Investigating the Average Glandular Dose and Exposure Parameters in Mammography Based on Compressed Breast Thickness and Imaging Projection: A Single-Centre Study in Jeddah, Saudi Arabia

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The study addresses a clinically relevant topic and provides useful local data on mammography dose metrics, supported by a reasonably large dataset and standard statistical methods. It contributes to regional benchmarking and reinforces established radiological principles in a real-world clinical setting.

Areas of Improvement:

• Incorporate advanced analysis (e.g., multivariate regression, predictive modeling) or include additional variables such as breast density and patient BMI.
• Expand to a multicenter study with multiple mammography systems to improve external validity.
• Perform multivariate analysis and report effect sizes to account for confounding factors.
• Acknowledge mathematical coupling and validate findings with independent or adjusted models.
• Conduct thorough language editing and professional proofreading.
• Strengthen discussion with practical implications, optimization strategies, and comparison with international benchmarks.

Author Response

We sincerely thank the reviewer for their valuable comments. We have responded to each point individually and revised the manuscript accordingly. All changes have been clearly highlighted in the revised manuscript.

Comment 1: Incorporate advanced analysis (e.g., multivariate regression, predictive modeling) or include additional variables such as breast density and patient BMI.

We thank the reviewer for this valuable and insightful suggestion. We agree that advanced statistical approaches, such as multivariate regression or predictive modeling, could further enhance the analytical depth of the study. However, the primary aim of this work was to investigate the relationship between average glandular dose (AGD) and exposure parameter under controlled CBT and imaging projection using a large retrospective clinical dataset and established dosimetric parameters. Accordingly, the analysis was designed using stratified and inferential statistical methods across CBT categories, which are widely used in similar dosimetric studies.

Regarding additional variables such as breast density and body mass index (BMI), these were not consistently available in the retrospective dataset and therefore could not be included in the analysis. This has now been acknowledged under "Limitations and future research directions" in the revised manuscript.

Comment 2: Expand to a multicenter study with multiple mammography systems to improve external validity.

Response 2: We thank the reviewer for this important suggestion. We agree that expanding the study to a multicenter design involving multiple mammography systems would further improve the external validity and generalizability of the findings. However, this aspect has already been acknowledged in the “Limitations and Future Research Directions” section of the revised manuscript.

Comment 3: Perform multivariate analysis and report effect sizes to account for confounding factors.

Response 3: We thank the reviewer for this valuable and insightful suggestion. We agree that multivariate analysis and reporting of effect sizes would provide additional depth by accounting for potential confounding factors and quantifying the strength of associations. However, the primary objective of this study was to investigate the correlation between average glandular dose (AGD) and exposure parameters based on selected CBT range and projection using stratified descriptive and inferential statistical methods, which are commonly employed in dosimetric studies of this nature. Given the retrospective design and the available dataset, the analysis focused on CBT-based stratification as the principal determinant of AGD. We acknowledge that additional multivariate analysis incorporating variables such as breast density and BMI could further strengthen the analysis. This has been highlighted as a direction for limitation and future research in the revised manuscript.

Comment 4: Acknowledge mathematical coupling and validate findings with independent or adjusted models.

Response 4: We thank the reviewer for this comment. In this study, AGD values were obtained using a standardized automated calculation (Dance algorithm) based on exposure parameters and breast thickness. To minimise bias, analyses were performed using stratified CBT categories rather than direct derived-to-derived ratio comparisons.

Comment 5: Conduct thorough language editing and professional proofreading.

Response 5: We thank the reviewer for this helpful comment. We asked the journal to send the manuscript for professional language editing through the journal to ensure it meets the high linguistic and stylistic standards required for publication. This process will improve the clarity, readability, and overall quality of the manuscript, thereby enhancing its scientific impact.

Comment 6: Strengthen discussion with practical implications, optimization strategies, and comparison with international benchmarks.

Response 6: We thank the reviewer for this valuable suggestion. In response, we have substantially strengthened the Discussion section by expanding on the practical clinical implications and dose optimisation strategies of our findings.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Mammography remains one of the best methods for detecting early-stage breast cancer. On the other hand, the mammary gland is highly sensitive to ionising radiation, which makes exposure to ionising radiation a great concern. On the other hand, the mammary gland is highly sensitive to ionizing radiation, which makes exposure to ionizing radiation a great concern. In view of this, any study which could bring a supplement of information concerning the dependency of the average glandular dose on exposure geometry or on the breast tissue geometry should be appreciated.

It is the case of the present manuscript, which, without being particularly original, makes a worthwhile contribution to this topic, especially since it is based on a rigorous statistical analysis of more than 600 individual cases.

At its present form, the manuscript needs some moderate revisions, but altogether, the manuscript is meritorious.

Concerning the Abstract, I would recommend restricting it to the main results by removing divisions such as Introduction, Methods, Results, and Conclusions.

Also, I would recommend posting the ORCID ID for each author, i.e.

