In vivo diffusion MRI of the human heart using a 300 mT/m gradient system

PURPOSE

This work reports for the first time on the implementation and application of cardiac diffusion-weighted MRI on a Connectom MR scanner with a maximum gradient strength of 300 mT/m. It evaluates the benefits of the increased gradient performance for the investigation of the myocardial microstructure.

METHODS

Cardiac diffusion-weighted imaging (DWI) experiments were performed on 10 healthy volunteers using a spin-echo sequence with up to second- and third-order motion compensation (M 2 $$ {M}_2 $$ andM 3 $$ {M}_3 $$ ) andb = 100 ,   450 $$ b=100,450 $$ , and 1000s / m m 2 $$ \mathrm{s}/\mathrm{m}{\mathrm{m}}^2 $$ (twice theb max $$ {b}_{\mathrm{max}} $$ commonly used on clinical scanners). Mean diffusivity (MD), fractional anisotropy (FA), helix angle (HA), and secondary eigenvector angle (E2A) were calculated for b = [100, 450]s / m m 2 $$ \mathrm{s}/\mathrm{m}{\mathrm{m}}^2 $$ and b = [100, 1000]s / m m 2 $$ \mathrm{s}/\mathrm{m}{\mathrm{m}}^2 $$ for bothM 2 $$ {M}_2 $$ andM 3 $$ {M}_3 $$ .

RESULTS

The MD values withM 3 $$ {M}_3 $$ are slightly higher than withM 2 $$ {M}_2 $$ withΔ MD = 0 . 05 ± 0 . 05 [ × 1 0 - 3 mm 2 / s ] ( p = 4 e - 5 ) $$ \Delta \mathrm{MD}=0.05\pm 0.05\kern0.3em \left[\times 1{0}^{-3}\kern0.3em {\mathrm{mm}}^2/\mathrm{s}\right]\kern0.3em \left(p=4e-5\right) $$ forb max = 450 s / mm 2 $$ {b}_{\mathrm{max}}=450\kern0.3em \mathrm{s}/{\mathrm{mm}}^2 $$ andΔ MD = 0 . 03 ± 0 . 03 [ × 1 0 - 3 mm 2 / s ] ( p = 4 e - 4 ) $$ \Delta \mathrm{MD}=0.03\pm 0.03\kern0.3em \left[\times \kern0.3em 1{0}^{-3}\kern0.3em {\mathrm{mm}}^2/\mathrm{s}\right]\kern0.3em \left(p=4e-4\right) $$ forb max = 1000 s / mm 2 $$ {b}_{\mathrm{max}}=1000\kern0.3em \mathrm{s}/{\mathrm{mm}}^2 $$ . A reduction in MD is observed by increasing theb max $$ {b}_{\mathrm{max}} $$ from 450 to 1000s / mm 2 $$ \mathrm{s}/{\mathrm{mm}}^2 $$ (Δ MD = 0 . 06 ± 0 . 04 [ × 1 0 - 3 mm 2 / s ] ( p = 1 . 6 e - 9 ) $$ \Delta \mathrm{MD}=0.06\pm 0.04\kern0.3em \left[\times \kern0.3em 1{0}^{-3}\kern0.3em {\mathrm{mm}}^2/\mathrm{s}\right]\kern0.3em \left(p=1.6e-9\right) $$ forM 2 $$ {M}_2 $$ andΔ MD = 0 . 08 ± 0 . 05 [ × 1 0 - 3 mm 2 / s ] ( p = 1 e - 9 ) $$ \Delta \mathrm{MD}=0.08\pm 0.05\kern0.3em \left[\times \kern0.3em 1{0}^{-3}\kern0.3em {\mathrm{mm}}^2/\mathrm{s}\right]\kern0.3em \left(p=1e-9\right) $$ forM 3 $$ {M}_3 $$ ). The difference between FA, E2A, and HA was not significant in different schemes (p > 0 . 05 $$ p>0.05 $$ ).

CONCLUSION

This work demonstrates cardiac DWI in vivo with higher b-value and higher order of motion compensated diffusion gradient waveforms than is commonly used. Increasing the motion compensation order fromM 2 $$ {M}_2 $$ toM 3 $$ {M}_3 $$ and the maximum b-value from 450 to 1000  s / mm 2 $$ \mathrm{s}/{\mathrm{mm}}^2 $$ affected the MD values but FA and the angular metrics (HA and E2A) remained unchanged. Our work paves the way for cardiac DWI on the next-generation MR scanners with high-performance gradient systems.

