Review Reports

Response to Reviewer 1 Comments

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes

The Introduction was revised to clarify the study gap and to state explicitly that CEUS was not performed.

Are all the cited references relevant to the research?

Yes

The references were retained and the Discussion wording was revised to avoid implying direct comparison with CEUS.

Is the research design appropriate?

Yes

We clarified the retrospective exploratory design, the multi-platform setting, and the limitations of the mixed reference standard.

Are the methods adequately described?

Yes

The Methods were expanded to specify follow-up modalities, follow-up assessment, platform allocation, and probe-related handling.

Are the results clearly presented?

Yes

The Results and tables were revised to improve clarity and to report platform distribution and diagnostic performance more transparently.

Are the conclusions supported by the results?

Can be improved

The Conclusion was made more cautious and now emphasizes the need for validation in larger prospective studies.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: Did they perform contrast-enhanced US? No, they explicitly did not. MVFI is presented as a contrast-free alternative. They mention CEUS only in the introduction/discussion as a comparison point, noting that CEUS shows heterogeneous enhancement and early washout in malignancies, but small lesions often show homogeneous enhancement regardless.

Response 1: Thank you for this important question. No, contrast-enhanced ultrasonography was not performed in this cohort. To make this explicit, we added a clear statement in the Introduction that “In this study, we did not perform contrast-enhanced ultrasonography; instead, we evaluated MVFI as a contrast-free approach to vascular assessment” . We also clarified this limitation in the Discussion.

Comments 2: Can they match MVFI with CEUS results? Since they didn't perform CEUS, no matching is possible. This is actually a missed opportunity – correlating MVFI findings with CEUS in the same lesions would have strengthened the paper considerably.

Response 2: Thank you for this valuable suggestion. We agree that direct correlation between MVFI and CEUS findings in the same lesions would have strengthened the study. However, CEUS was not performed in the present cohort; therefore, such a comparison was not possible. We have acknowledged this point as an additional limitation in the revised Discussion.

Comments 3: Follow-up methods and duration: Described in section 2.1 – non-malignant lesions were followed with TUS or other imaging modalities for at least 1 year. However, the exact intervals (e.g., every 6 months?), who interpreted follow-up scans, and what specific "other modalities" were used (CT? MRI?) are not specified.

Response 3: Thank you for this helpful comment. We have revised the Methods section to clarify the follow-up strategy for lesions without histopathological confirmation. Such lesions were classified as non-malignant when follow-up imaging performed at least 1 year after the initial examination showed no change in size or morphology. Follow-up modalities included TUS, EUS, contrast-enhanced CT, and MRI. The number and timing of follow-up examinations within the first year varied according to the individual clinical course, and lesion stability was assessed by comparison with the previous imaging findings.

Comments 4: Convex vs. linear probes: Section 2.2 mentions both were used – high-frequency linear probes for superficial lesions, convex for deeper ones. However, no subgroup analysis comparing diagnostic performance between probe types is provided. This is a limitation worth noting.

Response 4: Thank you for raising this important point. We have now included the distribution of probe types in Table 1. The frequencies of convex-type and linear-type probe use did not differ significantly between the invasive malignant and non-malignant groups. On this basis, we did not perform a separate subgroup analysis of diagnostic performance according to probe type. Nevertheless, we acknowledge that the absence of a probe-specific subgroup analysis is a limitation of the present study, and we have added this statement to the Discussion.

Comments 5: You state MVFI "enables visualization of low-flow vessels without contrast agents" – this is accurate, but please clarify whether any patients in this cohort also underwent contrast-enhanced US for comparison. If not, consider adding this as a stated limitation.

Response 5: Thank you for this valuable comment. No patients in this cohort underwent contrast-enhanced ultrasonography. This was because the aim of the present study was to establish the diagnostic utility of a contrast-free vascular imaging approach using MVFI, rather than to compare MVFI directly with CEUS. Nevertheless, we agree that the absence of a direct comparison with CEUS is a limitation of the study, and we have added this point to the revised Discussion.

Comments 6: You cite contrast-enhanced ultrasonography literature extensively (refs 11-13) and note its limitations. However, you do not state explicitly that the present study did NOT use CEUS. Please add a clear sentence: "In this study, we did not perform contrast-enhanced ultrasonography; instead, we evaluated MVFI as a contrast-free alternative."

