arXiv subjects

Michael Lustig

Publications and source records attributed to Michael Lustig.

At least 19 recordsLinked to original sources

arXiv2025

MRDust: Wireless Implant Data Uplink & Localization via Magnetic Resonance Image Modulation

Magnetic resonance imaging (MRI) exhibits rich and clinically useful endogenous contrast mechanisms, which can differentiate soft tissues and are sensitive to flow, diffusion, magnetic susceptibility, blood oxygenation level, and more. However, MRI sensitivity is ultimately constrained by Nuclear Magnetic Resonance (NMR) physics, and its spatiotemporal resolution is limited by SNR and spatial encoding. On the other hand, miniaturized implantable sensors offer highly localized physiological information, yet communication and localization can be challenging when multiple implants are present. This paper introduces the MRDust, an active ``contrast agent" that integrates active sensor implants with MRI, enabling the direct encoding of highly localized physiological data into MR images to augment the anatomical images. MRDust employs a micrometer-scale on-chip coil to actively modulate the local magnetic field, enabling MR signal amplitude and phase modulation for digital data transmission. Since MRI inherently captures the anatomical tissue structure, this method has the potential to enable simultaneous data communication, localization, and image registration with multiple implants. This paper presents the underlying physical principles, design tradeoffs, and design methodology for this approach. To validate the concept, a 900 $\times$ 990 $\mu$m$^2$ chip was designed using TSMC 28 nm technology, with an on-chip coil measuring 630 $\mu$m in diameter. The chip was tested with custom hardware in an MR750W GE3T MRI scanner. Successful voxel amplitude modulation is demonstrated with Spin-Echo Echo-Planar-Imaging (SE-EPI) sequence, achieving a contrast-to-noise ratio (CNR) of 25.58 with a power consumption of 130 $\mu$W.

eess.SY↗

arXiv2024

Design and Control of a Compact Series Elastic Actuator Module for Robots in MRI Scanners

Robotic assistance has broadened the capabilities of magnetic resonance imaging (MRI)-guided medical interventions, yet force-controlled actuators tailored for MRI environments remain limited. In this study, we present a novel MRI-compatible rotary series elastic actuator (SEA) module that employs velocity-sourced ultrasonic motors for force-controlled operation within MRI scanners. Unlike prior MRI-compatible SEA designs, our module uses a transmission force sensing SEA architecture, with four off-the-shelf compression springs placed between the gearbox and motor housings. To enable precise torque control, we develop a controller based on a disturbance observer, specifically designed for velocity-sourced motors. This controller improves torque regulation, even under varying external impedance, enhancing the actuator's suitability for MRI-guided medical interventions. Experimental validation confirms effective torque control in both 3 Tesla MRI and non-MRI settings, achieving a 5% settling time of 0.05 seconds and steady-state error within 2.5% of the actuator's maximum output torque. Notably, the controller maintains consistent performance across both low and high impedance conditions.

cs.RO↗

arXiv2024

Reference-Free Image Quality Metric for Degradation and Reconstruction Artifacts

Image Quality Assessment (IQA) is essential in various Computer Vision tasks such as image deblurring and super-resolution. However, most IQA methods require reference images, which are not always available. While there are some reference-free IQA metrics, they have limitations in simulating human perception and discerning subtle image quality variations. We hypothesize that the JPEG quality factor is representatives of image quality measurement, and a well-trained neural network can learn to accurately evaluate image quality without requiring a clean reference, as it can recognize image degradation artifacts based on prior knowledge. Thus, we developed a reference-free quality evaluation network, dubbed "Quality Factor (QF) Predictor", which does not require any reference. Our QF Predictor is a lightweight, fully convolutional network comprising seven layers. The model is trained in a self-supervised manner: it receives JPEG compressed image patch with a random QF as input, is trained to accurately predict the corresponding QF. We demonstrate the versatility of the model by applying it to various tasks. First, our QF Predictor can generalize to measure the severity of various image artifacts, such as Gaussian Blur and Gaussian noise. Second, we show that the QF Predictor can be trained to predict the undersampling rate of images reconstructed from Magnetic Resonance Imaging (MRI) data.

eess.IV↗

arXiv2023

K-band: Self-supervised MRI Reconstruction via Stochastic Gradient Descent over K-space Subsets