Dalal Almoudi ORCID 0009-0001-1717-6454

Amnana Babdgi ORCID 0009-0007-7493-3478

Lama Bazuhair ORCID 0009-0004-3025-3030

Shaza Alsharif ORCID 0000-0002-9860-0789

Amani Y. Alhalwani ORCID 0000-0003-2734-2767

Doaa Almaudi ORCID 0009-0002-1848-3429

Some other less important remarks follow:

Row 144: Please provide a reference to “Dance algorithm”

Row 155: Aaccordingly misspelling !!

Row 165: please more details on how the DRL was calculated

Tables 1 and 2: A single decimal for SD should be enough

Table 2: Please could you explain the higher values of p at small CBT

Given the great number of acronyms, I would suggest that authors include a List of Abbreviations before the References. This would make the manuscript most readable.

 

13 April 2026  Supplementary

The problem of evidencing the most appropriate metric in the case of breast dosimetry in any screening of breast cancer is of paramount importance due to its dual aspect: an earlier diagnosis of breast cancer and a reduction to a minimum of the stochastic effects inherent to any exposure to ionizing radiation.

Given the stochastic character of the ionising radiation interaction with matter and, as a consequence, of its negative effects on humans, any statistically documented study is welcomed.

This is the case of the presented study, which discusses the results of a cross-sectional analysis regarding 607 patients who were examined by digital mammography at King Abdulaziz Medical City, Jeddah, Saudi Arabia, during a period of 24 months.

According to the authors, this study demonstrates a direct correlation between the average glandular dose and the compressed breast thickness, in fact, a consequence of the exponential attenuation of the X-rays in organic tissue.

Although this finding did not bring new knowledge of the X-ray interaction with the living tissue, the number of investigated cases under different exposure geometry, the age span of investigated patients as well as the top level of statistical data analysis makes the manuscript meritorious.

In this regard, the Introduction presents some local characteristics of breast cancer in the considered area, i.e. Saudi Arabia, as well as the role of ALARA principle in recommending a mammography examination.

In this section, the Saudi Food & Drug Authority diagnostic reference levels are presented and discussed, including the European Union, Australia and United States regulatory dose limits, all of them in connection with the aim of the presented project, i.e the optimisation of the mammography examination for different exposure geometries.

In the next section, Materials and Methods, the authors deal with more aspects of the study, regarding the description of the mammography system, patient selection, data collection and statistical analysis of estimated diagnostic reference levels and, which is of remarkable importance for this kind of study, the Ethical approval, as the entire study is based on human subjects.

The final Results of the study are the object of the next sections, where numerical data make te object of the Table 1 and 2. In this regard, it should be mentioned that the fact that authors have used Spearman’s correlation coefficient, the most appropriate descriptor for non-normally distributed data. Unfortunately, the resulting correlation coefficients, despite the higher probability (p< 0.0001), pointed toward medium positive correlations between average glandular dose and the compressed breast thickness.

In my opinion, these results express the complexity of the interaction process af the X-Ray and glandular breast tissue. It represents an outcome of the experimental measurements which should be accepted and interpreted as such, without any remarks on the quality of the experimental determinations, which appear of good quality.

In the next section, Discussion, the authors analyze the results of their experimental data, the correlation between the average glandular dose and the X-Ray source parameters such as kVp and mAs, as well as other factors such as exposure geometry, ending up with a brief exposure of the limitations of this category of studies, as well as pointing towards possible future investigations.

The final Conclusion of this study makes the object of the final section, which in short summaries the most important achievements of the present study. 

 

Author Response

We sincerely thank the reviewers for their valuable comments. We have responded to each point individually and revised the manuscript accordingly. All changes have been clearly highlighted in the revised manuscript.

• Comment 1: Concerning the abstract, I would recommend restricting it to the main results by removing divisions such as Introduction, Methods, Results, and Conclusions.

Response 1: We thank the reviewer for this suggestion. We have chosen to retain the structured format (Introduction, Methods, Results, and Conclusion) because it enhances clarity, improves readability, and ensures that key methodological and statistical details are presented clearly and systematically. Structured abstracts are commonly used in radiology and medical imaging journals, especially for technical studies like this study, and we believe this format is appropriate for our study. We hope the reviewer finds this justification acceptable.

 

• Comment 2: Also, I would recommend posting the ORCID ID for each author, i.e.

Dalal Alamoudi ORCID 0009-0001-1717-6454

Amna Babdgi ORCID 0009-0007-7493-3478

Lama Bazuhair ORCID 0009-0004-3025-3030

Shaza Alsharif ORCID 0000-0002-9860-0789

Amani Y. Alhalwani ORCID 0000-0003-2734-2767

Doaa Alamoudi ORCID 0009-0002-1848-3429

Response 2: We thank the reviewer for this useful suggestion. In response, we have now added the ORCID IDs for all authors in the revised manuscript, in accordance with the journal’s submission guidelines. We agree that inclusion of ORCID identifiers improves author identification, transparency, and proper attribution of scholarly work.

 

Some other, less important remarks follow:

• Comment 3: Row 144: Please provide a reference to “Dance algorithm.”

Response 3: We thank the reviewer for this valuable comment. We have now added the appropriate reference for the Dance algorithm in the revised manuscript. The corresponding citation has been included in the Methods section (under the Mammography system description section).

• Comment 4: Row 155: Aaccordingly misspelling !!