Artificial intelligence for neuro MRI acquisition: a review

OBJECT

To review recent advances of artificial intelligence (AI) in enhancing the efficiency and throughput of the MRI acquisition workflow in neuroimaging, including planning, sequence design, and correction of acquisition artifacts.

MATERIALS AND METHODS

A comprehensive analysis was conducted on recent AI-based methods in neuro MRI acquisition. The study focused on key technological advances, their impact on clinical practice, and potential risks associated with these methods.

RESULTS

The findings indicate that AI-based algorithms have a substantial positive impact on the MRI acquisition process, improving both efficiency and throughput. Specific algorithms were identified as particularly effective in optimizing acquisition steps, with reported improvements in workflow efficiency.

DISCUSSION

The review highlights the transformative potential of AI in neuro MRI acquisition, emphasizing the technological advances and clinical benefits. However, it also discusses potential risks and challenges, suggesting areas for future research to mitigate these concerns and further enhance AI integration in MRI acquisition.

Clipped DeepControl: Deep neural network two-dimensional pulse design with an amplitude constraint layer

Advanced radio-frequency pulse design used in magnetic resonance imaging has recently been demonstrated with deep learning of (convolutional) neural networks and reinforcement learning. For two-dimensionally selective radio-frequency pulses, the (convolutional) neural network pulse prediction time (a few milliseconds) was in comparison more than three orders of magnitude faster than the conventional optimal control computation. The network pulses were from the supervised training capable of compensating scan-subject dependent inhomogeneities of B₀ and B₁⁺ fields. Unfortunately, the network presented with a small percentage of pulse amplitude overshoots in the test subset, despite the optimal control pulses used in training were fully constrained. Here, we have extended the convolutional neural network with a custom-made clipping layer that completely eliminates the risk of pulse amplitude overshoots, while preserving the ability to compensate for the inhomogeneous field conditions.

Analysing the effects of metallic biomaterial design and imaging sequences on MRI interpretation challenges due to image artefacts

Biometals cause signal loss and susceptibility artefacts in the surrounding tissue, resulting in deterioration in magnetic resonance (MR) images. This metal-artefact effect may lead to interpretation challenges for MR images. Therefore, artefact reduction is required to obtain higher-quality images. This paper aims to analyse the impact of imaging sequence and metallic biomaterial design on MR image artefacts. In this respect, implant specimens were designed in thin, thick, and pointed forms and manufactured using 316LVM, 316L, CoCr-alloy, and Ti-alloy, which are commonly utilized materials in the biomaterials field. Specimens were placed in a phantom that simulates average human anatomy separately and scanned in a 1.5 T MRI under four imaging conditions: "Axial-T₁-Gradient-Echo (GRE)", "Sagittal-T₁-GRE", "Axial-T₂-Spin-Echo (SE)" and "Sagittal-T₂-SE". Images were analysed regarding image artefact amount. The lower magnetic susceptibility of Ti-alloy specimens caused 84.76% less deterioration than 316LVM specimens in the MR images with the mean image artefact-to-specimen size ratio. Thinner implant designs provided better performance regarding the metal artefact by reducing the artefact-to-specimen size ratio. T₂SE decreased the image artefact by 44.7% for 316LVM and 54.6% for Ti-Alloy specimens and provided better image quality than T₁GRE for clinical interpretation. This study reveals that image artefacts directly depend on material content, implant volume, geometry, and imaging sequence selection. The minor artefact effect of T₂SE provides more accurate MR images than T₁GRE regarding the interpretation of the images of the patients with biometals. The higher magnetic susceptibility of biometals causes more deterioration of the images.

[Applications of Artificial Intelligence in MR Image Acquisition and Reconstruction]

Recently, artificial intelligence (AI) technology has shown potential clinical utility in a wide range of MRI fields. In particular, AI models for improving the efficiency of the image acquisition process and the quality of reconstructed images are being actively developed by the MR research community. AI is expected to further reduce acquisition times in various MRI protocols used in clinical practice when compared to current parallel imaging techniques. Additionally, AI can help with tasks such as planning, parameter optimization, artifact reduction, and quality assessment. Furthermore, AI is being actively applied to automate MR image analysis such as image registration, segmentation, and object detection. For this reason, it is important to consider the effects of protocols or devices in MR image analysis. In this review article, we briefly introduced issues related to AI application of MR image acquisition and reconstruction.

