Artificial neural network for suppression of banding artifacts in balanced steady-state free precession MRI

The balanced steady-state free precession (bSSFP) MR sequence is frequently used in clinics, but is sensitive to off-resonance effects, which can cause banding artifacts. Often multiple bSSFP datasets are acquired at different phase cycling (PC) angles and then combined in a special way for banding artifact suppression. Many strategies of combining the datasets have been suggested for banding artifact suppression, but there are still limitations in their performance, especially when the number of phase-cycled bSSFP datasets is small. The purpose of this study is to develop a learning-based model to combine the multiple phase-cycled bSSFP datasets for better banding artifact suppression. Multilayer perceptron (MLP) is a feedforward artificial neural network consisting of three layers of input, hidden, and output layers. MLP models were trained by input bSSFP datasets acquired from human brain and knee at 3T, which were separately performed for two and four PC angles. Banding-free bSSFP images were generated by maximum-intensity projection (MIP) of 8 or 12 phase-cycled datasets and were used as targets for training the output layer. The trained MLP models were applied to another brain and knee datasets acquired with different scan parameters and also to multiple phase-cycled bSSFP functional MRI datasets acquired on rat brain at 9.4T, in comparison with the conventional MIP method. Simulations were also performed to validate the MLP approach. Both the simulations and human experiments demonstrated that MLP suppressed banding artifacts significantly, superior to MIP in both banding artifact suppression and SNR efficiency. MLP demonstrated superior performance over MIP for the 9.4T fMRI data as well, which was not used for training the models, while visually preserving the fMRI maps very well. Artificial neural network is a promising technique for combining multiple phase-cycled bSSFP datasets for banding artifact suppression.     

Evaluating gradient artifact correction of EEG data acquired simultaneously with fMRI

Simultaneous electroencephalography (EEG) and functional magnetic resonance imaging (fMRI) has become a widely used application in spite of EEG perturbations due to electromagnetic interference in the MR environment. The most prominent and disturbing artifacts in the EEG are caused by the alternating magnetic fields (gradients) of the MR scanner. Different methods for gradient artifact correction have been developed. Here we propose an approach for the systematic evaluation and comparison of these gradient artifact correction methods. Exemplarily, we evaluate different algorithms all based on artifact template subtraction--the currently most established means of gradient artifact removal. We introduce indices for the degree of gradient artifact reduction and physiological signal preservation. The combination of both indices was used as a measure for the overall performance of gradient artifact removal and was shown to be useful in identifying problems during artifact removal. We demonstrate that the evaluation as proposed here allows to reveal frequency-band specific performance differences among the algorithms. This emphasizes the importance of carefully selecting the artifact correction method appropriate for the respective case.     

Fat-water selective excitation in balanced steady-state free precession using short spatial-spectral RF pulses

Fat suppression is important but challenging in balanced steady-state free precession (bSSFP) acquisitions, for a number of clinical applications. In the present work, the practicality of performing fat-water selective excitations using spatial-spectral (SPSP) RF pulses in bSSFP sequence is examined. With careful pulse design, the overall duration of these SPSP pulses was kept short to minimize detrimental effects on TR, scan time and banding artifact content. Fat-water selective excitation using SPSP pulses was demonstrated in both phantom and human bSSFP imaging at 3T, and compared to results obtained using a two-point Dixon method. The sequence with SPSP pulses performed better than the two-point Dixon method, in terms of scan time and suppression performance. Overall, it is concluded here that SPSP RF pulses do represent a viable option for fat-suppressed bSSFP imaging.     