Response 4: We thank the reviewer for carefully identifying this typographical error. The word “accordingly” has been corrected to “Accordingly” in the revised manuscript (under the patient selection section). We have also performed a thorough proofreading of the manuscript to ensure that no similar typographical errors remain.

• Comment 5: “Row 165: please more details on how the DRL was calculated.”

Response 5: We thank the reviewer for this important request for clarification. In response, we have revised the Methods section (under DRL estimation) to provide additional detail on how the Diagnostic Reference Level (DRL) was determined. For clarification, in this study, “In accordance with international recommendations, the DRL was defined as the 75th percentile of the AGD distribution for each projection.” The corresponding description and clarification have now been added to the revised manuscript.

• Comment 6: Tables 1 and 2: A single decimal for SD should be enough

Response 6: We thank the reviewer for this helpful suggestion. We agree that limiting decimal places can improve readability; however, we have chosen to retain two decimal places for the standard deviation (SD) values to preserve numerical precision, particularly given the relatively small variability in some of the reported measurements. Reducing the SD to a single decimal place may lead to loss of relevant detail and potential rounding effects. For consistency and accuracy across the dataset, SD values are therefore presented to two decimal places in Tables 1 and 2. We hope the reviewer finds this justification acceptable.

• Comment 7: Table 2: Please could you explain the higher values of p at small CBT

Response 7: We thank the reviewer for this insightful comment. We have added this explanation: “The higher p-values observed at lower CBT ranges can be explained by the smaller differences in AGD between projections and reduced variability within these groups. In contrast, at higher CBT, the increasing differences in AGD lead to lower p-values and stronger statistical significance.” to the Discussion section of the revised manuscript to improve clarity regarding the observed statistical trends.

• Comment 8: Given the great number of acronyms, I would suggest that authors include a List of Abbreviations before the references. This would make the manuscript most readable.

Response 8: We thank the reviewer and state that you added a List of Abbreviations before the References to improve readability and clarity, especially given the technical nature of the study.

 Abbreviations

The following abbreviations are used in this manuscript:

AGD: Average Glandular Dose

AEC: Automatic Exposure Control

ALARA: As Low As Reasonably Achievable

CBT: Compressed Breast Thickness

CC: Craniocaudal

CF: Compression Force

DRL: Diagnostic Reference Level

DICOM: Digital Imaging and Communication in Medicine

ESD: Entrance Surface Dose

ESAK: Entrance Surface Air Kerma 

HVL: half-value layer

kVp: Peak Kilovoltage

mAs: Milliampere-second

MGD: Mean glandular dose

MLO: Mediolateral Oblique

NDRL: National Diagnostic Reference Level

PACS: Picture Archiving and Communication System

QA: Quality Assurance 

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This paper investigates the relationship between average glandular dose (AGD) and exposure parameters in digital mammography across different compressed breast thicknesses and imaging projections in a single-center cohort. It reports that AGD increases with breast thickness and is higher in mediolateral oblique views, with strong correlations particularly between AGD and mAs. Multiple issues across different sections of the manuscript raise concerns:

- The title suggests an assessment of both average glandular dose (AGD) and exposure parameters across compressed breast thickness (CBT) and projections; however, the scope of the study is narrower than implied. The manuscript does not truly “assess” exposure parameters in a comprehensive or multivariable manner, but rather performs limited correlation analyses, primarily univariate, without accounting for interdependencies between variables such as mAs, kVp, and CBT. This creates a mismatch between the stated aim and the actual analytical depth.

- In the abstract, the statistical approach is described in a way that lacks clarity and precision. The description of normality testing and subsequent use of Mann–Whitney U tests does not specify whether multiple testing correction was applied despite repeated comparisons across multiple CBT categories and projections. Additionally, the statement that “significant differences between right and left breasts were observed across CBT groups” is misleading, as the statistical tests described compare CC vs MLO projections rather than right vs left symmetry; the analytical framework for side-to-side comparison is not properly defined.

- The introduction overstates the radiation risk associated with mammography in a manner that is not sufficiently balanced. Statements such as mammography being “linked to a high risk of breast cancer” are not aligned with established epidemiologic evidence and may misrepresent the risk-benefit balance of screening. The discussion of national and international diagnostic reference levels (DRLs) is extensive but not well integrated into a clear research gap. The authors claim limited local data, yet do not clearly define how their study advances beyond existing DRL or AGD literature.

- In the materials and methods section, the single-center design using a single mammography unit severely limits generalizability, yet this limitation is not adequately accounted for in the study design. The reliance on a single vendor system with automatic exposure control (AEC) introduces systematic bias, particularly in exposure parameter selection, making correlation analyses between AGD and technical factors partially tautological. The description of the mammography system is detailed but does not include calibration data, quality control metrics, or temporal stability of the system over the study period.

- The patient selection process raises concerns regarding potential selection bias. A large proportion of patients (over 60%) were excluded, yet the rationale for excluding “limited sample sizes in CBT categories” is unclear and methodologically problematic, as this introduces post hoc selection based on outcome distribution. Furthermore, excluding cases with CBT >100 mm and certain compression force values may systematically remove clinically relevant extremes, biasing the dose distribution.