Response 6: Thank you for this helpful comment. We agree that the original manuscript did not clearly state that contrast-enhanced ultrasonography was not performed in the present study. In the revised Introduction, we have now explicitly clarified this point by adding the following sentence: “In this study, we did not perform contrast-enhanced ultrasonography; instead, we evaluated MVFI as a contrast-free approach to vascular assessment.” We believe that this revision makes the study design and its intended clinical position clearer.

Comments 7: You state non-malignant lesions were followed "using TUS or other imaging modalities for at least 1 year." Please specify: (a) the follow-up intervals (e.g., every 6 months, annually), (b) what constituted "other imaging modalities" (CT? MRI? EUS?), and (c) who reviewed the follow-up images and whether they were blinded to the initial diagnosis.

Response 7: Thank you for this helpful comment. We have revised the Methods section to provide more detailed information regarding follow-up. Follow-up modalities included TUS, EUS, contrast-enhanced CT, and MRI, and the number and timing of follow-up examinations varied according to the individual clinical course. At each follow-up examination, lesion stability was assessed by comparison with the previous imaging findings. Because of the retrospective nature of the study, the follow-up assessments were not performed in a blinded manner with respect to the initial diagnosis.

Comments 8: You mention using both convex and linear probes. Did you perform any subgroup analysis comparing MVFI diagnostic performance between probe types? If not, please add a limitation statement. Also, were the three MVFI platforms systematically assigned to specific probe types, or was this operator-dependent?

Response 8: Thank you for this helpful comment. We did not perform a probe-specific subgroup analysis of diagnostic performance, and we have added this as a limitation of the study. The three MVFI platforms were not systematically assigned to specific probe types. Rather, the platform used in each case depended mainly on the ultrasound system available at the time of examination, resulting in a generally quasi-random distribution in routine clinical practice. The predominance of DFI reflects its more frequent routine use and earlier availability at our institution.

Comments 9: You report that hyperechoic lesions were more frequent in malignant than non-malignant groups (50% vs. 13%, p=0.022). This finding contradicts the typical expectation that malignancies are hypoechoic. Please discuss this unexpected finding – could it relate to lesion type (e.g., ICPN vs. conventional carcinoma) or technical factors?

Response 9: Thank you for this important comment. We agree that this finding was unexpected and does not fully align with the general impression from previous reports that malignant lesions are more often hypoechoic. Although we could not identify a definitive explanation, the relatively low interobserver agreement for echogenicity in our study suggests that this feature may be less reproducible than other imaging findings. In addition, echogenicity on transabdominal ultrasonography is assessed relative to surrounding reference structures, which may be limited or variable depending on lesion location and imaging conditions. Therefore, we believe that the discrepancy from previous reports may reflect the subjective nature and lower reproducibility of echogenicity assessment in this setting. We have added this point to the Discussion.

Comments 10: You compare your MVFI findings to prior CEUS studies showing heterogeneous enhancement and abnormal vascular architecture. Since you did not perform CEUS in this cohort, please avoid phrasing that implies direct comparison. Suggest revising to: "Our MVFI findings are broadly consistent with observations from contrast-enhanced studies, suggesting that contrast-free vascular assessment may provide similar diagnostic clues."

Response 10: Thank you for this important comment. We agree that the original wording may have implied a direct comparison between MVFI and CEUS, although contrast-enhanced ultrasonography was not performed in this cohort. In the revised Discussion, we modified the paragraph to avoid this implication and to clarify that our findings are only broadly consistent with previous contrast-enhanced studies. We also explicitly stated that direct comparison was not possible in the present cohort. The revised text now reads: “Previous studies using contrast-enhanced ultrasonography have reported heterogeneous enhancement, late-phase hypo-enhancement, and abnormal vascular architecture as useful findings for differentiating malignant from benign GBLs [11,17,19,26–32]. Our MVFI findings are broadly consistent with those observations. However, because contrast-enhanced ultrasonography was not performed in the present cohort, direct comparison was not possible. Therefore, our results should be interpreted as suggesting that contrast-free vascular assessment using MVFI may provide similar diagnostic clues in routine TUS.”

Comments 11: Please add two additional limitations: (1) the lack of direct comparison between MVFI and CEUS in the same patients, which would be needed to determine whether MVFI can truly replace contrast imaging; and (2) the absence of probe-type subgroup analysis, given that linear probes typically offer higher resolution for superficial vessels.

Response 11: Thank you for this valuable comment. We agree that these are important limitations. In the revised limitations section, we explicitly state that direct comparison between MVFI and CEUS was not possible because contrast-enhanced ultrasonography was not performed in this cohort. We also added that, although the distribution of probe types did not differ significantly between groups, a probe-specific subgroup analysis of diagnostic performance was not performed. We believe that the revised text now addresses both concerns more clearly.

