Sparse representation-based MRI super-resolution reconstruction
Citations
378 citations
281 citations
243 citations
Cites methods from "Sparse representation-based MRI sup..."
...We quantitatively compare the various resolution-enhancement methods using image quality metrics and qualitatively compare the methods through a reader study to evaluate the diagnostic potential of DeepResolve....
[...]
...An L2 loss function was chosen for DeepResolve....
[...]
...Such methods could also be especially useful for newer implementations of DESS that enable simultaneous T2 relaxometry, morphometry, and semiquantitative radiological assessment.47 The thicker slices could be used for generating high-SNR for quantitative T2 measurements, whereas the thin slices could be used for accurate morphometry and semi-quantitative whole-joint assessment.48 Bilateral knee imaging methods that acquire several hundred slices could also benefit from DeepResolve.49 The training data for DeepResolve consisted of 34% of patients with a KL OA grade of 2 and 59% of patients of grade 3....
[...]
...In addition to the TCI image inputs to DeepResolve, we also generated FI images from the simulated thick-slice images.37,38 For comparing against a state-of-the-art single MR image superresolution method, we generated ScSR images for the same 3443 3443 160 imaging volume as DeepResolve.22 The ScSR method creates sparse residual images using a 2D patch-based dictionary approach that iteratively tries to enhance low-resolution features based on image pairs of low resolution and high resolution (detailed description in Supporting Information)....
[...]
...For comparing against a state-of-the-art single MR image superresolution method, we generated ScSR images for the same 3443 3443 160 imaging volume as DeepResolve.(22) The ScSR method creates sparse residual images using a 2D patch-based dictionary approach that iteratively tries to enhance low-resolution features based on image pairs of low resolution and high resolution (detailed description in Supporting Information)....
[...]
150 citations
Cites background from "Sparse representation-based MRI sup..."
...In general, three requirements exist in a successful CS application: sparse representation (Wang et al., 2014; Zhang et al., 2012), incoherent undersampling artifacts (Greiser and von Kienlin, 2003; Tsai and Nishimura, 2000), and an effective nonlinear reconstruction algorithm (Aelterman et al.,…...
[...]
96 citations
References
22 citations
16 citations
11 citations
10 citations