Artifact Reduction in Compressed Sensing Averaging Techniques for High-Resolution Magnetic Resonance Images
Round 1
Reviewer 1 Report
The context of the research is MRI k-space imaging. When high-resolution imaging is required, using compressed sensing is a way to reduce the acquisition time. The paper adopts the concept of CSA which is a method for reducing the noise. By averaging multiple noise-affected imaging results, the noise can be reduced and the image quality can be improved. The proposed CSAKC method copies all the high-frequency k-space lines from subset 1 to subset 2 and 3. The results show that it reduces the artifacts (ringing and blurring), comparing to the CSAM and CSAK method.
Reducing noise by averaging images implicitly assumes that (1) the source image is stable enough; (2) the only unstable source(s) that affects the acquired result can be modeled as noise. For CSA based methods, the image quality can be improved if the under-sampling ratio is not too low. When we have insufficient k-space lines, probably the quality degradation is mainly cause by data loss, not noise. From table 1, the lower part shows the amount of data for reconstructing each image. It is obvious that CSAKC uses more data (even the data is copied) for reconstructing s1, s2, and s3. Therefore, the resulting artifacts are not as serious as those of the other two methods.
CSAKC keeps the acquisition time low and generates less visual artifacts. This is an interesting application of CSA concept. However, the concern is that all the high-frequency lines are copied from the first acquisition phase (s1). Sometimes different acquisition phases may be affected by different environmental conditions, such as temperature. It is not easy to justify that the reconstruction result is close to the expected one (the real one or the ground-truth one). Adding the comparisons to the full k-space reconstruction results would be better. The readers can justify the effectiveness of the three CSA based methods easier.
Author Response
Thank you for your positive and overall summary of our paper. As you concerned, phases can be affected by various environmental conditions. Therefore, we included your concern in this revised manuscript. In this study, we compared the full k-space reconstruction with different under-sampling ratios as shown in Figures 1 and 5. But further quantitatively evaluation whether the reconstructed images are close to the expected and/or ground-truth results should be followed.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The paper presents a compressed sensing averaging technique for the reduction of blurring and/or ringing artifacts depending on the k-space sampling ratio. This technique copies all high-frequency components of the undersampled first subset to the remaining subsets obtained only with keyhole data.
The method is described in detail. The results are well explained. The Disscussion is explained results.
I have only one comment: into Concussions, please add the limitation of the proposed technique. Authors describe that results must be evaluated by the doctors (l 236-239). However, it would be great to add how the proposed method can be used in the future from a scientific point of view.
Author Response
Thank you for your positive and valuable comments. We included some limitations and future applications in this revised manuscript as the reviewer suggested.
Author Response File: Author Response.docx