General comment: Strength: The interobserver agreement data (κ=0.91 for number of basal vessels) is excellent and clinically actionable. Consider emphasizing this more in the abstract – reproducibility is a major advantage over subjective B-mode features.

Weakness: The sample size (only 10 malignant lesions) and lack of external validation limit generalizability. A prospective multicenter study would be the logical next step.

Response : Thank you for this valuable comment. We agree that one of the major strengths of this study is the high reproducibility of key MVFI findings, particularly the number of basal vessels and vessel thickness. Accordingly, we revised the Abstract to emphasize the interobserver agreement of these findings more clearly. We also agree that the relatively small number of malignant lesions and the lack of external validation limit the generalizability of our results. Therefore, we revised the Conclusion to state more cautiously that multiple basal vessels and vessel dilation appear to be useful and reproducible imaging markers in this cohort and that larger prospective studies are needed to validate these findings.

4. Response to Comments on the Quality of English Language

Point 1: English language and readability

Response 1: The manuscript was carefully revised throughout for clarity and readability. In the revised manuscript, we also transparently disclosed the limited use of a generative AI tool for minor English-language editing and improvement of readability, while stating that no AI was used for data collection, statistical analysis, interpretation, or scientific conclusions.

5. Additional clarifications

All revisions described above are reflected in the clean revised manuscript and the tracked-changes version. We also revised the Discussion and Conclusion globally to maintain a cautious and exploratory interpretation consistent with the retrospective single-center design and limited number of malignant lesions.

Response to Reviewer 2 Comments

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes

The Introduction was further sharpened but already provided the rationale; the main revisions were made in Methods, Results, and Discussion.

Are all the cited references relevant to the research?

Yes

No inappropriate references were identified; the discussion of comparative standards was added as a limitation and future direction.

Is the research design appropriate?

Must be improved

We clarified the exploratory nature of the model development and the mixed reference standard.

Are the methods adequately described?

Can be improved

We added details on body habitus, dichotomization, model derivation, and the interobserver agreement setting.

Are the results clearly presented?

Must be improved

The Results and tables were revised to improve readability and to add platform distribution and confidence intervals.

Are the conclusions supported by the results?

Must be improved

The Conclusion was softened to reflect the exploratory nature of the study.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The authors describe that 42 consecutive patients with final diagnosis (line 62), which was defined as follow-up for at least one year (line 67) if no histopathology was present. It is unclear to this reviewer how reliable this measure is and if it is the appropriate comparator for such an evaluation if the final diagnosis is not confirmed by histopathology. Especially if a predictive model is suggested by the authors.

Response 1: Thank you for this important comment. We agree that the reliability of the reference standard should be described more clearly, particularly for non-malignant lesions without histopathological confirmation. In the revised Methods section, we clarified that such lesions were classified as non-malignant when follow-up imaging performed at least 1 year after the initial examination showed no change in lesion size or morphology, using TUS, EUS, contrast-enhanced CT, or MRI. We also revised the limitations section to explicitly acknowledge that not all non-malignant lesions were histopathologically confirmed and that the use of a mixed reference standard may have introduced verification bias.

Comments : The reader does not know anything about the patient except the small amount of information presented in table 1. This indicates that crucial information is missing given the fact that all investigations with two ultrasound machines were conducted with linear probes. In depths beyond five centimters, microvascular imaging performance is impaired, so this requires that patients are expected to be rather slim. Therefore, additional information on height, weight, and BMI would be helpful in the interpretation.

Response : Thank you for this important comment. We agree that additional information on body habitus would be helpful for interpreting the imaging conditions in this study. We therefore reviewed the clinical data and added height, weight, and BMI to Table 1. We also revised the Results section accordingly. No significant differences were observed between the invasive malignant and non-malignant groups in these variables. We believe that these additions improve the clinical interpretability of the cohort.

Comments : As Table 1 addresses the subjects, simply stating "size" is irritating as this certainly is not the patient size.

Response : Thank you for this helpful comment. We agree that the term “size” was too vague in Table 1 and could be misleading. We have revised it to “Lesion size, mm” for clarity.

Comments : The authors use a lot of dichotomization of continuous data in their analysis, but the justification for the dichotomization is missing. For example, why is the size dichotomized at 15mm or the age at 65 or the number of vessels in absent/single vs. multiple?

