Two-dimensional Reconstruction and Three-dimensional Visualization of MRI Images Based on Compressed Sensing / 中国医学影像学杂志

ABSTRACT

Compressed sensing (CS) is a novel theoretical framework for information acquisition and processing. Taking advantages of the sparsity or compressibility of the signals inherent in the real world, compressed sensing can collect compressed data at the sampling rate much lower than that needed in Shannon's theorem based on random measurement matrix. This technique is used in medical imaging to accelerate MRI's scanning speed, reduce radiation dosage and alleviate patients' suffering. The whole process of the proposed algorithm was as follows firstly, the wavelet transform was applied to achieve sparse representation of medical images and reserve certain parts with maximal coefficients; secondly, the reconstruction based on CS theory were achieved according to the improved optimized orthogonal matching pursuit (OOMP) algorithm;finally, maximum intensity projection algorithm was used to achieve three-dimensional volume reconstruction. The experimental results demonstrated that our proposed two-dimensional reconstruction method was accurate and effective, which was verified qualitatively by the local detail magnification of images and quantitatively by peak signal-to-noise ratio and sectional comparison. Therefore, the three-dimensional reconstruction can be rather helpful in clinic diagnosis and treatment.