The manuscript addresses an important methodological question regarding the use of medical CT scans for quantifying volumetric ice content in frozen sediments, which is highly relevant for permafrost research. The dataset is substantial, and the comparison with laboratory measurements is valuable. 

However, I believe that minor revisions are needed to clarify several aspects of the interpretation before the manuscript is ready for publication. Below, I outline comments and questions aimed at clarifying the interpretation of CT-derived versus laboratory-measured volumetric ice content, with particular attention to the direction and magnitude of bias, the proposed influence of organic matter, and the consistency between the figures and the accompanying discussion.

-L14: According to Figure 1, it appears that CT overestimates VIC in ice-poor sediments and underestimates it in ice-rich sediments. The text seems to describe the opposite trend and should be revised for consistency.

-L31: This sentence is unclear and would benefit from rephrasing.

-Line 69: Consider adding a typical HU range for organic matter to provide context for the classification.

-Line 71: Lapalme et al. (2017) suggested that CT imagery may be better suited to estimating EIC, particularly when pore-space diameters are within pixel resolution. Why was CT segmentation not compared with EIC in this study? Clarification would be helpful.

-Line 77: Please provide justification for the chosen organic matter classification thresholds (<10%, 10–20%, >20%).

-Line 87: The statement “Underestimation exceeded 60% at the low end of the 𝑉𝐿𝑎𝑏 𝑖𝑐𝑒 range, while overestimation reached about 30% in ice-rich, organic-rich samples” appears inconsistent with Figure 1b, which suggests the opposite pattern. This should be reconciled.

-Line 91: The text states systematic underestimation in ice-poor samples and overestimation in ice-rich samples. Based on Figure 1, the trend appears reversed.

-Line 95: The phrase “…indicating systematic underestimation” appears inconsistent with the plotted data, which suggest overestimation.

-Line 99: The statement that low-OM samples were underestimated and high-OM samples overestimated does not appear to align with Figure 1. Please clarify.

-Line 122: The general statement regarding systematic misestimation should reflect the direction of bias observed in Figure 1.

-Line 124: The explanation invoking mixed voxels would typically result in an underestimation of ice content. However, Figure 1 appears to show overestimation in ice-poor samples. Please clarify how this mechanism explains the observed bias.

-Line 129: The manuscript states that organic matter misclassification produces overestimation. However, Figure 1 does not clearly demonstrate that ≥20% OM samples exhibit the strongest overestimation; many appear underestimated. Quantitative support or clarification would strengthen this interpretation.

-Line 137: The summary statement describing underestimation in ice-poor samples and overestimation in organic-rich samples appears inconsistent with Figure 1 and should be revised.

This is a well-written manuscript that addresses the limitations of medical-CT in terms of routine application and quantitative analysis within permafrost research. However, please find below a few minor revisions that require clarification, particularly regarding the interpretation of the results. 

L58: Please write out the full term before using “ADAPT” as abbreviation. Additionally, provide more details on the sample dimensions, such as the core diameter range and length range of the collected specimens.

L61: Was the sample temperature maintained at  0 °C during scanning? If not, could melting of ice and associated phase transitions have affected the results?

L64: Please provide more details on the scanning parameters, such as tube voltage, tube current, and exposure time, as these directly influence the reconstructed Hounsfield Unit (HU) values.

L65: Please also provide more details on the image processing. Were any noise filtering steps applied to the data prior to segmentation? Furthermore, were any binary or morphological operations performed after segmentation before measuring the ice, sediments, and gas phase?

L66: Please provide more detailed explanation of the Regions of Interest (ROIs) definition process. The term “ROI” is commonly used to denote a specific area of the sample selected for further analysis, while a “trained classifier” in image processing usually refers to an AI-assisted algorithm trained to segment a specific phase.

L67: There appears to be an overlap of 1 HU between the gas-ice and ice-sediment ranges. This implies that some voxels may be counted twice, either as gas and ice or as ice and sediment. Could you clarify this HU overlap? Would it potentially lead to an overestimation of the phase volumes?

L75: Could you justify the selection of half of the total samples (81 out of 261) for the determination of the organic matter (OM)?

L87-88: The underestimation of 60% is stated here; however Figure 1(b) suggest a larger overestimation of the CT-derived volumetric ice content (). Could you please check whether this is correct, or maybe this is due to the typo on the x-axis label that differs from the definition in the text? Additionally, please clarify the 60% and 30% values-are they percentages out of the 261 samples?

L90: Similar to Lines 87-88, please clarify this point. The figure caption states an underestimation of CT-derived volumetric ice content () for ice-poor samples measured from laboratory. However, figure 1(a) appears to indicate an overestimation of the .

L90: To enhanced visualization, please maintain the same legend style for presenting the organic matter content (%) between figure 1(a, text) and (b, colormap)?

L102: Could the “widely varying thresholds” originate from differences in scanning parameters across different studies, as well as variations in ice type (e.g., pure ice vs. Sea ice), and sample composition? It may also be helpful to include this information in Table 1, as these parameters directly influence the reconstructed HU values.

L103: Could you please elaborate on how the two representative samples were selected? Were they chosen because their CT-derived volumetric ice content () is similar to the laboratory measured ice content()? Or were they selected simply to represent two different localities (Bylot Island and Beaver Creek)?

L134: Could you elaborate further on the calibration against reference data?

The Brief Communication: Limitations of Medical X-ray Computed Tomography for Estimating Ice Content in Permafrost Samples by Mahya Roustaei, Jonas Darey, Zakieh Mohammadi, and Daniel Fortier is well written, but also very concise.

Below you can find my main comments:

• The abstract does not fully show what has been done. It gives the impression that high-resolution micro-CT experiments have been done as well, and that there has been a comparison with the medical CT. However, this is not the case. Furthermore, high-resolution micro-CT is not discussed in the document at any moment. Therefore, I suggest making the abstract clearer and adding a paragraph in the discussion section about high-resolution micro-CT on ice cores. Does it really quantify the ice fraction better compared to medical CT? Has this been tested? If not, you can add it in future research, as this is missing as well.
• Another sentence of the abstract is, in my opinion, not discussed later as well: “we conclude that medical CT is better suited for visualizing cryostructures and heterogeneity than for routine quantification of ice content”. Can you elaborate more on this in the discussion or introduction? For which structures has it been successfully used?
• Introduction and also in other sections: make a distinction between medical-CT and high-resolution micro-CT, as it is sometimes confusing (don’t use just CT). Furthermore, it is important to note in line 41 that for high-resolution micro-CT or micro-CT, the voxels have the same size in all dimensions (in contrast to medical CT)
• Line 64-65: The voxel size was not constant for each sample? Did this affect the results of the ice quantification?
• Line 67-68: Did you visually check if the chosen HU ranges work well for your samples? Do they visually capture the phases of interest? How do you know that your chosen HU units are better or worse than the ones in Table 1? Otherwise, you can assume that it was just misclassified.
• Table 1: The variation seems so large that it suggests to me that medical CT is unable to quantify the ice content at all (and not just limited). How can the correct HU units be chosen? For me, this is a bigger issue than the misestimation of the ice phase. I think this should be addressed in more detail, and/or the discussion should be more critical (e.g., the final statement in line 140 should be stronger, including the title).