Balanced steady state free precession for arterial spin labeling MRI: Initial experience for blood flow mapping in human brain,retina, and kidney

We implemented pseudo-continuous ASL (pCASL) with 2D and 3D balanced steady state free precession (bSSFP) readout for mapping blood flow in the human brain, retina, and kidney, free of distortion and signal dropout, which are typically observed in the most commonly used echo-planar imaging acquisition. High resolution functional brain imaging in the human visual cortex was feasible with 3D bSSFP pCASL. Blood flow of the human retina could be imaged with pCASL and bSSFP in conjunction with a phase cycling approach to suppress the banding artifacts associated with bSSFP. Furthermore, bSSFP based pCASL enabled us to map renal blood flow within a single breath hold. Control and test–retest experiments suggested that the measured blood flow values in retina and kidney were reliable. Because there is no specific imaging tool for mapping human retina blood flow and the standard contrast agent technique for mapping renal blood flow can cause problems for patients with kidney dysfunction, bSSFP based pCASL may provide a useful tool for the diagnosis of retinal and renal diseases and can complement existing imaging techniques.     

Susceptibility artifacts from metallic markers and cardiac catheterization devices on a high-performance 0.55 T MRI system

IntroductionVisualization of passive devices during MRI-guided catheterizations often relies on a susceptibility artifact from the device itself or added susceptibility markers that impart a unique imaging signature. High-performance low field MRI systems offer reduced RF-induced heating of metallic devices during MRI-guided invasive procedures, but susceptibility artifacts are expected to diminish with field strength, reducing device visualization. In this study, field strength and orientation dependence of artifacts from susceptibility markers and metallic guidewires were evaluated using a prototype high-performance 0.55 T MRI system.Materials and methodsArtifact volume from nitinol and stainless steel passive susceptibility markers was quantified using histogram analysis of pixel intensities from three-dimensional gradient echo images at 0.55 T, 1.5 T and 3 T. In addition, visibility of commercially available clinical catheterization devices was compared between 0.55 T and 1.5 T using real-time bSSFP in phantoms and in vivo.ResultsA low-tensile strength stainless-steel marker produced field strength- and orientation-dependent artifact size (1.7 cm³, 1.95 cm³, 2.21 cm³ at 0.55 T, 1.5 T, 3 T, respectively). Whereas, a high-tensile strength steel marker, of the same alloy, produced field strength- and orientation-independent artifact size (3.35 cm³, 3.41 cm³, 3.42 cm³ at 0.55 T, 1.5 T, 3 T, respectively). Visibility of commercially available nitinol guidewires was reduced at 0.55 T, but imaging signature could be maintained using high-susceptibility stainless steel markers.Discussion and conclusionHigh-susceptibility stainless-steel markers generate field-independent artifacts between 0.55 T, 1.5 T and 3 T, indicating magnetic saturation at fields <0.55 T. Thus, artifact size can be tailored such that interventional devices produce identical imaging signatures across field strengths.     

An unsupervised deep learning technique for susceptibility artifact correction in reversed phase-encoding EPI images

Echo planar imaging (EPI) is a fast and non-invasive magnetic resonance imaging technique that supports data acquisition at high spatial and temporal resolutions. However, susceptibility artifacts, which cause the misalignment to the underlying structural image, are unavoidable distortions in EPI. Traditional susceptibility artifact correction (SAC) methods estimate the displacement field by optimizing an objective function that involves one or more pairs of reversed phase-encoding (PE) images. The estimated displacement field is then used to unwarp the distorted images and produce the corrected images. Since this conventional approach is time-consuming, we propose an end-to-end deep learning technique, named S-Net, to correct the susceptibility artifacts the reversed-PE image pair. The proposed S-Net consists of two components: (i) a convolutional neural network to map a reversed-PE image pair to the displacement field; and (ii) a spatial transform unit to unwarp the input images and produce the corrected images. The S-Net is trained using a set of reversed-PE image pairs and an unsupervised loss function, without ground-truth data. For a new image pair of reversed-PE images, the displacement field and corrected images are obtained simultaneously by evaluating the trained S-Net directly. Evaluations on three different datasets demonstrate that S-Net can correct the susceptibility artifacts in the reversed-PE images. Compared with two state-of-the-art SAC methods (TOPUP and TISAC), the proposed S-Net runs significantly faster: 20 times faster than TISAC and 369 times faster than TOPUP, while achieving a similar correction accuracy. Consequently, S-Net accelerates the medical image processing pipelines and makes the real-time correction for MRI scanners feasible. Our proposed technique also opens up a new direction in learning-based SAC.     