- The definition and calculation of DRLs are not methodologically consistent. The manuscript states that AGD per view was averaged to obtain a single value per examination, yet later reports projection-specific DRLs. This inconsistency suggests a lack of clarity in how DRLs were derived and whether they follow internationally accepted methodologies (e.g., per-view vs per-exam DRLs). Additionally, the sample size used in Table 1 (609 exams) does not align with the previously stated 607 patients and 2,428 projections, indicating inconsistency in dataset definition.

- The statistical analysis is fundamentally limited. The use of Mann–Whitney U tests for comparing projections is questionable, as CC and MLO views from the same patient are not independent samples; a paired statistical test would have been more appropriate. Similarly, the absence of multivariate regression analysis is a major limitation, given the known collinearity between CBT, mAs, and kVp. The reported correlations, particularly extremely high Spearman coefficients (ρ ≈ 0.995 between mAs and AGD), strongly suggest mathematical coupling rather than true physiological relationships, as AGD is directly derived from exposure parameters.

- Table 1 contains implausible variability metrics, particularly the standard deviation of kVp (±9.54), which is unusually large for mammography and suggests either data extraction errors or inappropriate aggregation of heterogeneous protocols. Additionally, the table mixes system-level and patient-level metrics without clear distinction, reducing interpretability.

- In Table 2, the statistical comparisons are not clearly defined. The reported p-values appear to compare CC vs MLO within each side, yet the narrative text repeatedly interprets them as differences between right and left breasts. This represents a fundamental misinterpretation of statistical testing. Moreover, multiple comparisons across CBT categories are performed without adjustment, increasing the risk of type I error.

- The reporting of results demonstrates internal inconsistencies. For example, statements about “very highly significant differences between both sides” contradict the numerical values presented, where differences between right and left breasts are minimal. The manuscript conflates statistical significance with clinical relevance without quantifying effect sizes.

- The correlation analysis presented in Figure 1 lacks sufficient methodological detail. It is unclear whether the correlations were computed across pooled data or stratified by CBT categories, and the reported values for specific CBT ranges contradict earlier statements about overall correlations. The interpretation of “strongest” and “weakest” correlations is not supported by a consistent analytical framework.

- Table 4 presents extremely high correlation coefficients between mAs and AGD, which are expected due to the computational relationship between these variables. However, the manuscript interprets this as a clinical finding rather than an inherent property of the AGD calculation algorithm. This reflects a lack of understanding of the underlying physics and introduces circular reasoning into the conclusions.

- The discussion largely reiterates known principles of mammographic physics rather than providing novel insights. Many statements are descriptive and confirmatory, without critically engaging with the limitations of the dataset or the analytical approach. The attribution of higher AGD to AEC behavior is speculative and not supported by direct analysis of AEC modes or system settings.

- The comparison with national DRLs is superficial. The manuscript notes that AGD values are slightly above Saudi reference levels but does not perform a formal benchmarking analysis or percentile-based comparison consistent with DRL methodology. Additionally, the interpretation that values are “acceptable” lacks a defined threshold or regulatory context.

- The discussion of compression force (CF) is inconsistent and at times contradictory. The manuscript reports weak correlations but then extrapolates broader conclusions about compression practices without sufficient evidence. The analysis does not account for the interaction between CF and CBT, which is critical in understanding dose optimization.

- The limitations section is insufficiently critical. While the single-center design is acknowledged, other major limitations, such as selection bias, lack of multivariate analysis, non-independence of observations, and potential errors in statistical interpretation, are not addressed.

- The conclusion overstates the findings, particularly the claim that mAs is the “dominant factor” influencing AGD, which is inherently true by definition and not a novel contribution. The statement that CF has negligible impact is not adequately supported by the limited and potentially confounded analysis presented.

- The manuscript also contains multiple inconsistencies in data reporting, including discrepancies in sample size, unclear definitions of units (per view vs per exam), and inconsistent terminology (e.g., mixing AGD, ESD, and exposure parameters without clear hierarchy).

Author Response

We sincerely thank the reviewer for their valuable comments. We have responded to each point individually and revised the manuscript accordingly. All changes have been clearly highlighted in the revised manuscript.

• Comment 1: The title suggests an assessment of both average glandular dose (AGD) and exposure parameters across compressed breast thickness (CBT) and projections; however, the scope of the study is narrower than implied. The manuscript does not truly “assess” exposure parameters in a comprehensive or multivariable manner, but rather performs limited correlation analyses, primarily univariate, without accounting for interdependencies between variables such as mAs, kVp, and CBT. This creates a mismatch between the stated aim and the actual analytical depth.

Response 1: We thank the reviewer for this insightful and important observation. We agree that the original title and aim may have suggested a broader multivariable analysis than was actually conducted.

In this work, the primary objective was to investigate the associations between AGD and individual exposure parameters (mAs, kVp, and compression force), focusing on the controlled CBT and imaging projection, as recorded in routine clinical practice. Accordingly, the statistical approach was intentionally based on stratified analyses and non-parametric correlation testing (Spearman’s rank correlation) to describe monotonic relationships under real-world clinical conditions.