High-fidelity control of spin ensemble dynamics via artificial intelligence: from quantum computing to NMR spectroscopy and imaging

High-fidelity control of spin ensemble dynamics is essential for many research areas, spanning from quantum computing and radio-frequency (RF) engineering to NMR spectroscopy and imaging. However, attaining robust and high-fidelity spin operations remains an unmet challenge. Using an evolutionary algorithm and artificial intelligence (AI), we designed new RF pulses with customizable spatial or temporal field inhomogeneity compensation. Compared with the standard RF shapes, the new AI-generated pulses show superior performance for bandwidth, robustness, and tolerance to field imperfections. As a benchmark, we constructed a spin entanglement operator for the weakly coupled two-spin-1/2 system of ¹³CHCl₃, achieving high-fidelity transformations under multiple inhomogeneity sources. We then generated band-selective and ultra-broadband RF pulses typical of biomolecular NMR spectroscopy. When implemented in multipulse NMR experiments, the AI-generated pulses significantly increased the sensitivity of medium-size and large protein spectra relative to standard pulse sequences. Finally, we applied the new pulses to typical imaging experiments, showing a remarkable tolerance to changes in the RF field. These AI-generated RF pulses can be directly implemented in quantum information, NMR spectroscopy of biomolecules, magnetic resonance imaging techniques for in vivo and materials sciences.

Optimal control gradient precision trade-offs: Application to fast generation of DeepControl libraries for MRI

We have recently demonstrated supervised deep learning methods for rapid generation of radiofrequency pulses in magnetic resonance imaging (https://doi.org/10.1002/mrm.27740, https://doi.org/10.1002/mrm.28667). Unlike the previous iterative optimization approaches, deep learning methods generate a pulse using a fixed number of floating-point operations - this is important in MRI, where patient-specific pulses preferably must be produced in real time. However, deep learning requires vast training libraries, which must be generated using the traditional methods, e.g., iterative quantum optimal control methods. Those methods are usually variations of gradient descent, and the calculation of the gradient of the performance metric with respect to the pulse waveform can be the most numerically intensive step. In this communication, we explore various ways in which the calculation of gradients in quantum optimal control theory may be accelerated. Four optimization avenues are explored: truncated commutator series expansions at zeroth and first order, a novel midpoint truncation scheme at first order, and the exact complex-step method. For the spin systems relevant to MRI, the first-order midpoint truncation is found to be sufficiently accurate, but also significantly faster than the machine precision gradient. This makes the generation of training databases for the machine learning methods considerably more realistic.

Deep Learning and Its Application to Function Approximation for MR in Medicine: An Overview

This article presents an overview of deep learning (DL) and its applications to function approximation for MR in medicine. The aim of this article is to help readers develop various applications of DL. DL has made a large impact on the literature of many medical sciences, including MR. However, its technical details are not easily understandable for non-experts of machine learning (ML).The first part of this article presents an overview of DL and its related technologies, such as artificial intelligence (AI) and ML. AI is explained as a function that can receive many inputs and produce many outputs. ML is a process of fitting the function to training data. DL is a kind of ML, which uses a composite of many functions to approximate the function of interest. This composite function is called a deep neural network (DNN), and the functions composited into a DNN are called layers. This first part also covers the underlying technologies required for DL, such as loss functions, optimization, initialization, linear layers, non-linearities, normalization, recurrent neural networks, regularization, data augmentation, residual connections, autoencoders, generative adversarial networks, model and data sizes, and complex-valued neural networks.The second part of this article presents an overview of the applications of DL in MR and explains how functions represented as DNNs are applied to various applications, such as RF pulse, pulse sequence, reconstruction, motion correction, spectroscopy, parameter mapping, image synthesis, and segmentation.

