Physical characteristics of deep learning-based image processing software in computed tomography: a phantom study

ConclusionDLIPFBP reduced image noise while maintaining a resolution similar to commercially available MBIR and DLR. The slight spatial frequency shift offP in DLIPFBP contributed to the noise texture degradation suppression. The NPS suppression in the low spatial frequency range effectively improved the low-contrast detectability.
Source: Australasian Physical and Engineering Sciences in Medicine - Category: Biomedical Engineering Source Type: research