We acknowledge that the relationships between exposure parameters (particularly mAs, kVp, and CBT under automatic exposure control conditions) are interdependent. However, the study was designed as an observational dosimetric analysis, and not a modelling study aiming to isolate independent predictors through multivariable regression. This has now been clarified in the revised Title, Aim, Methods, and Discussion sections to better align the stated objectives with the analytical approach.

To address the reviewer’s concern regarding consistency between the study scope and wording, we have also revised the manuscript title to avoid overstatement of analytical depth. We believe these revisions improve conceptual clarity and better align the title, aims, and statistical methodology.

“Investigating the Average Glandular Dose and Exposure Parameters in Mammography based on Compressed Breast Thickness and Imaging Projection: A Single-Centre Study in Jeddah, Saudi Arabia.”

• Comment 2: In the abstract, the statistical approach is described in a way that lacks clarity and precision. The description of normality testing and subsequent use of Mann–Whitney U tests does not specify whether multiple testing correction was applied, despite repeated comparisons across multiple CBT categories and projections.

Response 2: We thank the reviewer for this important comment. We have revised the abstract to improve clarity.

 

• Comment 3: Additionally, the statement that “significant differences between right and left breasts were observed across CBT groups” is misleading, as the statistical tests described compare CC vs MLO projections rather than right vs left symmetry; the analytical framework for side-to-side comparison is not properly defined.

Response 3: We thank the reviewer for this important observation. We agree that the original statement may be misleading, as the statistical analysis primarily focused on comparisons between CC and MLO projections rather than a formal side-to-side (right vs left) analysis. The statement has been revised to accurately reflect the analytical approach used in the study.

 

• Comment 4: The introduction overstates the radiation risk associated with mammography in a manner that is not sufficiently balanced. Statements such as mammography being “linked to a high risk of breast cancer” are not aligned with established epidemiologic evidence and may misrepresent the risk-benefit balance of screening. The discussion of national and international diagnostic reference levels (DRLs) is extensive but not well integrated into a clear research gap. The authors claim limited local data, yet do not clearly define how their study advances beyond existing DRL or AGD literature.

Response 4: We thank the reviewer for this constructive comment. The introduction has been revised to provide a more balanced and evidence-based discussion of radiation risk, emphasizing the well-established benefit–risk profile of mammography. In addition, we have clarified the specific contribution of this study, particularly in terms of CBT-stratified and projection-specific analysis using locally derived clinical data. We believe these revisions improve the clarity and relevance of the introduction.

Comment 5: In the materials and methods section, the single-center design using a single mammography unit severely limits generalizability, yet this limitation is not adequately accounted for in the study design. The reliance on a single vendor system with automatic exposure control (AEC) introduces systematic bias, particularly in exposure parameter selection, making correlation analyses between AGD and technical factors partially tautological. The description of the mammography system is detailed but does not include calibration data, quality control metrics, or temporal stability of the system over the study period.

Response 5: We thank the reviewer for the insightful comment. We have clarified that the single-center design and use of a single mammography system may limit generalisability. We have stated that the system operates under routine quality control procedures, and this has been clarified in the Methods section.

• Comment 6: The patient selection process raises concerns regarding potential selection bias. A large proportion of patients (over 60%) were excluded, yet the rationale for excluding “limited sample sizes in CBT categories” is unclear and methodologically problematic, as this introduces post hoc selection based on outcome distribution. Furthermore, excluding cases with CBT >100 mm and certain compression force values may systematically remove clinically relevant extremes, biasing the dose distribution.

 

Response 6: We thank the reviewer for this important comment. The exclusion criteria were applied primarily to ensure consistency in imaging technique and to enable reliable comparisons across standardized groups, which consequently reduced the final sample size. Regarding the exclusion of CBT categories >100 mm and those with limited sample sizes, this was undertaken to avoid unstable statistical estimates and unreliable group comparisons, rather than to selectively influence the outcome distribution. We have revised the manuscript to include a clarifying statement in the Patient Selection section: “Exclusion criteria were applied to ensure consistency in imaging technique and reliability of comparisons across groups.”

• Comment 7: The definition and calculation of DRLs are not methodologically consistent. The manuscript states that AGD per view was averaged to obtain a single value per examination, yet later reports projection-specific DRLs. This inconsistency suggests a lack of clarity in how DRLs were derived and whether they follow internationally accepted methodologies (e.g., per-view vs per-exam DRLs). Additionally, the sample size used in Table 1 (609 exams) does not align with the previously stated 607 patients and 2,428 projections, indicating inconsistency in dataset definition.

Response 7: We thank the reviewer for this valuable comment. We have revised the manuscript to improve clarity and methodological consistency regarding the definition and calculation of DRLs. Specifically, we have corrected the typographical error in the dataset description (607 replaced with 609 patients) to ensure consistency across the manuscript.

In addition, we have clarified the methodology for DRL estimation by adding a clear definition in the DRL section: “In accordance with international recommendations, the DRL was defined as the 75th percentile of the AGD distribution for each projection.” We believe these revisions address the reviewer’s concerns regarding clarity and internal consistency.

