Thank you for the interesting work. I have a question regarding the use of sinograms as prior knowledge. Since sinograms are line integrals of the CT image along projection rays, they do not directly encode spatial structural information. Therefore, using them to extract structural priors might not be theoretically sound. Could you kindly explain to a beginner like me how the harmonizer uses the sinogram to condition the INR? And do you think other types of priors, like image-domain features, might be more suitable?
Thank you for the interesting work. I have a question regarding the use of sinograms as prior knowledge. Since sinograms are line integrals of the CT image along projection rays, they do not directly encode spatial structural information. Therefore, using them to extract structural priors might not be theoretically sound. Could you kindly explain to a beginner like me how the harmonizer uses the sinogram to condition the INR? And do you think other types of priors, like image-domain features, might be more suitable?