Review Reports - Visibility of Intracranial Perforating Arteries Using Ultra-High-Resolution Photon-Counting Detector Computed Tomography (CT) Angiography

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Manuscript title
"Visibility of intracranial perforating arteries using ultrahigh-resolution photon-counting detector computed tomography (CT) angiography"

1. The main question addressed by the research is visibility of small intracranial arteries on contrast-enhanced photon-counting ultrahigh-resolution computed tomography, including various kernels.

2. The manuscript may be relevant to the field of CT imaging, neuroradiology and quality assurance. However, the potential clinical and academic usability of chief findings may be limited by several methodological shortcomings.

3. The manuscript expands prior published material by demonstrating via a retrospective study with two radiologists and a Likert-like scale that UHR PCD-CTA provided good visualization of perforating arteries, with kernel Hv72 providing optimal SNR and CNR.

4. The specific improvements to methodology include clarifications regarding:

- comments on sample size estimation
- comments on the need for two statistical packages (MedCalc and R)
- comments on data normality assessment
- expanding Discussion section with information on small intracranial artery visualization with magnetic resonance imaging
- expanding study limitations with information on small sample size

5. The conclusions are consistent with the evidence provided, given statistical considerations.

6. The references are appropriate.

7. The figures and tables are appropriate.

Author Response

1. The main question addressed by the research is visibility of small intracranial arteries on contrast-enhanced photon-counting ultrahigh-resolution computed tomography, including various kernels.

Reply: Yes. This research addresses the improved visualization of intracranial perforators on photon-counting CT angiography with ultrahigh-resolution mode.

2. The manuscript may be relevant to the field of CT imaging, neuroradiology and quality assurance. However, the potential clinical and academic usability of chief findings may be limited by several methodological shortcomings.

Reply: In this study, we assessed the visibility of the perforators according to the kernel and iterative reconstruction. We confirmed the good visibility of the perforators, which may aid in detailed CT angiography assessments before surgery and in understanding vascular disorders, but currently we did not assess the clinical utility. This should be a task for the future.

3. The manuscript expands prior published material by demonstrating via a

retrospective study with two radiologists and a Likert-like scale that UHR PCD-CTA provided good visualization of perforating arteries, with kernel Hv72 providing optimal SNR and CNR.

Reply: Yes. Thank you for your evaluation.

4. The specific improvements to methodology include clarifications regarding:

- comments on sample size estimation

Reply: The perforators themselves were only assessed qualitatively because direct quantitative analysis for such small arteries was difficult. Although the sample size was small, we assume an adequate minimum number of cases was secured to detect the difference in scores. This issue was added in the limitation section of Discussion as follows:

“The small sample size might affect the generalizability of the findings. However, we assume that an adequate minimum number of cases was secured to detect the difference in the assessed scores.”

- comments on the need for two statistical packages (MedCalc and R)

Reply: Data analysis, statistical analysis, and graph creation were done by MedCalc. Since MedCalc cannot analyze AC1, R was used only for this analysis. This issue was added in the Statistical analysis section as follows:

Interobserver agreement between the two readers was evaluated using Gwet’s AC1 due to imbalances in score frequencies [21], with analysis performed using R statistical software (version 4.3.1; R Foundation, Vienna, Austria) and the irrCAC package, because of the lack of this analysis in MedCalc.

- comments on data normality assessment

Reply: The dataset for signal-to-noise ratio and contrast-to-noise ratio contained both normally distributed and non-normally distributed data. Therefore, comparison analysis was performed by Friedman test. This issue was added in the Statistical analysis section as follows:

“SNR and CNR were compared among reconstruction methods for the LSA and PA regions, respectively, using the Friedman test with post-hoc pairwise comparisons, owing to the dataset comprising normally and non-normally distributed data.”

- expanding Discussion section with information on small intracranial artery visualization with magnetic resonance imaging

Reply: Thank you for your valuable comment. A new segment regarding magnetic resonance imaging was added in the Discussion section as follows:

“The perforators visibility was investigated on magnetic resonance (MR) angiography primarily on 5-T and 7-T MR units. The high-field MR units provide high SNR and high-resolution images. Harteveld et al.[30] assessed LSAs and PAs on contrast-enhanced MR angiography on 7-T MR unit and reported the mean (range) numbers of LSAs and PAs were 3.5 (1–8) and 5.0 (1–9), respectively. Kang et al.[31] investigated visibility of PAs on 7T-MR angiography and reported that the number of PAs was 7.14 ± 2.79. Shi et al.[32] qualitatively evaluated the visibility of LSAs and PAs on MR angiography and reported that excellent or adequate visibility for diagnosis were obtained in all cases for LSAs and more than half of the cases for PAs on 5T and 7T units. Thus, visibility of LSAs on UHD PCD-CTA is likely to be comparative with MR angiography on 5T or 7T. However, PA visibility may be lower on UHD PCD-CTA than MR angiography because precise counting of PAs was difficult in our UHD PCD-CTA. The advantages for MR angiography include no irradiation, and no necessity of contrast agents, whereas the disadvantages include contraindication for metal devices, relatively long scanning time, and less availability of such high-field MR units. Conversely, the CTA advantages include rapid scan, and no contraindication to patients with metallic devices; however, the disadvantages include irradiation, necessity of contrast agent, and bone related artifact particular in the posterior fossa. At present, there are no reports regarding the simultaneous comparison of the perforators between PCD-CTA and MR angiography. The CT and MR adaptations should be considered in the abovementioned issues.”