Although deep learning (DL) methods are powerful for solving inverse problems, their reliance on high-quality training data is a major hurdle. This is significant in high-dimensional (dynamic/volumetric) magnetic resonance imaging (MRI), where acquisition of high-resolution fully sampled k-space data is impractical. We introduce a novel mathematical framework, dubbed k-band, that enables training DL models using only partial, limited-resolution k-space data. Specifically, we introduce training with stochastic gradient descent (SGD) over k-space subsets. In each training iteration, rather than using the fully sampled k-space for computing gradients, we use only a small k-space portion. This concept is compatible with different sampling strategies; here we demonstrate the method for k-space "bands", which have limited resolution in one dimension and can hence be acquired rapidly. We prove analytically that our method stochastically approximates the gradients computed in a fully-supervised setup, when two simple conditions are met: (i) the limited-resolution axis is chosen randomly-uniformly for every new scan, hence k-space is fully covered across the entire training set, and (ii) the loss function is weighed with a mask, derived here analytically, which facilitates accurate reconstruction of high-resolution details. Numerical experiments with raw MRI data indicate that k-band outperforms two other methods trained on limited-resolution data and performs comparably to state-of-the-art (SoTA) methods trained on high-resolution data. k-band hence obtains SoTA performance, with the advantage of training using only limited-resolution data. This work hence introduces a practical, easy-to-implement, self-supervised training framework, which involves fast acquisition and self-supervised reconstruction and offers theoretical guarantees.

eess.IV↗

arXiv2023

Beat Pilot Tone (BPT): Simultaneous MR Imaging and RF Motion Sensing at Arbitrary Frequencies

Purpose: To introduce a simple system exploitation with the potential to turn MRI scanners into general-purpose RF motion monitoring systems. Methods: Inspired by Pilot Tone (PT), this work proposes Beat Pilot Tone (BPT), in which two or more RF tones at arbitrary frequencies are transmitted continuously during the scan. These tones create motion-modulated standing wave patterns that are sensed by the receiver coil array, incidentally mixed by intermodulation in the receiver chain, and digitized simultaneously with the MRI data. BPT can operate at almost any frequency as long as the intermodulation products lie within the bandwidth of the receivers. BPT's mechanism is explained in electromagnetic simulations and validated experimentally. Results: Phantom and volunteer experiments over a range of transmit frequencies suggest that BPT may offer frequency-dependent sensitivity to motion. Using a semi-flexible body receiver array, BPT appears to sense cardiac-induced body vibrations at microwave frequencies (1.2 GHz and greater). At lower frequencies, it exhibits a similar cardiac signal shape to PT, likely due to blood volume changes. Other volunteer experiments with respiratory, bulk, and head motion show that BPT can achieve greater sensitivity to motion than PT and greater separability between motion types. Basic multiple-input multiple-output (4x22 MIMO) operation with simultaneous PT and BPT in head motion is demonstrated using two transmit antennas and a 22-channel head-neck coil. Conclusion: BPT may offer a rich source of motion information that is frequency-dependent, simultaneous, and complementary to PT and the MRI exam.

physics.med-ph↗

arXiv2022

High-fidelity Direct Contrast Synthesis from Magnetic Resonance Fingerprinting

Magnetic Resonance Fingerprinting (MRF) is an efficient quantitative MRI technique that can extract important tissue and system parameters such as T1, T2, B0, and B1 from a single scan. This property also makes it attractive for retrospectively synthesizing contrast-weighted images. In general, contrast-weighted images like T1-weighted, T2-weighted, etc., can be synthesized directly from parameter maps through spin-dynamics simulation (i.e., Bloch or Extended Phase Graph models). However, these approaches often exhibit artifacts due to imperfections in the mapping, the sequence modeling, and the data acquisition. Here we propose a supervised learning-based method that directly synthesizes contrast-weighted images from the MRF data without going through the quantitative mapping and spin-dynamics simulation. To implement our direct contrast synthesis (DCS) method, we deploy a conditional Generative Adversarial Network (GAN) framework and propose a multi-branch U-Net as the generator. The input MRF data are used to directly synthesize T1-weighted, T2-weighted, and fluid-attenuated inversion recovery (FLAIR) images through supervised training on paired MRF and target spin echo-based contrast-weighted scans. In-vivo experiments demonstrate excellent image quality compared to simulation-based contrast synthesis and previous DCS methods, both visually as well as by quantitative metrics. We also demonstrate cases where our trained model is able to mitigate in-flow and spiral off-resonance artifacts that are typically seen in MRF reconstructions and thus more faithfully represent conventional spin echo-based contrast-weighted images.