Response : Thank you for this valuable comment. We agree that the basis for dichotomization should be stated more clearly. The thresholds for age (65 years) and lesion size (15 mm) were chosen as approximate and clinically interpretable values for exploratory analysis, rather than as formally optimized cut-off values. Similarly, the categorization of basal vessels as absent/single versus multiple was adopted as a pragmatic morphological classification, because multiple basal vessels were considered to suggest a sessile lesion, and this parameter was evaluated only in polypoid lesions. We acknowledge that these dichotomizations were exploratory and not formally validated, and we have clarified this in the revised manuscript.

Comments : The authors compared the concordance between the two gastroenterologists, but it is doubtful if that is possible so straightforward as they did not account for the moderator variable of the used ultrasound machine as it is not stated that the same patient had all the ultrasound investigations in the same machine even if the same set of images was analysed by both gastroenterologists in parallel.

Response : Thank you for this valuable comment. We agree that the use of different ultrasound systems may have affected image acquisition and represents a source of technical heterogeneity. In this study, however, the two gastroenterologists independently reviewed the same stored image sets, so the concordance analysis was intended to evaluate interobserver agreement in image interpretation. We have acknowledged the multi-platform design as a limitation in the revised manuscript and have added the distribution of MVFI platforms, showing no significant difference between the invasive malignant and non-malignant groups.

Comments : Based on what criteria have the analyzed parameters been included in the model?

Response : Thank you for this important comment. The analyzed parameters were not pre-specified for model construction. In the revised manuscript, we have clarified in the Methods section that the diagnostic models were constructed in an exploratory manner after consensus review. Among the findings associated with malignancy, clinically interpretable parameters with relatively high interobserver agreement were prioritized for inclusion in the models. We also agree that this approach was data-driven and that the resulting model may therefore appear arbitrary. Accordingly, we have added a statement in the Discussion acknowledging that, because the diagnostic models were developed and evaluated using the same dataset, their performance should be interpreted as exploratory and hypothesis-generating. We further note that prospective validation in an independent cohort will be necessary.

Comments : Regarding the model itself, it is doubtful whether the comparison is adequate: No one would interpret microvascular imaging without the B-mode and B-mode without any form of doppler imaging. Therefore, this analysis seems rather artificial.

Response : Thank you for this valuable comment. We agree that MVFI and B-mode cannot be completely separated in image interpretation, because B-mode information is inherently included in MVFI images. This is an unavoidable characteristic of the modality. Therefore, the present comparison should not be interpreted as a comparison between two fully independent imaging techniques. Rather, the purpose of this analysis was to explore whether MVFI-based assessment may provide additional diagnostic value beyond a pragmatic B-mode assessment in routine clinical practice. We have clarified this point in the revised manuscript.

Comments : Moreover, it is not reported what has led to the inclusion in the model via the consensus.

Response : Please see our response to Point 6. In the revised Methods section, we have clarified that the diagnostic models were constructed in an exploratory manner after consensus review, and that clinically interpretable findings associated with malignancy and showing relatively high interobserver agreement were prioritized for inclusion.

Comments : The discussion lacks crucial points: Why has the present study not been evaluated against a standard, may it b GB-RADS or CT? Does the misclassification of a potentially malign lesion as benign not support this reviewers point that only histopathology may provide a definitive diagnosis for this kind of study?

Response : Thank you for this valuable comment. We agree that comparison with established standards such as GB-RADS or CT would be informative. However, the present study was designed as an exploratory analysis of MVFI rather than a comparative study against existing diagnostic frameworks. We also acknowledge that not all non-malignant lesions were histopathologically confirmed. Because lesions considered benign are not always surgically resected in clinical practice, restricting the analysis to pathology-confirmed cases alone would substantially limit the feasibility of evaluating a new diagnostic modality and could introduce selection bias. Therefore, follow-up imaging was used for classification in some non-malignant lesions, and we have explicitly acknowledged this as a limitation. We have also noted in the revised Discussion that future studies should compare MVFI with established diagnostic frameworks such as GB-RADS or CT.

Comments : The authors do not discuss the level of certainty of the findings given the small sample size and the relevant distortion due to the dichotomization.

