Purpose: Prior-image-based reconstruction (PIBR) methods, which leverage high-quality patient-specific prior image information into the reconstruction of subsequent acquisitions, have shown great potential to reduce radiation dose and improve image quality for ultralow-dose CT. However, PIBR is challenged when the prior image and current image have differences in intensity due to kVp mismatch or scanner difference. In this work, we investigate a specific PIBR method â€“ normal dose image induced nonlocal means (ndiNLM) â€“ to address PIBR with such mismatched prior and achieve quantitative ultralow-dose CT imaging.
Methods: We present two corrective schemes for the original ndiNLM method, 1) a global histogram matching approach and 2) a local intensity correction approach, to account for the intensity differences between the prior and current images in PIBR. We validated the efficacy of the proposed schemes using monoenergetic images (50 keV and 70 keV) acquired from a dual-energy CT scanner of a lung cancer patient. Meanwhile, we utilized different CT slices to emulate anatomical changes between the prior image and the current ultralow-dose image.
Results: We observed that the traditional PIBR method introduces artifacts to the reconstruction when using abovementioned mismatched prior. In contrast, our proposed corrective schemes enabled PIBR to successfully handle intensity differences and anatomical changes between the prior and current images. We demonstrated that the proposed techniques permit PIBR to leverage the mismatched prior and achieve quantitative ultralow-dose CT reconstruction without artifacts from both low-flux and sparse-view acquisitions.
Conclusion: We have proposed and validated two corrective schemes enabling PIBR to leverage and handle mismatched prior. This work permits robust and reliable PIBR for CT data acquired using different settings or different scanners. The proposed methods have the potential to facilitate the application of PIBR in more sequential imaging studies that are faced with mismatched prior.