eess.IV↗

arXiv2022

SLfRank: Shinnar-Le-Roux Pulse Design with Reduced Energy and Accurate Phase Profiles using Rank Factorization

The Shinnar-Le-Roux (SLR) algorithm is widely used to design frequency selective pulses with large flip angles. We improve its design process to generate pulses with lower energy (by as much as 26%) and more accurate phase profiles. Concretely, the SLR algorithm consists of two steps: (1) an invertible transform between frequency selective pulses and polynomial pairs that represent Cayley-Klein (CK) parameters and (2) the design of the CK polynomial pair to match the desired magnetization profiles. Because the CK polynomial pair is bi-linearly coupled, the original algorithm sequentially solves for each polynomial instead of jointly. This results in sub-optimal pulses. Instead, we leverage a convex relaxation technique, commonly used for low rank matrix recovery, to address the bi-linearity. Our numerical experiments show that the resulting pulses are almost always globally optimal in practice. For slice excitation, the proposed algorithm results in more accurate linear phase profiles. And in general the improved pulses have lower energy than the original SLR pulses.

physics.med-ph↗

arXiv2022

Rapid Fabrication of Low Impedance, 3D Dry Electrodes for Physiological Sensing

Medical electrophysiological sensors that can study the body and diagnose diseases depend on consistently low impedance electrode-skin interfaces. Clinical-standard wet electrodes use hydrogels and skin abrasion to improve the interface and thus the recorded signal quality. These electrodes are challenging to self-administer and impede in-home care. Wearable dry electrodes are more practical; however, they show higher impedances than wet electrodes and are costly to customize. This work presents a fabrication method for rapidly producing low impedance, anatomically fit dry electrodes that do not require hydrogels. By using electroless copper and gold plating with 3D printing, biocompatible electrodes can be optimized for individuals at a fraction of the cost of existing vacuum deposition-based techniques. Example 3D dry electrodes made with this process are evaluated alongside clinical-standard devices in typical scenarios to compare electrical performance and comfort. The resulting dry electrodes exhibited an average electrode-skin impedance of 66.7 kΩ at 50Hz and DC offset of -20 mV without any hydrogel, which, when area normalized, is within the range achieved by wet electrodes without skin abrasion.

physics.med-ph↗

arXiv2021

Subtle Data Crimes: Naively training machine learning algorithms could lead to overly-optimistic results

While open databases are an important resource in the Deep Learning (DL) era, they are sometimes used "off-label": data published for one task are used for training algorithms for a different one. This work aims to highlight that in some cases, this common practice may lead to biased, overly-optimistic results. We demonstrate this phenomenon for inverse problem solvers and show how their biased performance stems from hidden data preprocessing pipelines. We describe two preprocessing pipelines typical of open-access databases and study their effects on three well-established algorithms developed for Magnetic Resonance Imaging (MRI) reconstruction: Compressed Sensing (CS), Dictionary Learning (DictL), and DL. In this large-scale study we performed extensive computations. Our results demonstrate that the CS, DictL and DL algorithms yield systematically biased results when naively trained on seemingly-appropriate data: the Normalized Root Mean Square Error (NRMSE) improves consistently with the preprocessing extent, showing an artificial increase of 25%-48% in some cases. Since this phenomenon is generally unknown, biased results are sometimes published as state-of-the-art; we refer to that as subtle data crimes. This work hence raises a red flag regarding naive off-label usage of Big Data and reveals the vulnerability of modern inverse problem solvers to the resulting bias.

cs.LG↗

arXiv2021

High Fidelity Deep Learning-based MRI Reconstruction with Instance-wise Discriminative Feature Matching Loss

Purpose: To improve reconstruction fidelity of fine structures and textures in deep learning (DL) based reconstructions. Methods: A novel patch-based Unsupervised Feature Loss (UFLoss) is proposed and incorporated into the training of DL-based reconstruction frameworks in order to preserve perceptual similarity and high-order statistics. The UFLoss provides instance-level discrimination by mapping similar instances to similar low-dimensional feature vectors and is trained without any human annotation. By adding an additional loss function on the low-dimensional feature space during training, the reconstruction frameworks from under-sampled or corrupted data can reproduce more realistic images that are closer to the original with finer textures, sharper edges, and improved overall image quality. The performance of the proposed UFLoss is demonstrated on unrolled networks for accelerated 2D and 3D knee MRI reconstruction with retrospective under-sampling. Quantitative metrics including NRMSE, SSIM, and our proposed UFLoss were used to evaluate the performance of the proposed method and compare it with others. Results: In-vivo experiments indicate that adding the UFLoss encourages sharper edges and more faithful contrasts compared to traditional and learning-based methods with pure l2 loss. More detailed textures can be seen in both 2D and 3D knee MR images. Quantitative results indicate that reconstruction with UFLoss can provide comparable NRMSE and a higher SSIM while achieving a much lower UFLoss value. Conclusion: We present UFLoss, a patch-based unsupervised learned feature loss, which allows the training of DL-based reconstruction to obtain more detailed texture, finer features, and sharper edges with higher overall image quality under DL-based reconstruction frameworks.