Response : Thank you for this important comment. We agree that the certainty of the present findings is limited by both the relatively small sample size and the use of exploratory dichotomization. In the revised manuscript, we have addressed this point more explicitly in the Discussion. Specifically, we now state that the observed diagnostic performance should be interpreted cautiously because of the small number of malignant lesions and the use of exploratory, clinically pragmatic thresholds rather than formally validated cut-off values. We also emphasize that the diagnostic models were developed and evaluated using the same dataset and therefore should be regarded as exploratory and hypothesis-generating until validated in larger prospective cohorts.

Comments : Given the shortcomings, the conclusions are not justified by the present study.

Response : Thank you for this important comment. We agree that the conclusion in the original manuscript was stronger than warranted given the exploratory nature of the present study. We therefore revised the Conclusion to make it more cautious by limiting the interpretation to the present cohort, emphasizing the reproducible nature of the key MVFI findings, and explicitly stating that these findings should be interpreted cautiously and require validation in larger prospective studies.

4. Response to Comments on the Quality of English Language

Point 1: The English is fine and does not require any improvement.

Response 1: The manuscript was carefully revised throughout for clarity and readability. In the revised manuscript, we also transparently disclosed the limited use of a generative AI tool for minor English-language editing and improvement of readability, while stating that no AI was used for data collection, statistical analysis, interpretation, or scientific conclusions.

5. Additional clarifications

All revisions described above are reflected in the clean revised manuscript and the tracked-changes version. We also revised the Discussion and Conclusion globally to maintain a cautious and exploratory interpretation consistent with the retrospective single-center design and limited number of malignant lesions.

Response to Reviewer 3 Comments

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files.

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Yes

The manuscript already had a clear rationale, and the revised version retains this while improving statistical precision in reporting elsewhere.

Are all the cited references relevant to the research?

Yes

The references remain relevant, and the wording around comparative interpretation was refined.

Is the research design appropriate?

Can be improved

We clarified the mixed reference standard and its limitations more explicitly.

Are the methods adequately described?

Yes

The follow-up definition and the statistical wording regarding McNemar testing were revised.

Are the results clearly presented?

Can be improved

The Results were revised to improve statistical wording and the tables now report 95% confidence intervals.

Are the conclusions supported by the results?

Yes

The revised wording is now more cautious and aligned with the data.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: A substantial proportion of the non-malignant lesions (20 cases, 62%) were not pathologically confirmed but were considered based on clinical follow-up. The description of “clinical follow-up using TUS or other imaging modalities for at least 1 year” is not sufficient, and a follow-up period of 1 year is relatively short to define non-malignant lesions. More detailed information is needed on how these 20 lesions were determined to be non-malignant.

Response :Thank you for this important comment. We agree that the original description of follow-up-based diagnosis for non-malignant lesions was insufficient. In the revised manuscript, we clarified in the Methods section that lesions without histopathological confirmation were classified as non-malignant when follow-up imaging performed at least 1 year after the initial examination showed no change in lesion size or morphology. We also specified that follow-up modalities included TUS, EUS, contrast-enhanced CT, and MRI, and that lesion stability was assessed at each follow-up examination by comparison with previous imaging findings. In addition, we revised the limitations section to explicitly acknowledge that not all non-malignant lesions were histopathologically confirmed and that the use of a mixed reference standard may have introduced verification bias.

Comments 2: In the Abstract, Methods (statistical analysis), Results, and Discussion, it is stated that the diagnostic accuracy of MVFI was significantly higher than that of B-mode US based on the exact McNemar test. This is not an appropriate expression. The exact McNemar test evaluates the significance of differences in paired comparisons between two methods, and it does not directly indicate that the diagnostic accuracy itself is significantly higher.

Response 2: Thank you for this important comment. We agree that the original wording was not statistically precise. The exact McNemar test evaluates the significance of discordance in paired comparisons between two methods and does not directly demonstrate that diagnostic accuracy itself is significantly higher. Accordingly, we revised the wording in the Abstract, Methods, Results, and Discussion to avoid this overstatement. We now state that paired comparison using the exact McNemar test showed a significant difference between the MVFI-based and B-mode models, while separately reporting the observed accuracy values for each model.

4. Response to Comments on the Quality of English Language

Point 1: The English is fine and does not require any improvement.

Response 1: The manuscript was carefully revised throughout for clarity and readability. In the revised manuscript, we also transparently disclosed the limited use of a generative AI tool for minor English-language editing and improvement of readability, while stating that no AI was used for data collection, statistical analysis, interpretation, or scientific conclusions.

5. Additional clarifications

All revisions described above are reflected in the clean revised manuscript and the tracked-changes version. We also revised the Discussion and Conclusion globally to maintain a cautious and exploratory interpretation consistent with the retrospective single-center design and limited number of malignant lesions.