Correction of out-of-FOV motion artifacts using convolutional neural network

PurposeSubject motion during MRI scan can result in severe degradation of image quality. Existing motion correction algorithms rely on the assumption that no information is missing during motions. However, this assumption does not hold when out-of-FOV motion happens. Currently available algorithms are not able to correct for image artifacts introduced by out-of-FOV motion. The purpose of this study is to demonstrate the feasibility of incorporating convolutional neural network (CNN) derived prior image into solving the out-of-FOV motion problem.Methods and materialsA modified U-net network was proposed to correct out-of-FOV motion artifacts by incorporating motion parameters into the loss function. A motion model based data fidelity term was applied in combination with the CNN prediction to further improve the motion correction performance. We trained the CNN on 1113 MPRAGE images with simulated oscillating and sudden motion trajectories, and compared our algorithm to a gradient-based autofocusing (AF) algorithm in both 2D and 3D images. Additional experiment was performed to demonstrate the feasibility of transferring the networks to different dataset. We also evaluated the robustness of this algorithm by adding Gaussian noise to the motion parameters. The motion correction performance was evaluated using mean square error (NMSE), peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM).ResultsThe proposed algorithm outperformed AF-based algorithm for both 2D (NMSE: 0.0066 ± 0.0009 vs 0.0141 ± 0.008, P < .01; PSNR: 29.60 ± 0.74 vs 21.71 ± 0.27, P < .01; SSIM: 0.89 ± 0.014 vs 0.73 ± 0.004, P < .01) and 3D imaging (NMSE: 0.0067 ± 0.0008 vs 0.070 ± 0.021, P < .01; PSNR: 32.40 ± 1.63 vs 22.32 ± 2.378, P < .01; SSIM: 0.89 ± 0.01 vs 0.62 ± 0.03, P < .01). Robust reconstruction was achieved with 20% data missed due to the out-of-FOV motion.ConclusionIn conclusion, the proposed CNN-based motion correction algorithm can significantly reduce out-of-FOV motion artifacts and achieve better image quality compared to AF-based algorithm.     

Combined modified-Dixon and PROPELLER method with low refocusing flip angle for contrast-enhanced fat-suppressed T1-weighted MRI: A prospective cross-sectional study

PurposeTo propose the combined modified-Dixon and PROPELLER sequence with low refocusing flip angle (RFA) and investigate whether this sequence can acquire clinical contrast-enhanced (CE), fat-suppressed T1-weighted (T1W) images of the head and neck.MethodsThe optimal RFA for T1W imaging was investigated in the brain of a healthy volunteer. The motion artifacts, water–fat separation error, contrast ratio (CR), and comprehensive quality were evaluated through comparison with a standard Cartesian modified-Dixon sequence in 50 patients. Two radiologists independently scored motion artifacts and water–fat separation error using a 4-point scale (1, unacceptable; 4, excellent) and comprehensive quality using a 5-point scale (1, substantially inferior; 5, substantially superior). The CR between CE lesions and non-CE muscle was calculated.ResultsThe optimal RFA of 40° was determined. In the motion artifact assessment, ratings of 3 or 4 points were assigned to 83% (observer-1, 42/50; observer-2, 41/50) and 99% (50/50; 49/50) of cases for the standard and proposed sequences, respectively (p < 0.001; p < 0.001). For the water–fat separation error assessment, ratings of 3 or 4 points were assigned to 100% (50/50; 50/50) and 97% (48/50; 49/50) of cases, respectively (p < 0.001; p = 0.02). In comprehensive evaluation, the proposed sequence was equal, slightly superior, or substantially superior to the standard sequence in 85% (39/50; 46/50). The CR was significantly higher with the proposed sequence [2.27 (1.99–2.97) vs. 2.08 (1.88–2.42), p < 0.001].ConclusionThe proposed sequence acquired stable fat-suppressed CE T1W images without motion artifacts and yielded superior overall image quality compared with the standard sequence.     