Three-dimensional static and dynamic parallel transmission of the human heart at 7 T

Three-dimensional (3D) human heart imaging at ultra-high fields is highly challenging due to respiratory and cardiac motion-induced artifacts as well as spatially heterogeneous B1+ profiles. In this study, we investigate the feasibility of applying 3D flip angle (FA) homogenization targeting the whole heart via static phase-only and dynamic kT-point in vivo parallel transmission at 7 T. 3D B1+ maps of the thorax were acquired under free breathing in eight subjects to compute parallel transmission pulses that improve excitation homogeneity in the human heart. To analyze the number of kT-points required, excitation homogeneity and radiofrequency (RF) power were compared using different regions of interest in six subjects with different body mass index (BMI) values of 20-34 kg/m² for a wide range of regularization parameters. One subset of the optimized subject-specific pulses was applied in vivo on a 7 T scanner for six subjects in Cartesian 3D breath-hold scans as well as in two subjects in a radial phase-encoded 3D free-breathing scan. Across all subjects, 3-4 kT-points achieved a good tradeoff between RF power and nominal FA homogeneity. For subjects with a BMI in the normal range, the 4 kT-point pulses reliably improved the coefficient of variation by less than 10% compared with less than 25% achieved by static phase-only parallel transmission. in vivo measurements on a 7 T scanner validated the B1+ estimations and the pulse design, despite neglecting ΔB₀ in the optimizations and Bloch simulations. This study demonstrates in vivo that kT-point pTx pulses are highly suitable for mitigating nominal FA heterogeneities across the entire 3D heart volume at 7 T. Furthermore, 3-4 kT-points demonstrate a practical tradeoff between nominal FA heterogeneity mitigation and RF power.

Multi-task convolutional neural network-based design of radio frequency pulse and the accompanying gradients for magnetic resonance imaging

Modern MRI systems usually load the predesigned RFs and the accompanying gradients during clinical scans, with minimal adaption to the specific requirements of each scan. Here, we describe a neural network-based method for real-time design of excitation RF pulses and the accompanying gradients' waveforms to achieve spatially two-dimensional selectivity. Nine thousand sets of radio frequency (RF) and gradient waveforms with two-dimensional spatial selectivity were generated as the training dataset using the Shinnar-Le Roux (SLR) method. Neural networks were created and trained with five strategies (TS-1 to TS-5). The neural network-designed RF and gradients were compared with their SLR-designed counterparts and underwent Bloch simulation and phantom imaging to investigate their performances in spin manipulations. We demonstrate a convolutional neural network (TS-5) with multi-task learning to yield both the RF pulses and the accompanying two channels of gradient waveforms that comply with the SLR design, and these design results also provide excitation spatial profiles comparable with SLR pulses in both simulation (normalized root mean square error [NRMSE] of 0.0075 ± 0.0038 over the 400 sets of testing data between TS-5 and SLR) and phantom imaging. The output RF and gradient waveforms between the neural network and SLR methods were also compared, and the joint NRMSE, with both RF and the two channels of gradient waveforms considered, was 0.0098 ± 0.0024 between TS-5 and SLR. The RF and gradients were generated on a commercially available workstation, which took ~130 ms for TS-5. In conclusion, we present a convolutional neural network with multi-task learning, trained with SLR transformation pairs, that is capable of simultaneously generating RF and two channels of gradient waveforms, given the desired spatially two-dimensional excitation profiles.

DeepControl: 2DRF pulses facilitating B 1 + inhomogeneity and B0 off-resonance compensation in vivo at 7 T

PURPOSE

Rapid 2DRF pulse design with subject-specific B 1 + inhomogeneity and B₀ off-resonance compensation at 7 T predicted from convolutional neural networks is presented.

METHODS

The convolution neural network was trained on half a million single-channel transmit 2DRF pulses optimized with an optimal control method using artificial 2D targets, B 1 + and B₀ maps. Predicted pulses were tested in a phantom and in vivo at 7 T with measured B 1 + and B₀ maps from a high-resolution gradient echo sequence.

RESULTS

Pulse prediction by the trained convolutional neural network was done on the fly during the MR session in approximately 9 ms for multiple hand-drawn regions of interest and the measured B 1 + and B₀ maps. Compensation of B 1 + inhomogeneity and B₀ off-resonances has been confirmed in the phantom and in vivo experiments. The reconstructed image data agree well with the simulations using the acquired B 1 + and B₀ maps, and the 2DRF pulse predicted by the convolutional neural networks is as good as the conventional RF pulse obtained by optimal control.

CONCLUSION

The proposed convolutional neural network-based 2DRF pulse design method predicts 2DRF pulses with an excellent excitation pattern and compensated B 1 + and B₀ variations at 7 T. The rapid 2DRF pulse prediction (9 ms) enables subject-specific high-quality 2DRF pulses without the need to run lengthy optimizations.

