Efficient Reconstruction Technique for Multi-Slice CS-MRI Using Novel Interpolation and 2D Sampling Scheme

Abstract

Compressed Sensing (CS) theory breaks the Nyquist theorem through random under-sampling and enables us to reconstruct a signal from 10%-50% samples. Magnetic Resonance Imaging (MRI) is a good candidate for application of compressed sensing techniques due to i) implicit sparsity in MR images and ii) inherently slow data acquisition process. In multi-slice MRI, strong inter-slice correlation has been exploited for further scan time reduction through interpolated compressed sensing (iCS). In this paper, a novel fast interpolated compressed sensing (FiCS) technique is proposed based on 2D variable density under-sampling (VRDU) scheme. The 2D-VRDU scheme improves the result by sampling the high energy central part of the k-space slices. The novel interpolation technique takes two consecutive slices and estimates the missing samples of the target slice (T slice) from its left slice (L slice). Compared to the previous methods, slices recovered with the proposed FiCS technique have a maximum correlation with their corresponding original slices. The proposed FiCS technique is evaluated by using both subjective and objective assessment. In subjective assessment, our proposed technique shows less partial volume loss compared to existing techniques. For objective assessment different performance metrics, such as structural similarity index measurement (SSIM), peak signal to noise ratio (PSNR), mean square error (MSE) and correlation, are used and compared with existing interpolation techniques. Simulation results on knee and brain dataset shows that the proposed FiCS technique has improved image quality and performance with even reduced scan time, lower computational complexity and maximum information content.

Radial Undersampling-Based Interpolation Scheme for Multislice CSMRI Reconstruction Techniques

“…FiCS and CS, the proposed radial undersampling strategy of EiCS equally undersamples the k-space multislice MRI sequence but using a much lower undersampling ratio. Table 1 shows a comparison of the proposed radial undersampling schemes with the 2D-VRDU undersampling scheme of FiCS [38], 1D-VRDU scheme of iCS [32], and CS [10]. The assessment has been performed using all the seven assessment parameters for three successive slices and averaged.…”

“…The concept of iCS has been introduced by Pang et al [24,25] in 2012. His interpolation technique has later been explored by Datta and Deka [31,32], but their undersampling approaches do not produce clinically acceptable results by causing information loss in most (67%) of their reconstructed slices [38]. Secondly, their interpolation techniques are computationally inefficient with redundant Fou-rier steps.…”

“…In a recent work [38], the authors proposed a new and fast interpolation technique (FiCS) based on a 2D-VRDU sampling scheme. Their outcome shows more clinically acceptable results with less partial volume loss and lower average sampling ratio and by using a computational efficient interpolation technique.…”

“…The proposed EiCS technique is evaluated using all the seven assessment parameters and compared with recent interpolation techniques [32,38] and CS [10]. The proposed technique is also evaluated on whole datasets for different undersampling ratios and compared.…”

Multi-Layer Basis Pursuit for Compressed Sensing MR Image Reconstruction

“…A test set size of 5% is used to validate the performance of the testing framework in both data sets for analyzing PSNR and SSIM on reconstructed images. The third dataset comprising brain images is of the normal aging coronal plane with 123 slices, matrix size (256 × 256), and is publicly available on AANLIB database of Harvard medical school at http://www.med.harvard.edu/AANLIB/home.html [49] used by M.Murad et al in [50]. We have also used a single image consisting of a single-slice (axial T2-weighted reference brain image) dataset of size (256x256), vivo MR scans from American Radiology Services as used by Parasad et al in [10].…”