-
K-band: Self-supervised MRI Reconstruction via Stochastic Gradient Descent over K-space Subsets
Authors:
Frederic Wang,
Han Qi,
Alfredo De Goyeneche,
Reinhard Heckel,
Michael Lustig,
Efrat Shimron
Abstract:
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…
▽ More
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.
△ Less
Submitted 23 May, 2024; v1 submitted 5 August, 2023;
originally announced August 2023.
-
A Cryogenic Torsion Balance Using a Liquid-Cryogen Free, Ultra-Low Vibration Cryostat
Authors:
S. M. Fleischer,
M. P. Ross,
K. Venkateswara,
C. A. Hagedorn,
E. A. Shaw,
E. Swanson,
B. R. Heckel,
J. H. Gundlach
Abstract:
We describe a liquid-cryogen free cryostat with ultra-low vibration levels which allows for continuous operation of a torsion balance at cryogenic temperatures. The apparatus uses a commercially available two-stage pulse-tube cooler and passive vibration isolation. The torsion balance exhibits torque noise levels lower than room temperature thermal noise by a factor of about four in the frequency…
▽ More
We describe a liquid-cryogen free cryostat with ultra-low vibration levels which allows for continuous operation of a torsion balance at cryogenic temperatures. The apparatus uses a commercially available two-stage pulse-tube cooler and passive vibration isolation. The torsion balance exhibits torque noise levels lower than room temperature thermal noise by a factor of about four in the frequency range of 3-10mHz, limited by residual seismic motion and by radiative heating of the pendulum body. In addition to lowering thermal noise below room-temperature limits, the low-temperature environment enables novel torsion balance experiments. Currently, the maximum duration of a continuous measurement run is limited by accumulation of cryogenic surface contamination on the optical elements inside the cryostat.
△ Less
Submitted 8 November, 2022; v1 submitted 6 June, 2022;
originally announced June 2022.
-
New Test of the Gravitational $1/r^2$ Law at Separations down to 52 $μ$m
Authors:
J. G. Lee,
E. G. Adelberger,
T. S. Cook,
S. M. Fleischer,
B. R. Heckel
Abstract:
We tested the gravitational $1/r^2$ law using a stationary torsion-balance detector and a rotating attractor containing test bodies with both 18-fold and 120-fold azimuthal symmetries that simultaneously tests the $1/r^2$ law at two different length scales. We took data at detector-attractor separations between $52~μ$m and 3.0 mm. Newtonian gravity gave an excellent fit to our data, limiting with…
▽ More
We tested the gravitational $1/r^2$ law using a stationary torsion-balance detector and a rotating attractor containing test bodies with both 18-fold and 120-fold azimuthal symmetries that simultaneously tests the $1/r^2$ law at two different length scales. We took data at detector-attractor separations between $52~μ$m and 3.0 mm. Newtonian gravity gave an excellent fit to our data, limiting with 95\% confidence any gravitational-strength Yukawa interactions to ranges $< 38.6~μ$m.
△ Less
Submitted 26 February, 2020;
originally announced February 2020.
-
Deep Phase Decoder: Self-calibrating phase microscopy with an untrained deep neural network
Authors:
Emrah Bostan,
Reinhard Heckel,
Michael Chen,
Michael Kellman,
Laura Waller
Abstract:
Deep neural networks have emerged as effective tools for computational imaging including quantitative phase microscopy of transparent samples. To reconstruct phase from intensity, current approaches rely on supervised learning with training examples; consequently, their performance is sensitive to a match of training and imaging settings. Here we propose a new approach to phase microscopy by using…
▽ More
Deep neural networks have emerged as effective tools for computational imaging including quantitative phase microscopy of transparent samples. To reconstruct phase from intensity, current approaches rely on supervised learning with training examples; consequently, their performance is sensitive to a match of training and imaging settings. Here we propose a new approach to phase microscopy by using an untrained deep neural network for measurement formation, encapsulating the image prior and imaging physics. Our approach does not require any training data and simultaneously reconstructs the sought phase and pupil-plane aberrations by fitting the weights of the network to the captured images. To demonstrate experimentally, we reconstruct quantitative phase from through-focus images blindly (i.e. no explicit knowledge of the aberrations).