 

• Comment 8: The statistical analysis is fundamentally limited. The use of Mann–Whitney U tests for comparing projections is questionable, as CC and MLO views from the same patient are not independent samples; a paired statistical test would have been more appropriate. Similarly, the absence of multivariate regression analysis is a major limitation, given the known collinearity between CBT, mAs, and kVp. The reported correlations, particularly extremely high Spearman coefficients (ρ ≈ 0.995 between mAs and AGD), strongly suggest mathematical coupling rather than true physiological relationships, as AGD is directly derived from exposure parameters.

 

Response 8: We thank the reviewer for this valuable comment. The analysis has been revised to use the Wilcoxon signed-rank test for paired CC and MLO comparisons. We acknowledge the lack of multivariate analysis and note this as a limitation, while our CBT-stratified approach was used to reduce confounding. The very high correlation between mAs and AGD is recognised as likely due to mathematical coupling, and interpretations have been revised accordingly.

• Comment 9: The reporting of results demonstrates internal inconsistencies. For example, statements about “very highly significant differences between both sides” contradict the numerical values presented, where differences between right and left breasts are minimal. The manuscript conflates statistical significance with clinical relevance without quantifying effect sizes.

Response 9: We thank the reviewer for this important comment. We acknowledge that the wording may have overstated the significance of the findings. The manuscript has been revised to remove ambiguous terms such as “very highly significant” and to better distinguish between statistical significance and clinical relevance. The results are now presented more clearly and interpreted with appropriate caution.

• Comment 10: The discussion largely reiterates known principles of mammographic physics rather than providing novel insights. Many statements are descriptive and confirmatory, without critically engaging with the limitations of the dataset or the analytical approach. The attribution of higher AGD to AEC behavior is speculative and not supported by direct analysis of AEC modes or system settings.

Response 10: We thank the reviewer for this valuable comment. We have revised the Discussion to reduce descriptive and confirmatory statements and to improve critical analysis of the findings. The statements regarding the influence of the AEC system have been moderated to avoid speculative interpretation.

 

• Comment 11: The comparison with national DRLs is superficial. The manuscript notes that AGD values are slightly above Saudi reference levels, but does not perform a formal benchmarking analysis or percentile-based comparison consistent with DRL methodology. Additionally, the interpretation that values are “acceptable” lacks a defined threshold or regulatory context.

Response 11: We thank the reviewer for this important comment. We have clarified the DRL estimation methodology by explicitly stating that DRLs were defined as the 75th percentile of AGD and used for comparison with national reference levels. This information has been added to the DRL estimation section. 

• Comment 12: The discussion of compression force (CF) is inconsistent and at times contradictory. The manuscript reports weak correlations but then extrapolates broader conclusions about compression practices without sufficient evidence. The analysis does not account for the interaction between CF and CBT, which is critical in understanding dose optimization.

Response 12: We thank the reviewer for this important comment. The Discussion has been revised to ensure consistency and to avoid unsupported conclusions.

 

• Comment 13: The limitations section is insufficiently critical. While the single-center design is acknowledged, other major limitations, such as selection bias, lack of multivariate analysis, non-independence of observations, and potential errors in statistical interpretation, are not addressed.

Response 13: We thank the reviewer for this valuable comment. We have revised the manuscript to further strengthen the Limitations section and improve the clarity and depth of the discussion on study constraints and future research directions. In particular, we have expanded the section to include additional methodological considerations and to better contextualize the findings.

In the revised manuscript, we have also added a statement highlighting the importance of incorporating patient-specific and operator-dependent factors in future research, including breast composition, body mass index (BMI), and radiographer-dependent positioning techniques, to improve the accuracy of individualised dose estimation and optimisation strategies.

• Comment 14: The conclusion overstates the findings, particularly the claim that mAs is the “dominant factor” influencing AGD, which is inherently true by definition and not a novel contribution. The statement that CF has negligible impact is not adequately supported by the limited and potentially confounded analysis presented.

Response 14: We thank the reviewer for the comment. The conclusions have been revised to avoid overstatement, clarify that the role of mAs reflects its inherent relationship with AGD, and present the impact of compression force (CF) more cautiously.

 

Comment 15: The manuscript also contains multiple inconsistencies in data reporting, including discrepancies in sample size, unclear definitions of units (per view vs per exam), and inconsistent terminology (e.g., mixing AGD, ESD, and exposure parameters without clear hierarchy).

Response 15: We thank the reviewer for highlighting these important issues. We have carefully reviewed the manuscript and revised it to correct inconsistencies in data reporting, including sample size discrepancies, and standardization of terminology. We believe these revisions have addressed the inconsistencies raised by the reviewer.

Comment 16: Table 1 contains implausible variability metrics, particularly the standard deviation of kVp (±9.54), which is unusually large for mammography and suggests either data extraction errors or inappropriate aggregation of heterogeneous protocols. Additionally, the table mixes system-level and patient-level metrics without a clear distinction, reducing interpretability.

Response 16: We acknowledge the reviewer’s concern regarding the large standard deviation for kVp in Table 1. This variability resulted from aggregation across all projections (CC and MLO) and across different automatic exposure control (AEC) selection ranges, rather than from a single fixed acquisition protocol.

