Risk estimation algorithm reduces noise in MR images

Pairing the established denoising algorithm NeighShrink with chi-square unbiased risk estimation (CURE) was superior to conventional methods at reducing noise in MR images, reported researchers of a study published in Artificial Intelligence in Medicine.

Typical methods for MRI denoising, such as filtering, transform domain and statistical methods, are unable to produce good results for Rician noise. Even NeighShrink alone, which has successfully reduced white Gaussian noise, falls short when it comes to Rician noise, according to first author Chang-Jiang Zhang of Zhejiang Normal University in Jinhua, China.

Applying the combined NeighShrink algorithm with CURE (NeighShrinkCURE) and adding bilateral filtering and fast cycle-spin technology to MR images resulted in improved peak signal-to-noise ratio and structural similarity compared to other denoising methods. The results were better than two wavelet domain noising algorithms—Iterative bilateral filtering and LMMSE.

“Both quantitatively and qualitatively, the results show the efficiency of the proposed algorithm to MR image denoising,” Zhang et al. wrote. “Note that simple threshold shrinkage denoising methods optimized by CURE can also yield good results, however, the NeighShrink method makes use only of intra-scale dependencies between wavelet coefficients. Further denoising gains are likely to improve NeighShrink by introducing inter-scale dependencies.

Read the entire story in AI in Healthcare.