Fifty-seventh annual meeting of the American association of physicists in medicine
MO-FG-204-07: Nonlocal Total Variation Based Spectral CT Image Reconstruction
When divided into narrow energy bins, the projection data for spectral CT contains spectral information with reduced signal-to-noise ratio (SNR). When used together, the total projection data has improved SNR without spectral information. The purpose of this work is to develop an image reconstruction method that maximizes the utility of spectral data in terms of synergized spectral information and SNR.
The proposed method is based on nonlocal total variation (TV) (NLTV). That is, we first reconstruct a spectrally averaged image using total projection data without energy binning, and build the NLTV weights from this image that characterize the nonlocal image feature with good SNR, such as texture and fine structures. Then these NLTV weights are incorporated into a NLTV-based iterative spectral image reconstruction scheme using the binned projection data with reduced SNR, so that the weights serve as the reference to regularize the image quality during spectral image reconstruction. In addition, the spectral reconstruction incorporates the material decomposition by directly reconstructing material compositions instead of the conventional approach, that separates image reconstruction and material decomposition. The spectral image reconstruction is formulated as an iterative reconstruction method with the NLTV regularization. The split Bergman method is developed for solving the above sparse optimization problem.
The proposed NLTV-based method was validated using both simulated and experimental data. In comparison with FBP and TV- based method, NLTV had improved image quality, e.g., contrast-to-noise ratio (CNR) and modulation transfer function (MTF).
NLTV-based spectral CT image reconstruction is developed that maximizes the utility of spectral data in terms of synergized spectral information, and it offers significant improvement in CNR without scarifying spatial resolution of the image.
Jiulong Liu and Hao Gao were partially supported by the NSFC (#11405105), the 973 Program (#2015CB856000) and the Shanghai Pujiang Talent Program (#14PJ1404500).