Comment 17: In Table 2, the statistical comparisons are not clearly defined. The reported p-values appear to compare CC vs MLO within each side, yet the narrative text repeatedly interprets them as differences between right and left breasts. This represents a fundamental misinterpretation of statistical testing. Moreover, multiple comparisons across CBT categories are performed without adjustment, increasing the risk of type I error.

Response 17: We thank the reviewer for this important comment. We have revised Table 2 and the corresponding text to clearly define the statistical comparisons performed, ensuring correct interpretation of CC vs MLO within each breast.

Comment 18: Table 4 presents extremely high correlation coefficients between mAs and AGD, which are expected due to the computational relationship between these variables. However, the manuscript interprets this as a clinical finding rather than an inherent property of the AGD calculation algorithm. This reflects a lack of understanding of the underlying physics and introduces circular reasoning into the conclusions.

Response 18: We thank the reviewer for this insightful comment. We acknowledge that the strong correlation observed between mAs and AGD reflects the inherent physical and computational relationship between these variables, as mAs directly influences X-ray output and consequently the estimated glandular dose. The relevant sections in the discussion have been updated accordingly.

• Comment 19: The correlation analysis presented in Figure 1 lacks sufficient methodological detail. It is unclear whether the correlations were computed across pooled data or stratified by CBT categories, and the reported values for specific CBT ranges contradict earlier statements about overall correlations. The interpretation of “strongest” and “weakest” correlations is not supported by a consistent analytical framework.

Response 19: We thank the reviewer for this important comment. We have revised the manuscript to improve clarity and consistency. The description has been updated to ensure alignment between the text and Figure 1, and the interpretation of correlation strength has been revised to follow a consistent analytical framework throughout the manuscript.

 

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

This paper examines the relationship between average glandular dose (AGD) and compressed breast thickness (CBT) and several imaging parameters in mammography. The experiments in the study were conducted on data obtained from 607 patients. The results showed that AGD increased with CBT and that higher dose values ​​were observed in MLO projections. The study can be considered clinically valuable in terms of radiation dose optimization and quality assurance. However, it is observed that the paper needs significant improvements in its current form and has some generalization limitations:
- The Introduction section of the article is superficial. Given the absence of a section on relevant studies, this section needs improvement to more clearly present the original contribution.
- The contributions of the article to the literature should be detailed by listing them point by point. Since the focus of this study, which examines the relationship between MGD and CBT and imaging parameters, has been addressed in many studies, the unique contributions of this study should be emphasized. It should be stated in what way it differs from existing research.
- The methodology of the article is superficially described in Section 2. It should be conveyed to the reader with a visual aid or flowchart.
- Details regarding data collection are missing. Patient selection or exclusion should be detailed. Whether the scans were repeated or for screening/diagnostic purposes should be specified.
- The fairness of experimental results should be reviewed. Potential confounding variables should be controlled. Factors that may affect AGD, such as patient age and breast density, need to be controlled.
- Tests need to be strengthened with various statistical analyses. Visual and tabular presentations should be improved. AGD vs CBT graphs and scatter or boxplot comparisons can be added.
- The article does not provide a rationale for the CBT ranges determined. What is the reason for choosing these ranges? If they are standards, this should be stated. Also, the specific formula or software used in AGD calculation should be detailed.
- The results related to CF in the article should be analyzed in depth. The results have not been compared with the literature. They should be compared with similar studies.
- The ALARA principle is used in the article. However, no evaluation has been made regarding how the parameters can be optimized in practice.

Author Response

We sincerely thank the reviewers for their valuable comments. We have responded to each point individually and revised the manuscript accordingly. All changes have been clearly highlighted in the revised manuscript.

Comment 1: The Introduction section of the article is superficial. Given the absence of a section on relevant studies, this section needs improvement to more clearly present the original contribution. The contributions of the article to the literature should be detailed by listing them point by point. Since the focus of this study, which examines the relationship between MGD and CBT and imaging parameters, has been addressed in many studies, the unique contributions of this study should be emphasized. It should be stated in what way it differs from existing research.

Response 1: We thank the reviewer for this constructive comment. We have clearly outlined the specific contributions of this study and emphasised how it differs from existing work, particularly in terms of its focus on local data, stratification by CBT, and projection-specific analysis under routine clinical conditions. These revisions aim to improve the clarity and originality of the study’s contribution.

Comment 2: The methodology of the article is superficially described in Section 2. It should be conveyed to the reader with a visual aid or flowchart.

Response 2: We thank the reviewer for this valuable suggestion. While we appreciate the benefit of including a visual aid, we believe that a flowchart is not necessary for this study. The methodology is already clearly structured and presented using subheadings that systematically describe each stage of the work, including study design, ethical approval, patient mammography system description, data collection, patient selection, DRL estimation, and statistical analysis. This organization allows the reader to follow the study design and workflow clearly and logically without the need for an additional figure.

Comment 3: Details regarding data collection are missing. Patient selection or exclusion should be detailed. Whether the scans were repeated or for screening/diagnostic purposes should be specified.