- expanding study limitations with information on small sample size

Reply: This issue was added to the limitation section as follows:

“The small sample size might affect the generalizability of the findings.”

“Further research is needed to determine the optimal reconstruction parameters, including kernel types, QIR levels, and matrix sizes with larger sample sizes.

5. The conclusions are consistent with the evidence provided, given statistical considerations.

Reply: Thank you for your evaluation.

6. The references are appropriate.

Reply: Thank you for your evaluation.

7. The figures and tables are appropriate.

Reply: Thank you for your evaluation.

Reviewer 2 Report

Comments and Suggestions for Authors

The main content of the manuscipt is about Photon-counting detector computed tomography (PCD-CT) for retrospective analysis of intracranial. The manuscript used a quantum iterative reconstruction (QIR) algorithm for the reconstruction of PCD-CT data of 30 patients. The peripheral visibility, vessel sharpness, and image noise were subjective evaluation. The signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were objective analyzed. The lenticulostriate arteries (LSAs) and pontine arteries (PAs) were assessed by the method. The issue is below:

1. The innovative of the manuscript is not clear, since the PCD-CT is a mature instrument.

2. The principle and experment system of the PCD-CT are wanted.

3. The quantum iterative reconstruction (QIR) should be introduced.

4. The deep analysis of the experimental data are necessary.

5. As the present writing, I believe it not suitable for the publication in Tomography Journal.

Author Response

1. The innovative of the manuscript is not clear, since the PCD-CT is a mature instrument.

Reply: Although the principle of PCD-CT has been known, its utility has not been fully elucidated, particularly for the small vessel visualization. We found a relatively good visibility of the brain perforators using high number of kernel and QIR.

2. The principle and experment system of the PCD-CT are wanted.

Reply: The principle of PCD-CT is described in Introduction. We agree with the need for experimental study of the small vessels. This issue was added in the limitation section in Discussion as follows:

“An experimental data using phantom may effectively investigate the relationship between the image quality and various reconstruction methods.”

3. The quantum iterative reconstruction (QIR) should be introduced.

Reply: The sentence for explanation for QIR was added in Material and methods, CT imaging section as follows:

“QIR is a new, tailored IR method for PCD-CT. In QIR process, the raw detector data is divided into two streams based on energy levels, with each stream entering its iterative loop for artifacts cancellation and noise reduction. To maintain precise geometric alignment between these energy levels, synchronization points are introduced in the projection and image data loops. Subsequently, these fully synchronized streams proceed through spectral processing to produce spectral maps and monoenergetic images [21] (Figure 1). QIR offers four levels of IR reconstruction strength. Herein, strength level 3 was used for QIR.”

4. The deep analysis of the experimental data are necessary.

Reply: same as #2.

Reviewer 3 Report

Comments and Suggestions for Authors

In this manuscript, Okazaki et al. applied kernels with different dimension and different reconstruction strategies to compare the visibility of intracranial perforating arteries with ultrahigh-resolution photon-counting CT angiogram (UHR PCD-CTA). The reviewer believes the scientific soundness of the manuscript can be further improved if the following comments can be addressed.

1. The methods section lacks technical details about the QIR algorithm and CT angiogram reconstruction. Since QIR seems to be an important factor to improve the image quality, it would be beneficial to summarize from high-level how it works and what were previous applications of this method. For CTA reconstruction, what is the frame rate (or volume rate), how did the authors address the motion artifacts, and how was the overall processing steps? Would be good if a flowchart and example images of the image processing pipeline can be inserted to better demonstrate the process.

2. Figure 1 only showed the score range and the median score. Have the authors ever considered using box plots like Figure 5? This can add more statistical information to the figure. Moreover, significance level can be indicated and overlaid, if exists.

3. Please add scale bars to Figure 2 and Figure 3.

4. The reviewer is wondering why is the standard deviation of artery considered to be the noise level in Line 126. Artery should be a part of the signal, since there is blood flow, hence presents as signal in angiograms. White matter, on the other hand, is relatively static and can be considered as the noise term. Usually SNR is defined as the signal level divided by the standard deviation of noise. The CNR is usually defined as the contrast divided by the standard deviation of the noise. SNR = HUartery/√𝑆𝐷white matter seems to be a better way to do this.

