Super-Resolution Reconstruction Using Cross-Scale Self-similarity in Multi-slice MRI

Abstract

In MRI, the relatively thick slices of multi-slice acquisitions often hamper visualization and analysis of the underlying anatomy. A group of post-processing techniques referred to as super-resolution reconstruction (SRR) have been developed to address this issue. In this study, we present a novel approach to SRR in MRI, which exploits the highresolution content usually available in the 2D slices of MRI slice stacks to reconstruct isotropic high-resolution 3D images. Relying on the assumption of local self-similarity of anatomical structures, the method can be applied both to a single slice stack and to the combination of multiple slice stacks that differ in the orientation of their field of view. We evaluate the method quantitatively on synthetic brain MRI and qualitatively on MRI of the lungs. The results show that the method outperforms state-of-the-art MRI super-resolution methods.

Context-aware single image super-resolution using sparse representation and cross-scale similarity

“…To improve the precision of the comparison, we use corresponding HR example patches fH k i♢j g 9 k ¼ 1 instead of P i♢j to draw the comparison in Eq. (16), where HoG designates the histogram of oriented gradient (HOG) descriptor [28]. In our test, we find that such a descriptor performs better than the simple sum of square difference (SSD) descriptor when comparing the similarity between HR patches.…”

“…Incorporating Eqs. (15) and (16) into Eq. (14), we obtain the probability pðW i Þ, which is used to measure whether an LR patch is suitable for reconstruction by example-based SR on the basis of currently used image pyramids.…”

“…This example-based segment is combined with the multi-image SR approach to yield the final HR image. Plenge [16] also exploits the cross-scale self-similarity idea to realize the multi-slice MRI super resolution. In addition, Yang [17] proposes the in-place self-similarity which refines the recently proposed local self-similarity by proving that a patch in the upper scale image has good matches around its origin location in the lower scale image.…”

Performance evaluation of MIND demons deformable registration of MR and CT images in spinal interventions

“…Binning used downsampling factors ranging 1 – 11 voxels (in increments of 1) and gave a set of 11 MR I 0 images with the slice thickness varying from 0.9 – 9.9 mm (figure 3(a)). Such a broad range of slice thickness was selected in order to test performance at extremes (i.e., find where the algorithm might break down at very large slice thickness) and to cover the range of thickness investigated in previous work (Manjón et al 2010, Plenge et al 2013, Shi et al 2013, Jafari-Khouzani 2014, Ahmadi and Salari 2015). The CT I 1 image was acquired on a Somatom Definition Flash (Siemens Healthcare, Erlangen, Germany) using a scan technique of 100 kVp and 291 mAs, and reconstructed at 0.6×0.6×0.8 mm 3 with a size of 256×256×312 voxels (figure 3(b)).…”

Population Based Image Imputation

“…Unfortunately, clinical slices are often sampled too sparsely for functional interpolation, such as linear, cubic or spline [19], to succeed. Similarly, patch-based methods that rely on redundancy within a single scan to “hallucinate” missing fine scale structure [13–15] fail to produce anatomically plausible reconstructions at this level of sparsity. Superresolution algorithms that use multiple images of the same subject to improve a single scan [2,10,15] are unsuitable for clinical data where multiple similar acquisitions are not commonly available.…”