Response to Reviewer 4 Comments

1. Summary

Thank you very much for taking the time to review this manuscript. Please find the detailed responses below and the corresponding revisions/corrections highlighted/in track changes in the re-submitted files. [This is only a recommended summary. Please feel free to adjust it. We do suggest maintaining a neutral tone and thanking the reviewers for their contribution although the comments may be negative or off-target. If you disagree with the reviewer's comments please include any concerns you may have in the letter to the Academic Editor.]

2. Questions for General Evaluation

Reviewer’s Evaluation

Response and Revisions

Does the introduction provide sufficient background and include all relevant references?

Can be improved

The Introduction was revised to state the study gap and novelty more explicitly.

Are all the cited references relevant to the research?

Can be improved

The references are appropriate; the revised Discussion now more carefully positions the findings relative to prior literature.

Is the research design appropriate?

Yes

We explicitly describe the study as retrospective, single-center, and exploratory, and acknowledge the risk of optimistic model performance estimates.

Are the methods adequately described?

Can be improved

We clarified the reference standard, platform distribution, and model derivation in the Methods and Results.

Are the results clearly presented?

Yes

Diagnostic performance tables were reformatted and now include 95% confidence intervals.

Are the conclusions supported by the results?

Can be improved

The Discussion and Conclusion now frame the findings as preliminary and requiring external prospective validation.

3. Point-by-point response to Comments and Suggestions for Authors

Comments 1: The Introduction is generally clear and provides an acceptable rationale for the study, but it would benefit from sharper positioning within the existing literature. The authors cite prior work on contrast-enhanced ultrasound and MVFI-related techniques, yet the manuscript should state more explicitly what gap this study fills beyond prior gallbladder ultrasound studies, especially given that some cited references appear closely related to the present topic. The novelty seems to lie in the cross-platform MVFI assessment and in the specific vascular criteria evaluated, and that should be stated more directly.

Response 1: Thank you for this valuable comment. We have revised the Introduction to state the novelty of the present study more explicitly. Specifically, we clarified that the gap addressed by this study is the lack of an established practical classification of vascular findings for gallbladder lesion diagnosis across different MVFI platforms. We also emphasized that the present study focused on both the diagnostic utility and reproducibility of specific MVFI vascular features.

Comments : The Methods section is understandable, but several points need clarification. First, the reference standard is mixed: all malignant lesions were confirmed surgically, whereas many non-malignant lesions were classified on the basis of follow-up rather than histopathology. This is acceptable in a retrospective diagnostic study, but it should be presented more clearly as a limitation because it introduces the possibility of verification bias. The authors should report more explicitly what imaging modalities were used during follow-up, how stability was defined, and whether any lesions changed category during surveillance.

Response : Thank you for this important comment. We agree that the reference standard should be described more clearly. In the revised Methods section, we clarified that non-malignant lesions without histopathological confirmation were classified on the basis of follow-up imaging using TUS, EUS, contrast-enhanced CT, and MRI. We also specified that such lesions were regarded as non-malignant only when follow-up imaging performed at least 1 year after the initial examination demonstrated no change in lesion size or morphology, and that lesion stability was assessed at each follow-up examination by comparison with previous imaging findings. In addition, lesions that changed category during follow-up were omitted from the analysis. We also revised the limitations section to explicitly acknowledge that the use of a mixed reference standard may have introduced verification bias.

Comments : Second, the use of three different MVFI platforms is interesting and may improve generalizability, but it also introduces heterogeneity. The manuscript should clarify how many lesions were assessed on each platform and whether there were any meaningful differences in lesion type, lesion depth, or pathology across systems. At present, the paper states that B-flow, DFI, and SMI were used, but the distribution of cases across platforms is not clearly reported. Without this information, it is difficult to judge whether the reported diagnostic performance reflects a broadly shared MVFI signal or is driven disproportionately by one system.

Response : Thank you for this valuable comment. We have added the case distribution for each MVFI platform to the Results section and Table 1, and we confirmed that the distribution of invasive malignant and non-malignant lesions did not differ significantly among the platforms. Lesion depth was not measured in this study, and additional subgroup analyses according to lesion type were not performed. Therefore, platform-related heterogeneity cannot be completely excluded. Nevertheless, the relatively high interobserver agreement for the key MVFI findings suggests that these findings were robust across the evaluated image sets.