Artifact-free black-blood cine cardiac imaging in a single breath-hold

Fast imaging using the STimulated Echo Acquisition Mode (STEAM) sequence can produce cine images of the heart with black-blood contrast. Nevertheless, correction of deformation-related artifacts is required in order to maintain myocardial signal throughout the cardiac cycle. Recent work by our group has eliminated this artifact by combining two STEAM sequences acquired with two different demodulation gradients. Unfortunately, these two STEAM sequences were acquired on two separate breath-holds; thus, scan time doubled. In this work, we present a technique to reduce the total scan time by one half, without sacrificing image quality. The technique is based on interleaving two demodulations within one acquisition in order to obtain quality cine images of the heart in a single breath-hold. The technique was tested on animal models and human subjects, and the impact of interleaved acquisition on image quality was studied using quantitative and qualitative measures.     

Feasibility of fat-saturated T2-weighted magnetic resonance imaging with slice encoding for metal artifact correction (SEMAC) at 3T

Background and Purpose

Fluid-sensitive MR imaging in postoperative evaluation is important, however, metallic artifacts is inevitable. The purpose is to investigate the feasibility of fat-saturated slice encoding for metal artifact correction (SEMAC)-corrected T2-weighted magnetic resonance (MR) at 3T in patients with spinal prostheses.

Methods

Following institutional review board approval, 27 SEMAC-encoded spinal MRs between September 2012 and October 2013 in patients with spinal metallic prostheses were analyzed. The MR images were scanned on a 3T MR system including SEMAC-corrected and uncorrected fast spin echo (FSE) T2-weighted MR images with fat-saturation. Two musculoskeletal radiologists compared the image sets and qualitatively analyzed the images using a five-point scale in terms of artifact reduction around the prosthesis, visualization of the prosthesis and pedicle, and intervertebral neural foramina. Quantitative assessments were performed by calculating the ratio of signal intensity from the fixated vertebra and that from upper level vertebra. For statistical analyses, paired t-test was used.

Results

Fat-saturated SEMAC-corrected T2-weighted MR images enabled significantly improved metallic artifact reduction (P < 0.05). Quantitative evaluation of the signal intensity ratio of screw-fixated vertebra and upper level vertebra showed a significantly lower ratio on fat-saturated SEMAC images (P < 0.05), however, the high signal intensity of signal pile-up could be not completely corrected.

Conclusion

SEMAC correction in fat-suppressed T2-weighted MR images can overcome the signal loss of metallic artifacts and provide improved delineation of the pedicle screw and peri-prosthetic region. Signal pile-up, however, could not be corrected completely, therefore readers should be cautious in the evaluation of marrow around the prosthesis.     

Large angle spin-echo imaging

A study was undertaken to assess the use of excitation flip angles greater than 90° for T₁ weighted spin-echo (SE) imaging with a single 180° refocusing pulse and short TR values. Theoretical predictions of signal intensity for SE images with excitation pulse angles of 90–180° were calculated based on the Bloch equations and then measured experimentally from MR images of MnCl₂ phantoms of various concentrations. Liver signal-to-noise ratios (SNR) and liver-spleen contrast-to-noise ratios (CNR) were measured from breathhold MR images of the upper abdomen in 16 patients using 90 and 110° excitation flip angles. The theoretical predictions showed significant improvements in SNR with excitation flip angles >90°, which were more pronounced at small TR values. The phantom studies showed reasonably good agreement with the theoretical predictions in correlating the excitation pulse angle with signal intensity. In the human imaging studies, the 110° excitation pulse angle resulted in a 7.4% (p < .01) increase in liver SNR and an 8.2% (p = .2) increase in liver-spleen CNR compared to the 90° pulse angle at TR = 275 ms. Increased signal intensity resulting from the use of large flip angle excitation pulses with a single echo SE pulse sequence was predicted and confirmed experimentally in phantoms and humans.     

An optimized pulse sequence for isotropically weighted diffusion imaging.