Author Response

Reply: Thank you for your valuable comments. Explanation sentences regarding QIR were added in Material and methods, CT imaging section. Also, a part of the sentences was modified. A flowchart for QIR explanation was added as Figure 1. Accordingly, the number of previous figures were renumbered.

Flow chart of QIR was provided as Figure 1.

Reply: Due to the low variance in scores, a box could not be formed, making the graph difficult to read (please refer to the sample graph with box-and-whisker plot). Therefore, bar charts were used. Significance levels were added in Figures.

3. Please add scale bars to Figure 2 and Figure 3.

Reply: Scale bars were added in Figures.

4. The reviewer is wondering why is the standard deviation of artery considered to be the noise level in Line 126. Artery should be a part of the signal, since there is blood flow, hence presents as signal in angiograms. White matter, on the other hand, is relatively static and can be considered as the noise term. Usually SNR is defined as the signal level divided by the standard deviation of noise. The CNR is usually defined as the contrast divided by the standard deviation of the noise. SNR = HUartery/√??white matter seems to be a better way to do this.

Reply: Thank you for pointing out. SNR estimation may be reasonable using noise as SDwhite matter. We estimated SNR according to the recent studies using SDartery below. We would like to use the current calculation method to ensure the consistency with the past studies.

1.Fransson V, Mellander H, Ramgren B et al. Image quality of spectral brain computed tomography angiography using halved dose of iodine contrast medium. Neuroradiology 2023;65:1333-1342.

2.Mergen V, Sartoretti T, Baer-Beck M et al. Ultra-high-resolution coronary CT angiography with photon-counting detector CT: feasibility and image characterization. Invest Radiol 2022;57:780-788.

3.Toth A, Chetta JA, Yazdani M et al. Neurovascular imaging with ultra-high-resolution photon-counting CT: preliminary findings on image-quality evaluation. AJNR Am J Neuroradiol 2024;45(10):1450-1457.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Although the manuscript has been revised as the requirements of the previous comments, there are still some aspects to be improved.

1. A large amount of data was introduced in the manuscript especially in the part of results. The table is strongly recommended since it is clear and concise.

2. The principle of the QIR is not clear, since it is the core of the manuscript. The detailed introduction is suggested including its principal figure, and the flow chart should be provided if necessary.

3. The figure.2 and figure.6 are not well discussed. What do the line with "*" stand for on the top of each subplot?

4. Figures 3 and 4 are raw data. In the manuscript, the authors compared them only using eyes. How to know one picture is clearer than another?

5. The innovative are still not clear. Although the authors answered: "particularly for the small vessel visualization. We found a relatively good visibility of the brain perforators using high number of kernel and QIR". It is necessary to compare with the latest research to illustrate the innovation of the manuscript.

Author Response

Although the manuscript has been revised as the requirements of the previous comments, there are still some aspects to be improved.

1. A large amount of data was introduced in the manuscript especially in the part of results. The table is strongly recommended since it is clear and concise.

Reply: Tables were provided for the corresponding results of Figures 2 and 6.

Reply: The principal and flow chart are described in “CT imaging” section and Figure 1. QIR is installed in NAEOTOM Alpha. It is possible to select whether to use QIR and the strength of its application at image reconstruction. This issue was added in “CT imaging” as follows:

“QIR is installed in NAEOTOM Alpha; thus, selecting whether to use QIR and the strength of its application at image reconstruction is possible.”

3. The figure.2 and figure.6 are not well discussed. What do the line with "*" stand for on the top of each subplot?

Reply: * indicates significant differences with post-hoc pairwise comparison. This explanation is described in each legend.

4. Figures 3 and 4 are raw data. In the manuscript, the authors compared them only using eyes. How to know one picture is clearer than another?

Reply: In this study, the visibility of the perforators was qualitatively assessed by two experienced neuroradiologists, focusing on peripheral visibility, vessel sharpness, and image noise. Interobserver agreement was generally good or almost perfect. High image quality correlates with good peripheral visibility, sharp delineation of vessels, and minimal noise. We assessed these factors. Due to the very thin nature of the perforators, only a visual qualitative assessment was conducted for them, which is a limitation of this study. This issue was added in the limitation section as follows:

“In addition, the perforators themselves were only assessed qualitatively because of their thin structures.”

Reply: Comparison of the previous study was added in Discussion as follows:

“Tóth et al. evaluated intracranial major arteries on UHR PCD-CT with kernel 40’s and QIR strength level 1. Excellent image quality was reportedly observed in most cases. However, the visibility of the perforators was not assessed.”