Comments : Third, the process used to construct the diagnostic models should be explained more carefully. The manuscript states that clinically interpretable findings associated with malignancy after consensus review were used to build the models. However, because the same dataset appears to have been used both to identify the relevant features and to test the resulting model, there is a risk of overfitting and optimistic performance estimates. This does not invalidate the analysis, but it should be acknowledged explicitly. The authors should make clear that the model is exploratory and hypothesis-generating rather than validated.

Response : Thank you for this valuable comment. We agree that the diagnostic models in this study were developed in an exploratory manner. Specifically, individual imaging findings associated with invasive malignancy were first identified, and diagnostic criteria were then constructed using clinically interpretable findings with relatively high interobserver agreement. Because these models were also evaluated in the same cohort, their performance may be optimistic. We have therefore clarified in the revised manuscript that the models are exploratory and hypothesis-generating rather than validated.

Comments : The Results section is generally well organized. The sequence from patient characteristics to interobserver agreement, consensus imaging findings, and diagnostic performance is logical and easy to follow. The strongest aspect of the results is that the authors do not rely only on p values, but also provide sensitivity, specificity, and accuracy for the main diagnostic criteria. The very high agreement for number of basal vessels and the good agreement for vessel thickness are particularly relevant because these are the two features incorporated into the final MVFI model.
However, the results would be stronger if the authors added confidence intervals for diagnostic performance measures, especially given the small sample size and the very small malignant subgroup. Reporting only point estimates may make the findings appear more precise than they actually are. Confidence intervals would help readers interpret the uncertainty around the observed sensitivity, specificity, and accuracy. This is an essential revision because it directly affects how the results are understood.

Response : Thank you for this important comment. We agree that reporting only point estimates may overstate the precision of the findings, particularly given the small sample size and the limited number of malignant lesions. In response to your suggestion, we have added 95% confidence intervals for sensitivity, specificity, and accuracy to Table 4. We believe that these additions improve the interpretation of the uncertainty surrounding the observed diagnostic performance.

Comments : There are also some issues in the tables that should be corrected. Table 2 is difficult to read in its current form because the percentages and raw counts are presented in a format that is not immediately intuitive. Table 4 also needs minor formatting cleanup, including the broken line in the B-mode model row. In addition, the manuscript should ensure that all denominators are immediately clear wherever only polypoid lesions are included in the analysis. These are not major scientific problems, but they do affect readability and should be fixed.

Response : Thank you for this helpful comment. We agree that the readability of Tables 2 and 4 needed improvements. In the revised manuscript, we reformatted Table 2 so that percentages and raw counts are presented more intuitively, and we revised Table 4 to improve the layout and readability of the diagnostic performance data. We also clarified denominators wherever analyses were restricted to polypoid lesions. In addition, for consistency and readability across the Results section, we revised Table 3 to use the same % (n/N) format. We believe that these revisions have improved the clarity of the tabulated data.

Comments : The Discussion is thoughtful overall and appropriately attempts to interpret the vascular findings in clinical terms. I found the explanation of why multiple basal vessels and vessel dilation may be practically useful to be reasonable and clinically interpretable. The limitations section is present and includes several important points, including the retrospective design, small sample size, mixed lesion types, and use of multiple platforms. However, it should be strengthened further by explicitly discussing the lack of external validation, the potential inflation of model performance due to derivation and testing in the same cohort, and the limitations of comparing a data-driven MVFI model with a simplified pragmatic B-mode model.

Response : Thank you for this important comment. We agree that the Discussion should clearly acknowledge the lack of external validation, the potential inflation of model performance due to derivation and testing in the same cohort, and the limitations of comparing an MVFI-based model with a simplified pragmatic B-mode model. In the revised manuscript, we have explicitly addressed these points. Specifically, we now state that the diagnostic models were developed and evaluated using the same dataset and should therefore be interpreted as exploratory and hypothesis-generating. We also clarified that the comparison between the MVFI-based and B-mode models should be regarded as an exploratory assessment of incremental diagnostic value rather than a comparison between two fully independent modalities. In addition, we revised the Conclusion to emphasize the need for cautious interpretation and prospective validation in larger cohorts.

Comments : The conclusion is broadly supported by the results, but it should be phrased more conservatively. The current wording is close to acceptable, but the manuscript should consistently avoid implying that MVFI is already established as a reliable diagnostic discriminator in general practice. Based on the presented data, it is fair to conclude that MVFI appears promising and may improve diagnostic discrimination in this cohort, particularly when multiple basal vessels or vessel dilation are present. A stronger claim would not yet be justified.