Single-shot echo-planar imaging is becoming the most widely used technique for magnetic resonance diffusion imaging, since it enables measurement of diffusion coefficients in human brain without motion artifacts. However, its reliability is limited by geometrical distortions due to eddy currents. In this report, an isotropically weighted echo-planar pulse sequence, optimized to give the maximum signal-to-noise ratio in the computed trace image and designed to produce inherently low distortions, is presented. It is also shown how the residual translational distortion can be easily characterized and removed by postprocessing. A full characterization of the distortion artifact involves a few measurements on a phantom, in order to estimate the distortion as a function of slice orientation, which can then be used to correct any slice orientation. Results of applying the image translation correction to data collected from a patient are presented.     

Balanced steady-state free precession with parallel imaging gives distortion-free fMRI with high temporal resolution

Research on the functions of the human brain requires that functional magnetic resonance imaging (MRI) moves towards producing images with less distortion and higher temporal and spatial resolution. This study compares passband balanced steady-state free precession (bSSFP) acquisitions with and without parallel imaging (PI) to investigate whether combining PI with this pulse sequence is a viable option for functional MRI. Such a novel combination has the potential to offer the distortion-free advantages of bSSFP with the reduced acquisition time of PI. Scans were done on a Philips 3T Intera, using the installed bSSFP pulse sequence, both with and without the sensitivity encoding (SENSE) PI option. The task was a visual flashing checkerboard, and the viewing window covered the visual cortex. Sensitivity comparisons with and without PI were done using the same manually drawn region of interest for each time course of the subject, and comparing the z-score summary statistics: number of voxels with z>2.3, the mean of those voxels, their 90th percentile and their maximum value. We show that PI greatly improves the temporal resolution in bSSFP, reducing the volume acquisition time by more than half in this study to 0.67 s with 3-mm isotropic voxels. At the same time, a statistically significant increase was found for the maximum z-score using bSSFP with PI as compared to without it (P=.02). This improvement can be understood in terms of physiological noise, as demonstrated by noise measurements. This produces observed increases in the overall temporal signal to noise of the functional time series, giving greater sensitivity to functional activations with PI. This study demonstrates for the first time the possibility of combining PI with bSSFP to achieve distortion-free functional images without loss of sensitivity and with high temporal resolution.     

Artifact removal in Fourier-domain optical coherence tomography with a piezoelectric fiber stretcher

We describe an artifact removal setup swept-source optical coherence tomography (OCT) system that enables high-speed full-range imaging. We implement a piezoelectric fiber stretcher to generate a periodic phase shift between successive A-scans, thus introducing a transverse modulation. The depth ambiguity is then resolved by performing a Fourier filtering in the transverse direction before processing the data in the axial direction. The dc artifact is also removed. The key factor is that the piezoelectric fiber stretcher can be used to generate discrete phase shifts with a high repetition rate. The proposed experimental setup is a much improved version of the previously reported B-M mode scanning for spectral-domain OCT in that it does not generate additional artifacts. It is a simple and low-cost solution for artifact removal that can easily be applied.     

Determination of the Rydberg D and F° energy levels of Cs-like ions by the method of interpolation of relativistic quantum defects

Data are presented that allow one to determine the energy values of the Rydberg levels nD _3/2, nD _5/2, nF _5/2°, and nF _7/2° for ions of the cesium isoelectronic sequence by the method of interpolation of rela- tivistic quantum defects in the Dirac-Fock approximation. To improve agreement with the experimental data, it is proposed to use an empirical correction for the quantum defect. The obtained energy values are compared with the experimental data. The anomalous behavior of the quantum defects of the nf levels is discussed. The calculations are carried out for ten ions of the cesium isoelectronic sequence.     

Single-shot line scan imaging using stimulated echoes

A new high-speed MRI method is described for single-shot line scan imaging (LSI) based on stimulated echoes (STE). To allow for multislice imaging, the technique comprises a series of slice-selective preparation pulses (each corresponding to the first RF pulse of a STE sequence), a slab-selective refocusing pulse (second RF pulse), and multiple line-selective read pulses (third RF pulses). An alternative version employs packages of two slice-selective pulses followed by multiple line-selective read pulses. Experimental applications deal with human brain imaging on a clinical MRI system at 2.0 T. The technique offers user-selectable trade-offs between volume coverage (1-15 sections) and in-plane spatial resolution (1-5 mm linear pixel dimension) within total acquisition times of less than 500 ms. Although LSI yields a lower signal-to-noise ratio than Fourier imaging, single-shot LSI with STEs is free from resonance offset effects (e.g., magnetic field inhomogeneities and susceptibility differences) that are typical for echo-planar imaging. Moreover, the technique exhibits considerable robustness against motion and provides access to arbitrary fields-of-view, i.e., localized imaging of inner volumes without aliasing artifacts due to phase wrapping.     