eess.IV↗

arXiv2021

Memory-efficient Learning for High-Dimensional MRI Reconstruction

Deep learning (DL) based unrolled reconstructions have shown state-of-the-art performance for under-sampled magnetic resonance imaging (MRI). Similar to compressed sensing, DL can leverage high-dimensional data (e.g. 3D, 2D+time, 3D+time) to further improve performance. However, network size and depth are currently limited by the GPU memory required for backpropagation. Here we use a memory-efficient learning (MEL) framework which favorably trades off storage with a manageable increase in computation during training. Using MEL with multi-dimensional data, we demonstrate improved image reconstruction performance for in-vivo 3D MRI and 2D+time cardiac cine MRI. MEL uses far less GPU memory while marginally increasing the training time, which enables new applications of DL to high-dimensional MRI.

eess.IV↗

arXiv2020

SURE-based Automatic Parameter Selection For ESPIRiT Calibration

Purpose: Parallel imaging methods in MRI have resulted in faster acquisition times and improved noise performance. ESPIRiT is one such technique that estimates coil sensitivity maps from the auto-calibration region using an eigenvalue-based method. This method requires choosing several parameters for the the map estimation. Even though ESPIRiT is fairly robust to these parameter choices, occasionally, poor selection can result in reduced performance. The purpose of this work is to automatically select parameters in ESPIRiT for more robust and consistent performance across a variety of exams. Theory and Methods: Stein's unbiased risk estimate (SURE) is a method of calculating an unbiased estimate of the mean squared error of an estimator under certain assumptions. We show that this can be used to estimate the performance of ESPIRiT. We derive and demonstrate the use of SURE to optimize ESPIRiT parameter selection. Results: Simulations show SURE to be an accurate estimator of the mean squared error. SURE is then used to optimize ESPIRiT parameters to yield maps that are optimal in a denoising/data-consistency sense. This improves g-factor performance without causing undesirable attenuation. In-vivo experiments verify the reliability of this method. Conclusion: Simulation experiments demonstrate that SURE is an accurate estimate of expected mean squared error. Using SURE to determine ESPIRiT parameters allows for automatic parameter selections.In-vivo results are consistent with simulation and theoretical results.

physics.med-ph↗

arXiv2020

How to do Physics-based Learning

The goal of this tutorial is to explain step-by-step how to implement physics-based learning for the rapid prototyping of a computational imaging system. We provide a basic overview of physics-based learning, the construction of a physics-based network, and its reduction to practice. Specifically, we advocate exploiting the auto-differentiation functionality twice, once to build a physics-based network and again to perform physics-based learning. Thus, the user need only implement the forward model process for their system, speeding up prototyping time. We provide an open-source Pytorch implementation of a physics-based network and training procedure for a generic sparse recovery problem

eess.IV↗

arXiv2020

Memory-efficient Learning for Large-scale Computational Imaging -- NeurIPS deep inverse workshop

Computational imaging systems jointly design computation and hardware to retrieve information which is not traditionally accessible with standard imaging systems. Recently, critical aspects such as experimental design and image priors are optimized through deep neural networks formed by the unrolled iterations of classical physics-based reconstructions (termed physics-based networks). However, for real-world large-scale systems, computing gradients via backpropagation restricts learning due to memory limitations of graphical processing units. In this work, we propose a memory-efficient learning procedure that exploits the reversibility of the network's layers to enable data-driven design for large-scale computational imaging. We demonstrate our methods practicality on two large-scale systems: super-resolution optical microscopy and multi-channel magnetic resonance imaging.