Other corrections

Results

The errors in the value notation have been corrected.

Discussion

Correction of spelling mistakes: UHD to UHR

The typographical errors have been corrected.

Round 3

Reviewer 2 Report

Comments and Suggestions for Authors

The key point is not answered. What is the QIR? Since it is the core of the manuscript and the authors compared the experimental results with and without QIR, I think the readers can hardly understand and gain from the article if the principle and procedure of QIR are not introduced. As the present version, I think it is not suitable for the publication.

Author Response

Reply: The QIR is a new model-based IR method designed specifically for PCD-CT. The explanation is added as follows: “QIR is a new model-based IR method designed specifically for PCD-CT. In QIR process, the raw detector data are divided into two streams based on energy levels, with each stream entering its iterative loop for artifacts cancellation and noise reduction. To maintain precise geometric alignment between these energy levels, synchronization points are introduced in the projection and image data loops. Subsequently, these fully synchronized streams proceed through spectral processing to produce spectral maps and monoenergetic images.”

QIR builds upon the principles of the previous Siemens IR method, Advanced Modeled Iterative Reconstruction (ADMIRE). However, the QIR algorism differs from ADMIRE due to its utilization of multi-energy data in PCD-CT. While the specific mechanisms of the QIR algorithm are proprietary and only limited information is publicly available, we were unable to provide a comprehensive description of QIR's underlying principles. Nevertheless, QIR is already integrated into commercially available PCD-CT system. Previous studies have shown that QIR enhances image quality across various anatomical regions, including the lung [1], coronary arteries [2], liver lesions [3-5], and hip joints [6]. Although these studies did not delve into the underlying principles of QIR, they highlighted noise reduction as its main benefit.

In our study, we also observed that QIR effectively reduces noise, particularly in brain parenchyma. However, we attribute the enhanced visibility of fine perforator vessels to the high-resolution capability of the PCD-CT detector itself, as peripheral vessel clarity was consistently achieved regardless of the reconstruction kernel or use of QIR. In this paper, we primary aim to demonstrate that PCD-CT provides superior visualization of fine vessels, including perforators, thereby reinforcing the technology's potential for vascular imaging.

References

Sartoretti, T.; Racine, D.; Mergen, V.; Jungblut, L.; Monnin, P.; Flohr, T.G.; Martini, K.; Frauenfelder, T.; Alkadhi, H.; Euler, A. Quantum iterative reconstruction for low-dose ultra-high-resolution photon-counting detector CT of the lung. Diagnostics (Basel) 2022, 12, 522, doi:10.3390/diagnostics12020522.
Vecsey-Nagy, M.; Varga-Szemes, A.; Schoepf, U.J.; Tremamunno, G.; Fink, N.; Zsarnoczay, E.; Szilveszter, B.; Graafen, D.; Halfmann, M.C.; Vattay, B.; et al. Ultra-high resolution coronary CT angiography on photon-counting detector CT: bi-centre study on the impact of quantum iterative reconstruction on image quality and accuracy of stenosis measurements. Eur J Radiol 2024, 176, 111517, doi:10.1016/j.ejrad.2024.111517.
Graafen, D.; Stoehr, F.; Halfmann, M.C.; Emrich, T.; Foerster, F.; Yang, Y.; Duber, C.; Muller, L.; Kloeckner, R. Quantum iterative reconstruction on a photon-counting detector CT improves the quality of hepatocellular carcinoma imaging. Cancer Imaging 2023, 23, 69, doi:10.1186/s40644-023-00592-5.
Racine, D.; Mergen, V.; Viry, A.; Eberhard, M.; Becce, F.; Rotzinger, D.C.; Alkadhi, H.; Euler, A. Photon-counting detector CT with quantum iterative reconstruction: Impact on liver lesion detection and radiation dose reduction. Invest Radiol 2023, 58, 245-252, doi:10.1097/RLI.0000000000000925.
Sartoretti, T.; Landsmann, A.; Nakhostin, D.; Eberhard, M.; Roeren, C.; Mergen, V.; Higashigaito, K.; Raupach, R.; Alkadhi, H.; Euler, A. Quantum iterative reconstruction for abdominal photon-counting detector CT improves image quality. Radiology 2022, 303, 339-348, doi:10.1148/radiol.211931.
Huflage, H.; Grunz, J.P.; Kunz, A.S.; Patzer, T.S.; Sauer, S.T.; Christner, S.A.; Petritsch, B.; Ergun, S.; Bley, T.A.; Luetkens, K.S. Potential of employing a quantum iterative reconstruction algorithm for ultra-high-resolution photon-counting detector CT of the hip. Radiography (Lond) 2023, 29, 44-49, doi:10.1016/j.radi.2022.09.010.