Development of an Automatic Computer Program to Determine the Optimal Dental Implant Size and Position for Fibula Free Flap Surgery

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review this work. The work addresses a clinically relevant problem and demonstrates good execution. However, several points require improvement:

• The rationale why the sample size is 91 patients is absent.
• Little mentioned comparing this to the available commercial planning software
• Want further information as to the clinical validation of the 1mm bony collar requirement
• Small coverage of the limits of algorithms or cases of failure
• Increasing the inter-rater reliability analysis may be more solid (Table S1 reveal worrying disagreements, only 2 raters)
• Lacking comparison to the existing clinical workflow efficiency
• The methods section can be rearranged to be more clear
• A bit of overlapping of introduction and discussion
• Description of algorithm development process is poor - was it an iteration? What was the process of optimization of parameters?

• Cross-sectional analysis limited to 7 points - justification needed

• Lack of information of CT scanning protocols and resolution requirements
• Nothing to compare to actual surgical outcomes
• Few citations to rival automated planning systems
• More international research about fibula anatomy could be welcomed
• Lack of references to the validation of similar automated systems

Critical Issue: It has been demonstrated in Table S1 that the difference in measurements between the raters could be as high as 6.53mm and this is clinically important. This should be discussed in detail and possibly suggests the reliability issues.

Missing Elements:

• Sample size power analysis
• Relative to manual planning times
• External verification of other populations

Critical comments:

1. Algorithm Selection: The black algorithm is overwhelming (94%) which indicates that the rest of the algorithms are not necessary. It would be in the interest of the paper to concentrate on the streamlining of this one approach.
2. Validation Concerns: Table S1 with the inter-rater disagreement data indicates a difference of up to 6.53mm and this is clinically unacceptable. This should be dealt with prior to publication.
3. Clinical Translation: The article does not give a discussion of how this can fit into current surgical workflows and software systems.
4. Statistical Analysis: ICC of 0.981 appears to contradict the differences in measurements as depicted in Table S1 which are large. This calculation needs to be checked.

The fundamental principle is good and the technical implementation is promising but the problems with reliability will have to be addressed.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1. The present study focuses on virtual analysis, without assessing whether the proposed implant positions correspond to actual surgical outcomes or postoperative implant survival.

Recommendation: While this may be beyond the scope of the current study, the authors should discuss how they plan to validate the algorithm clinically. Future correlation with postoperative CT scans or implant success rates would be valuable.

2. Line 223: "58.4 ± 16.7 years" is repeated with “±” 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This is an interesting paper on the cutting edge of computerised assisted planning for jaw reconstructive surgery.  It is well written and constructed.

The aim was to describe the development and verification of an automatic program to determine the optimal size and angulation of simultaneous dental implants in functional jaw reconstruction of jaw defects with fibula free flap.

The sample size seems sufficient with 91 CT scans of the lower limbs. This is a convenience sample collated retrospectivity so the reference concerning sample size is not necessary (20 Bonet).  There was CT data of the recipient jaw and donor fibula in what one assumes is a consecutive series from January 2018 to February 2023.

The study reports the program accuracy and Inter-rater agreement measures.

The limitation of the article is perhaps its retrospective nature, which is to be accepted if the technology did not exist when the sample collection started.

Also the study data does not include the clinical aspects such as the actual position of the fibula placed in the cases, the position of the implants in relation to the arch of the mandible and the maxilla, skeletal class, dental status (edentulous, part dentate, filly dentate), the number of implants in the actual patients, the number of implants used for rehabilitation, nuisance consideration regarding site of the mandible (symphysis, body) and soft tissue components, actual time taken to complete the virtual planning in each case. There is no data to support the statement ‘The automated program could potentially reduce virtual implant planning time’. The inclusion of clinical aspects would strength the premise of the paper and add context but its lack is an acceptable limitation.

It would be very useful to have a control group or comparison group.

It is interesting to read that ‘Fibula free flap with simultaneous dental implant placement is now routinely performed for functional jaw reconstruction in our center’. This is probably true in a lot of cases where there is virtual planning and 3D models however the selection of suitable cases will depend on dental status (Edentulous, dentate, smokers, significant comorbidly, poor dental motivation and previous oral health). Also the important issue of patents completing rehabilitation, the timing to complete and the number of implants used.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thanks for the opportunity to review this revised work. The authors have responded to a few previously expressed concerns, but there remain important issues which must be resolved beforehand to ensure publication:

• Inter-rater Reliability Remains Problematic: The corrected ICC of 0.963 does not address the problem from a basic view. Measurement differences up to 6.53mm are still clinically unacceptable when it comes to planning surgery. The explanation by the authors in respect to triangular fibula morphology at the proximal end is not rationale for the magnitude of disagreement. This is a fundamental problem of reliability, which compromises clinical applicability.
• Statistical Contradiction Unresolved: The high ICC continues to defy the large measurement variations in Table S1. An ICC of 0.963 with 8% disagreements that contain clinically significant variations implies the reliability metric does not represent true variation in measurements.
• Algorithm Design Inefficiency: The black algorithm's dominance (94% selection rate) proves the other three algorithms to be largely redundant. Rather than accept this is possible work in the future, the basic design should be simplified before publication.
• Clinical Integration Pathway Unclear: While the authors do note that they are developing proprietary CAS software, concrete integration strategies with existing workflows are still lacking. The need for software switching imposes major practical barriers that are poorly addressed.
• External population validation limited to Southern Chinese cohort
• Failure cases and algorithmic limitations not sufficiently analyzed
• Lack of surgical outcome validation in spite of retrospective claims

The work solves a clinically relevant problem, but basic reliability and validation limitations remain. The magnitude of the inter-rater disagreement (up to 6.53mm) is clinically unacceptable for applications in surgical planning.

The manuscript needs either: (1) supplemental validation studies with improved inter-rater protocols, or (2) reclassification as a technical development paper with explicit limitations regarding clinical readiness.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf