Search

Purpose: To develop and evaluate an accelerated 3D self‐gated cardiac cine imaging technique at 3 Tesla without the use of external electrocardiogram triggering or respiratory gating. Methods: A 3D stack‐of‐stars balanced steady‐state free precession sequence with a tiny golden angle sampling scheme was developed to reduced eddy current effect‐related artefacts at 3 Tesla. Respiratory and cardiac motion were derived from a central 5‐point self‐gating signal extraction approach. The data acquired around the end‐expiration phases were then sorted into individual cardiac bins and used for reconstruction with compressed sensing. To evaluate the performance of the proposed method, image quality (1: the best; 4: the worst) was quantitatively compared using both the proposed method and the conventional 3D golden‐angle self‐gated method. Linear regression and Bland‐Altman analysis were used to assess the functional measurements agreement between the proposed method and the routine 2D breath‐hold multi‐slice technique. Results: Compared to the conventional 3D golden‐angle self‐gated method, the proposed method yielded images with much less streaking artifact and higher myocardium edge sharpness (0.50 ± 0.06 vs. 0.45 ± 0.05, P = 0.004). The proposed method provided an inferior image quality score to the routine 2D technique (2.13 ± 0.35 vs. 1.38 ± 0.52, P = 0.063) but a superior one to the conventional self‐gated method (2.13 ± 0.35 vs. 3.13 ± 0.64, P = 0.031). Left ventricular functional measurements between the proposed method and routine 2D technique were all well in agreement. Conclusion: This study presents a novel self‐gating approach to realize rapid 3D cardiac cine imaging at 3 Tesla. [ABSTRACT FROM AUTHOR]

The high computational complexity of tree-based multipath search approaches makes putting them into practical use difficult. However, reselection of candidate atoms could make the search path more accurate and efficient. We propose a multipath greedy approach called fast sparsity adaptive multipath matching pursuit (fast SAMMP), which performs a sparsity adaptive tree search to find the sparsest solution with better performances. Each tree branch acquires K atoms, and fast SAMMP reselects the best K atoms among 2K atoms. Fast SAMMP adopts sparsity adaptive techniques that allow more practical applications for the algorithm. We demonstrated the reconstruction performances of the proposed fast scheme on both synthetically generated one-dimensional signals and two-dimensional images using Gaussian observation matrices. The experimental results indicate that fast SAMMP achieves less reconstruction time and a much higher exact recovery ratio compared with conventional algorithms. [ABSTRACT FROM AUTHOR]

ABSTRACT In the magnetic resonance imaging (MRI) field, total variation (TV) which is the [ABSTRACT FROM AUTHOR]

Magnetic Resonance Imaging (MRI) is a leading diagnostic imaging modality that supports high contrast of soft tissues with no invasiveness or radiation. Nonetheless, it suffers from long scan time owing to the inherent physics in its data acquisition process, hampering its development and applications. Traditional strategies such as Compressed Sensing (CS) and Parallel Imaging (PI) allow for MRI acceleration via sub-sampling strategy, and multiple coils, respectively. When Deep Learning (DL) joins in, both strategies get re-vitalized to achieve even faster reconstruction in various reconstruction methods, among which the variational network is a previously proposed method that combines the mathematical structure of variational models with DL for fast MRI reconstruction. However, in our study we observe that the information of MR features is either not efficiently or explicitly exploited in former works based on the variational network. Instead, we introduce a variational network with explicit feature fusion that combines the CS, PI, with DL for accelerated multi-coil MRI reconstruction. By explicitly leveraging the extra information via feature fusion following feature extraction, our proposed method achieves comparably satisfying performance to the state-of-the-art methods without too much computation overhead on a public multi-coil brain dataset under 5-fold and 10-fold acceleration. [ABSTRACT FROM AUTHOR]

Magnetic resonance imaging produces detailed anatomical and physiological images of the human body that can be used in the clinical diagnosis and treatment of diseases. However, MRI suffers its comparatively longer acquisition time than other imaging methods and is thus vulnerable to motion artifacts, which ultimately lead to likely failed or even wrong diagnosis. In order to perform faster reconstruction, deep learning-based methods along with traditional strategies such as parallel imaging and compressed sensing come into play in recent years in this field. Meanwhile, in order to better analyze the diseases, it is also often necessary to acquire images in the same region of interest under different modalities, which yield images with different contrast levels. However, most of these aforementioned methods tend to use single-modal images for reconstruction, neglecting the correlation and redundancy information embedded in MR images acquired with different modalities. While there are works on multi-modal reconstruction, the information is yet to be efficiently explored. In this paper, we propose an end-to-end neural network called MLMFNet, which helps the reconstruction of the target modality by using information from the auxiliary modality across feature channels and layers. Specifically, this is highlighted by three components: (I) An encoder based on UNet with a single-stream strategy that fuses auxiliary and target modalities; (II) a decoder that tends to multi-level features from all layers of the encoder, and (III) a channel attention module. Quantitative and qualitative analyses are performed on a public brain dataset and knee brain dataset, which show that the proposed method achieves satisfying results in MRI reconstruction within the multi-modal context, and also demonstrate its effectiveness and potential to be used in clinical practice. [ABSTRACT FROM AUTHOR]

High spatiotemporal resolution MRI is a challenging topic in dynamic MRI field. Partial separability (PS) model has been successfully applied to dynamic cardiac MRI by exploiting data redundancy. However, the model requires substantial preprocessing data to accurately estimate the model parameters before image reconstruction. Since compressed sensing (CS) is a potential technique to accelerate MRI by reducing the number of acquired data, the combination of PS and CS, named as Stepped-SparsePS, was introduced to accelerate the preprocessing data acquisition of PS in this work. The proposed Stepped-SparsePS method sequentially reconstructs a set of aliased dynamic images in each channel based on PS model and then the final dynamic images from the aliased images using CS. The results from numerical simulations and in vivo experiments demonstrate that Stepped-SparsePS could significantly reduce data acquisition time while preserving high spatiotemporal resolution. [ABSTRACT FROM AUTHOR]

Abstract: Joint estimation of coil sensitivities and output image (JSENSE) is a promising approach that improves the reconstruction of parallel magnetic resonance imaging (pMRI). However, when acceleration factor increases, the signal to noise ratio (SNR) of JSENSE reconstruction decreases as quickly as that of the conventional pMRI. Although sparse constraints have been used to improve the JSENSE reconstruction in recent years, these constraints only use the sparsity of the output image, which cannot fully exploit the prior information of pMRI. In this paper, we use the sparsity of coil images, instead of the output image, to exploit more prior information for JSENSE. Numerical simulation, phantom and in vivo experiments demonstrate that the proposed method has better performance than the SparseSENSE method and the constrained JSENSE method using the sparsity of the output image only. [Copyright &y& Elsevier]