CSF flow artifact reduction using cardiac cycle ordered phase-encoding method

The flow effects appear as a change of phase as well as signal intensity in NMR imaging. Since the flow of blood and CSF (cerebrospinal fluid) is pulsatile due to the heart pumping, their velocities are not constant during NMR imaging. This type of velocity fluctuation such as the blood or CSF flow induces irregular flow-dependent phase shifts, which have been one of the main causes of flow artifacts in NMR imaging. In order to reduce the flow artifacts, especially the CSF flow artifacts, a new cardiac cycle ordered phase encoding method is proposed and has been studied. This proposed technique utilizes the cardiac cycle as a precursor for the phase encoding gradients similar to the ROPE (respiratory ordered phase encoding) technique which has been used for respiratory motion artifact reduction. The basic concept and its applications are discussed together with the experimental results obtained with human volunteers using the KAIS 2.0 2.0 T whole-body NMR imaging system.     

Parallel EPI artifact correction (PEAC) for N/2 ghost suppression in neuroimaging applications

Echo Planar Imaging (EPI) is a neuroimaging tool for clinical practice and research investigation. Due to odd-even echo phase inconsistencies, however, EPI suffers from Nyquist N/2 ghost artifacts. In standard neuroimaging protocols, EPI artifacts are suppressed using phase correction techniques that require reference data collected from a reference scan. Because reference-scan based techniques are sensitive to subject motion, EPI performance is sub-optimal in neuroimaging applications. In this technical note, we present a novel EPI data processing technique which we call Parallel EPI Artifact Correction (PEAC). By introducing an implicit data constraint associated with multi-coil sensitivity in parallel imaging, PEAC converts phase correction into a constrained problem that can be resolved using an iterative algorithm. This enables “reference-less” EPI that can improve neuroimaging performance. In the presented work, PEAC is investigated using a standard functional magnetic resonance imaging (fMRI) protocol with multi-slice 2D EPI. It is demonstrated that PEAC can suppress ghost artifacts as effectively as the standard reference-scan based phase correction technique used on a clinical MRI system. We also found that PEAC can achieve dynamic phase correction when motion occurs.     

Determination of the <Emphasis Type="Italic">D</Emphasis> and <Emphasis Type="Italic">F</Emphasis>° Rydberg energy levels of rubidium-like ions by the method of interpolation of relativistic quantum defects

Data are presented for determination of the nD _3/2, nD _5/2, nF°_5/2, and nF°_7/2 Rydberg energy levels for ions of the Rb isoelectronic sequence in the Dirac-Fock approximation by the method of interpolation of relativistic quantum defects. For better agreement with experiment, it is proposed to use an empirical correlation correction to the quantum defect. The energy values obtained are compared to experimental data. The anomalous behavior of quantum defects for nf levels is discussed. The calculations are performed for 17 members of the Rb isoelectronic sequence (from Rb to Fr⁵⁰⁺).     

Phase cycling in double-quantum-filtered COSY

High repetition rates generate artifacts in double-quantum-filtered COSY experiments, because incomplete relaxation causes a modulation of the detected signal; the resultant anomalies appear at integral multiples of the resonance offset in the F₁ dimension of the final two-dimensional spectrum. Product-operator calculations are used to predict rearrangements of the basic DQCOSY phase cycle in which these artifacts are canceled. The success of these phase cycles is tested experimentally. It is shown that a phase cycle of only four steps is necessary to provide repetition-rate artifact suppression but that it becomes easier to suppress artifacts as the length of the cycle increases, provided that attention is paid to the interaction between various types of phase cycling.