Denoising Diffusion-Weighted MR Images Using Low Rank Structure and Edge Constraints

Keywords

absence abstract access account accurately achieving acquisition address advantage advantages alternating anisotropy applied applying approaches artifacts attention authors champaign characterization coded coil coils columns combine combines comments complementary complex components computer consistent constant constraints convenient correlated correlation determines diffusion diseases domain edge effectiveness effort eigenvector either electrical encoding engineering estimation evaluated expect experimental exploiting features finite form formulation frame frames free full function goal great illustrates illustration important improvement independent institute integrating intrinsic issue iteratively joint jointly kind likelihood limitation made magnitude martin matrices matrix medical model modeling newton noise noisy novel operator optimization orientation originally partial penalized performance precision previously primary prior problem properties property propose proposed quasi rank reconstructed reconstruction recovering recovers reduces reducing referred reflecting representative rows scanner sets settings side simulation smoothing solely solve specifically spin square squares stage stands strength structure superior taken taking tensor tensors thank theoretical theory useful utilizing valuable variance various vector zero