Response 3: We thank the reviewer for this comment. Details regarding patient selection are provided under the “Patient Selection” subheading in the Methods section. However, to address the reviewer’s concern, we have revised and expanded this section to more clearly describe the nature of the examinations (screening/diagnostic). These revisions improve the clarity and transparency of the methodology, and the updated information has been incorporated into the revised manuscript.

 

Comment 4: The fairness of experimental results should be reviewed. Potential confounding variables should be controlled. Factors that may affect AGD, such as patient age and breast density, need to be controlled.

Response 4: We thank the reviewer for this important comment. In response, we have clarified how potential confounding variables were addressed in the study. Compressed breast thickness (CBT) was used as the primary stratification variable, as it is the main determinant of average glandular dose (AGD). Analyses were therefore performed within CBT categories to minimise variability related to breast size and tissue composition. Breast density data were not consistently available in the dataset and could not be included; however, CBT was used as a practical surrogate, as it reflects breast size and attenuation characteristics relevant to dose estimation.

 

Comment 5: Tests need to be strengthened with various statistical analyses. Visual and tabular presentations should be improved. AGD vs CBT graphs and scatter or boxplot comparisons can be added.

Response 5: We thank the reviewer for this valuable suggestion. The current analysis already includes the key statistical assessments required to address the study objectives, including stratified correlation analysis across CBT ranges and projection-specific comparisons. We will therefore focus on improving clarity in the existing figures.

Comment 6: The article does not provide a rationale for the CBT ranges determined. What is the reason for choosing these ranges? If they are standards, this should be stated. Also, the specific formula or software used in AGD calculation should be detailed.

Response 6: We thank the reviewer for this important comment. In response, we have clarified the rationale for the CBT categorisation in the Methods section. Importantly, our results further support the appropriateness of this stratification, as shown in Figure 1, which demonstrates a consistent and statistically significant positive monotonic relationship between CBT and AGD across all projections. This confirms that the chosen CBT intervals appropriately capture meaningful variations in dose behaviour with increasing breast thickness.

We have also clarified the AGD calculation method under “Mammography system description” in the methodology section, which is automatically derived using the built-in Dance algorithm from exposure parameters and breast thickness data and retrieved from the system DICOM headers. In addition, we have added the appropriate reference for the Dance algorithm in the revised manuscript as suggested by Reviewer 2.

Comment 7: The results related to CF in the article should be analyzed in depth. The results have not been compared with the literature. They should be compared with similar studies.

Response 7: We thank the reviewer for this valuable suggestion. In response, the revised manuscript now includes relevant literature to place our findings within the existing evidence base.

Comment 8: The ALARA principle is used in the article. However, no evaluation has been made regarding how the parameters can be optimized in practice.

Response 8: We thank the reviewer for the comment. While this study did not include formal parameter optimization, the manuscript has been revised to better relate the findings to the ALARA principle and their potential relevance for future optimization of mammography practice.

 

 

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have adequately addressed most of the reviewer comments and improved the manuscript, particularly in terms of discussion and clarity. The study design and statistical approach are acceptable for the stated objectives, and limitations are largely acknowledged.

However, before acceptance, the authors should address the following minor points:

1. Explicitly acknowledge the potential impact of mathematical coupling between AGD and exposure parameters in the limitations.
2. Clarify that the reported correlations do not imply causal relationships due to the absence of multivariate adjustment.

With these minor clarifications, the manuscript is suitable for publication.

Author Response

Reviewer comment: The authors have adequately addressed most of the reviewer comments and improved the manuscript, particularly in terms of discussion and clarity. The study design and statistical approach are acceptable for the stated objectives, and limitations are largely acknowledged.

However, before acceptance, the authors should address the following minor points:

1. Explicitly acknowledge the potential impact of mathematical coupling between AGD and exposure parameters in the limitations.
2. Clarify that the reported correlations do not imply causal relationships due to the absence of multivariate adjustment.

With these minor clarifications, the manuscript is suitable for publication.

 

Response: We sincerely thank the reviewer for the constructive feedback and for recognizing the improvements made to the manuscript.

In response to the reviewer’s comments, we have incorporated the requested revisions, which are highlighted in the Limitations section (Pages 10–11, Lines 404–409).

We believe these revisions have further strengthened the manuscript and adequately addressed the reviewer’s remaining concerns.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The revisions have improved the clarity, structure, and overall presentation of the work. In particular, the authors have made a clear effort to address the previously raised concerns, and the responses are generally appropriate and satisfactory. Overall, the comments have been addressed adequately.

Author Response

We sincerely thank the reviewer for the positive evaluation of our revised manuscript. We appreciate the reviewer’s recognition of our efforts to improve the clarity, structure, and overall presentation of the work, as well as their acknowledgment that the comments have been addressed satisfactorily.

Reviewer 4 Report

Comments and Suggestions for Authors

It appears that the authors have fully implemented the suggested corrections. I believe the paper doesn't need further improvements before publication.

Author Response

Reviewer comment: It appears that the authors have fully implemented the suggested corrections. I believe the paper doesn't need further improvements before publication.

Response: We sincerely thank the reviewer for the careful evaluation of our manuscript and for acknowledging that the suggested corrections have been fully implemented. We appreciate the reviewer’s positive feedback and support for the publication of our work.