eess.IV↗

arXiv2020

Memory-efficient Learning for Large-scale Computational Imaging

Critical aspects of computational imaging systems, such as experimental design and image priors, can be optimized through deep networks formed by the unrolled iterations of classical model-based reconstructions (termed physics-based networks). However, for real-world large-scale inverse problems, computing gradients via backpropagation is infeasible due to memory limitations of graphics processing units. In this work, we propose a memory-efficient learning procedure that exploits the reversibility of the network's layers to enable data-driven design for large-scale computational imaging systems. We demonstrate our method on a small-scale compressed sensing example, as well as two large-scale real-world systems: multi-channel magnetic resonance imaging and super-resolution optical microscopy.

cs.CV↗

arXiv2020

Unsupervised Deep Basis Pursuit: Learning inverse problems without ground-truth data

Basis pursuit is a compressed sensing optimization in which the l1-norm is minimized subject to model error constraints. Here we use a deep neural network prior instead of l1-regularization. Using known noise statistics, we jointly learn the prior and reconstruct images without access to ground-truth data. During training, we use alternating minimization across an unrolled iterative network and jointly solve for the neural network weights and training set image reconstructions. At inference, we fix the weights and pass the measurements through the network. We compare reconstruction performance between unsupervised and supervised (i.e. with ground-truth) methods. We hypothesize this technique could be used to learn reconstruction when ground-truth data are unavailable, such as in high-resolution dynamic MRI.

eess.SP↗

arXiv2020

Extreme MRI: Large-Scale Volumetric Dynamic Imaging from Continuous Non-Gated Acquisitions

Purpose: To develop a framework to reconstruct large-scale volumetric dynamic MRI from rapid continuous and non-gated acquisitions, with applications to pulmonary and dynamic contrast enhanced (DCE) imaging. Theory and Methods: The problem considered here requires recovering hundred-gigabytes of dynamic volumetric image data from a few gigabytes of k-space data, acquired continuously over several minutes. This reconstruction is vastly under-determined, heavily stressing computing resources as well as memory management and storage. To overcome these challenges, we leverage intrinsic three dimensional (3D) trajectories, such as 3D radial and 3D cones, with ordering that incoherently cover time and k-space over the entire acquisition. We then propose two innovations: (1) A compressed representation using multi-scale low rank matrix factorization that constrains the reconstruction problem, and reduces its memory footprint. (2) Stochastic optimization to reduce computation, improve memory locality, and minimize communications between threads and processors. We demonstrate the feasibility of the proposed method on DCE imaging acquired with a golden-angle ordered 3D cones trajectory and pulmonary imaging acquired with a bit-reversed ordered 3D radial trajectory. We compare it with "soft-gated" dynamic reconstruction for DCE and respiratory resolved reconstruction for pulmonary imaging. Results: The proposed technique shows transient dynamics that are not seen in gating based methods. When applied to datasets with irregular, or non-repetitive motions, the proposed method displays sharper image features. Conclusion: We demonstrated a method that can reconstruct massive 3D dynamic image series in the extreme undersampling and extreme computation setting.

physics.med-ph↗

arXiv2020

Iterative Motion Compensation reconstruction ultra-short TE(iMoCo UTE) for high resolution free breathing pulmonary MRI

Purpose, To develop a high scanning efficiency, motion corrected imaging strategy for free-breathing pulmonary MRI by combining a motion compensation reconstruction with a UTE acquisition, called iMoCo UTE. Methods, An optimized golden angle ordering radial UTE sequence was used to continuously acquire data for 5 minutes. All readouts were grouped to different respiratory motion states based on self-navigator signals, then motion resolved data was reconstructed by XD Golden angle RAdial Sparse Parallel reconstruction (XD GRASP). One state from the motion resolved images was selected as a reference, and motion fields from the other states to the reference were derived via non-rigid registration. Finally, all motion resolved data and motion fields were reconstructed by using an iterative motion compensation reconstruction with a total generalized variation sparse constraint. Results, The iMoCo UTE strategy was evaluated in volunteers and non-sedated pediatric patient(4-6 y/o) studies. Images reconstructed with iMoCo UTE provided sharper anatomical lung structures, and higher apparent SNR and CNR, compared to using other motion correction strategies, such as soft-gating, motion resolved reconstruction, and non-rigid motion compensation(MoCo). iMoCo UTE also showed promising results in an infant study. Conclusions, The proposed iMoCo UTE combines self-navigation, motion modeling, and a compressed sensing reconstruction to increase scan efficiency, SNR, and reduce respiratory motion in lung MRI. This proposed strategy shows improvements in free breathing lung MRI scans, especially in very challenging application situations, such as pediatric MRI studies.

eess.IV↗