Development, validation, and pilot MRI safety study of a high-resolution, open source, whole body pediatric numerical simulation model

Numerical body models of children are used for designing medical devices, including but not limited to optical imaging, ultrasound, CT, EEG/MEG, and MRI. These models are used in many clinical and neuroscience research applications, such as radiation safety dosimetric studies and source localization. Although several such adult models have been reported, there are few reports of full-body pediatric models, and those described have several limitations. Some, for example, are either morphed from older children or do not have detailed segmentations. Here, we introduce a 29-month-old male whole-body native numerical model, "MARTIN", that includes 28 head and 86 body tissue compartments, segmented directly from the high spatial resolution MRI and CT images. An advanced auto-segmentation tool was used for the deep-brain structures, whereas 3D Slicer was used to segment the non-brain structures and to refine the segmentation for all of the tissue compartments. Our MARTIN model was developed and validated using three separate approaches, through an iterative process, as follows. First, the calculated volumes, weights, and dimensions of selected structures were adjusted and confirmed to be within 6% of the literature values for the 2-3-year-old age-range. Second, all structural segmentations were adjusted and confirmed by two experienced, sub-specialty certified neuro-radiologists, also through an interactive process. Third, an additional validation was performed with a Bloch simulator to create synthetic MR image from our MARTIN model and compare the image contrast of the resulting synthetic image with that of the original MRI data; this resulted in a "structural resemblance" index of 0.97. Finally, we used our model to perform pilot MRI safety simulations of an Active Implantable Medical Device (AIMD) using a commercially available software platform (Sim4Life), incorporating the latest International Standards Organization guidelines. This model will be made available on the Athinoula A. Martinos Center for Biomedical Imaging website.

Deep Reinforcement Learning Designed Shinnar-Le Roux RF Pulse Using Root-Flipping: DeepRFSLR

A novel approach of applying deep reinforcement learning to an RF pulse design is introduced. This method, which is referred to as DeepRF_SLR, is designed to minimize the peak amplitude or, equivalently, minimize the pulse duration of a multiband refocusing pulse generated by the Shinar Le-Roux (SLR) algorithm. In the method, the root pattern of SLR polynomial, which determines the RF pulse shape, is optimized by iterative applications of deep reinforcement learning and greedy tree search. When tested for the designs of the multiband pulses with three and seven slices, DeepRF_SLR demonstrated improved performance compared to conventional methods, generating shorter duration RF pulses in shorter computational time. In the experiments, the RF pulse from DeepRF_SLR produced a slice profile similar to the minimum-phase SLR RF pulse and the profiles matched to that of the computer simulation. Our approach suggests a new way of designing an RF by applying a machine learning algorithm, demonstrating a "machine-designed" MRI sequence.

Predicting in vivo MRI Gradient-Field Induced Voltage Levels on Implanted Deep Brain Stimulation Systems Using Neural Networks

Introduction

MRI gradient-fields may induce extrinsic voltage between electrodes and conductive neurostimulator enclosure of implanted deep brain stimulation (DBS) systems, and may cause unintended stimulation and/or malfunction. Electromagnetic (EM) simulations using detailed anatomical human models, therapy implant trajectories, and gradient coil models can be used to calculate clinically relevant induced voltage levels. Incorporating additional anatomical human models into the EM simulation library can help to achieve more clinically relevant and accurate induced voltage levels, however, adding new anatomical human models and developing implant trajectories is time-consuming, expensive and not always feasible.

Methods

MRI gradient-field induced voltage levels are simulated in six adult human anatomical models, along clinically relevant DBS implant trajectories to generate the dataset. Predictive artificial neural network (ANN) regression models are trained on the simulated dataset. Leave-one-out cross validation is performed to assess the performance of ANN regressors and quantify model prediction errors.

Results

More than 180,000 unique gradient-induced voltage levels are simulated. ANN algorithm with two fully connected layers is selected due to its superior generalizability compared to support vector machine and tree-based algorithms in this particular application. The ANN regression model is capable of producing thousands of gradient-induced voltage predictions in less than a second with mean-squared-error less than 200 mV.

Conclusion

We have integrated machine learning (ML) with computational modeling and simulations and developed an accurate predictive model to determine MRI gradient-field induced voltage levels on implanted DBS systems.