in the deformable simplex section. 

4. Section 2.1 needs to be made more concise since the majority of the text should be dedicated to the approach the authors would like to propose in the paper.

This section is now 2.2 and has experienced some cuts. We did not want to completely eliminate the section on ambiguous cases, however, since this is one of the refinements over our previous publications.

5. In Equation (4) (in page 15), how can one determine the values of Fin and Fext? Are they related to the j’s in Equations (9) and(10)?

This shortcoming is addressed in equations 5 to 12, by defining an ideal point for each force  and  .

6. In the abstract, the authors claim that a comparison between the proposed approach and the traditional DC was made. However, I could not find the comparison in the text. I believe the authors may need to create a table to clearly show the result of the comparison and clearly describe the performance indices they have used to evaluate the effectiveness of both methods in the text.

The abstract is refined in relation to previous claims, but the advantages of Dual Contouring over Marching Cubes are well described in the paper by Ju (citation 8). What our multi-material method offers over single-surface DC is made explicit in the paper.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

I have no other concerns and recommend the acceptance of the paper.