Response : Thank you for this valuable comment. We agree that the conclusion should be more cautious. Accordingly, we revised the Conclusion to avoid overstatement and to clarify that MVFI appears promising in this cohort, particularly when multiple basal vessels or vessel dilation are present, but that these findings require prospective validation in larger studies.

Comments : Clarify the reference standard in more detail, particularly the criteria and imaging modalities used for follow-up diagnosis in non-malignant lesions.

Response : Thank you for this helpful comment. We have revised the Methods section to clarify the reference standard in more detail. Specifically, we now state that follow-up diagnosis for non-malignant lesions was based on TUS, EUS, contrast-enhanced CT, and MRI, and that lesions without histopathological confirmation were regarded as non-malignant only when no change in size or morphology was observed on follow-up imaging performed at least 1 year after the initial examination.

Comments : Report how many lesions were examined with each MVFI platform and discuss possible platform-related heterogeneity.

Response : Thank you for this important comment. In the revised manuscript, we added the distribution of lesions across the three MVFI platforms to Table 1 and the Results section, and confirmed that there was no significant difference in the distribution of invasive malignant and non-malignant lesions among the platforms. Lesion depth was not measured in this study, and we did not perform further subgroup analyses according to lesion type across platforms. However, we have now clarified in the Discussion that the three platforms had different practical imaging characteristics in routine clinical use, which may have introduced platform-related heterogeneity. We therefore acknowledge this point as a limitation of the study.

Comments : Explain more explicitly how the diagnostic models were derived and acknowledge that model development and testing were performed in the same dataset.

Response : Thank you for this valuable comment. We have revised the Discussion, limitations section, and Conclusion to address these points more explicitly. Specifically, we now state that the diagnostic models were derived and tested in the same dataset and should be interpreted as exploratory and hypothesis-generating, and that the comparison with the simplified B-mode model represents an exploratory assessment of incremental diagnostic value rather than a comparison between fully independent modalities. We also emphasize the lack of external validation and the need for prospective confirmation.

Comments : Add confidence intervals for sensitivity, specificity, and accuracy in Table 4 and, if possible, in the text.

Response : Thank you for this valuable comment. We have revised Table 4 and the Results section to include 95% confidence intervals for sensitivity, specificity, and accuracy.

Comments : Revise the discussion and conclusion to emphasize that the findings are preliminary and require external prospective validation.

Response : Thank you for this valuable comment. We agree that the findings should be interpreted as preliminary. Accordingly, we revised the Discussion and Conclusion to emphasize the exploratory nature of the results and the need for external prospective validation in larger cohorts.

Comments : Improve the presentation of Tables 2 and 4 so that denominators, percentages, and subgroup restrictions are immediately clear.

Response : Thank you for this valuable comment. We revised Tables 2 and 4 to improve readability by clarifying denominators, percentages, and subgroup restrictions. We also revised Table 3 to use the same % (n/N) format for consistency.

Comments : Strengthen the Introduction by stating more explicitly what is novel in this study compared with prior ultrasound and MVFI literature on gallbladder lesions.

Response : Thank you for this important comment. We agree that the novelty of the present study should be stated more explicitly in the Introduction. In the revised manuscript, we clarified that the main unresolved issue in transabdominal ultrasonography for gallbladder lesions is the differential diagnosis based on vascular morphology. We also emphasized that, although contrast-enhanced ultrasonography can evaluate vascularity, assessment of fine vessel architecture is often difficult because the lesion is visualized as whole-lesion enhancement. In addition, we clarified that previous MVFI studies in gallbladder lesions have been limited, including mainly single-platform evaluations, whereas the novelty of the present study lies in the assessment of specific vascular findings across multiple MVFI platforms without the use of contrast agents. We believe that these revisions make the gap in the literature and the contribution of the present study clearer.

4. Response to Comments on the Quality of English Language

Point 1: The English could be improved to more clearly express the research.

Response 1: The manuscript was carefully revised throughout for clarity and readability. In the revised manuscript, we also transparently disclosed the limited use of a generative AI tool for minor English-language editing and improvement of readability, while stating that no AI was used for data collection, statistical analysis, interpretation, or scientific conclusions.

5. Additional clarifications

All revisions described above are reflected in the clean revised manuscript and the tracked-changes version. We also revised the Discussion and Conclusion globally to maintain a cautious and exploratory interpretation consistent with the retrospective single-center design and limited number of malignant lesions.