△ Less
Submitted 24 January, 2020;
originally announced January 2020.
-
Reduced Limit on the Permanent Electric Dipole Moment of $^{199}$Hg
Authors:
B. Graner,
Y. Chen,
E. G. Lindahl,
B. R. Heckel
Abstract:
This paper describes the results of the most recent measurement of the permanent electric dipole moment (EDM) of neutral $^{199}$Hg atoms. Fused silica vapor cells containing enriched $^{199}$Hg are arranged in a stack in a common magnetic field. Optical pum** is used to spin-polarize the atoms orthogonal to the applied magnetic field, and the Faraday rotation of near-resonant light is observed…
▽ More
This paper describes the results of the most recent measurement of the permanent electric dipole moment (EDM) of neutral $^{199}$Hg atoms. Fused silica vapor cells containing enriched $^{199}$Hg are arranged in a stack in a common magnetic field. Optical pum** is used to spin-polarize the atoms orthogonal to the applied magnetic field, and the Faraday rotation of near-resonant light is observed to determine an electric-field-induced perturbation to the Larmor precession frequency. Our results for this frequency shift are consistent with zero; we find the corresponding $^{199}$Hg EDM $d_{Hg} = (2.20 \pm 2.75_{stat} \pm 1.48_{syst}) \times 10^{-30} e\cdot \text{cm}$. We use this result to place a new upper limit on the $^{199}$Hg EDM $|d_{Hg}| < 7.4\times 10^{-30} e\cdot \text{cm}$ (95\% C.L.), improving our previous limit by a factor of 4. We also discuss the implications of this result for various $CP$-violating observables as they relate to theories of physics beyond the standard model.
△ Less
Submitted 17 August, 2017; v1 submitted 17 January, 2016;
originally announced January 2016.
-
Short-range spin-dependent interactions of electrons: a probe for exotic pseudo-Goldstone bosons
Authors:
W. A. Terrano,
E. G. Adelberger,
J. G. Lee,
B. R. Heckel
Abstract:
We used a torsion pendulum and rotating attractor with 20-pole electron-spin distributions to probe dipole-dipole interactions mediated by exotic pseudo-Goldstone bosons with $m_{\rm b}\leq 500 \; μ$eV and coupling strengths up to 14 orders of magnitude weaker than electromagnetism. This corresponds to symmetry-breaking scales $F \leq 70$ TeV, the highest reached in any laboratory experiment. We u…
▽ More
We used a torsion pendulum and rotating attractor with 20-pole electron-spin distributions to probe dipole-dipole interactions mediated by exotic pseudo-Goldstone bosons with $m_{\rm b}\leq 500 \; μ$eV and coupling strengths up to 14 orders of magnitude weaker than electromagnetism. This corresponds to symmetry-breaking scales $F \leq 70$ TeV, the highest reached in any laboratory experiment. We used an attractor with a 20-pole unpolarized mass distribution to improve laboratory bounds on $CP$-violating monopole-dipole interactions with $1.5\:μ$eV$<m_{\rm b}<400\:μ$eV by a factor of up to 1000.
△ Less
Submitted 10 August, 2015;
originally announced August 2015.
-
Apparatus for Measurement of the Electric Dipole Moment of the Neutron using a Cohabiting Atomic-Mercury Magnetometer
Authors:
C. A. Baker,
Y. Chibane,
M. Chouder,
P. Geltenbort,
K. Green,
P. G. Harris,
B. R. Heckel,
P. Iaydjiev,
S. N. Ivanov,
I. Kilvington,
S. K. Lamoreaux,
D. J. May,
J. M. Pendlebury,
J. D. Richardson,
D. B. Shiers,
K. F. Smith,
M. van der Grinten
Abstract:
A description is presented of apparatus used to carry out an experimental search for an electric dipole moment of the neutron, at the Institut Laue-Langevin (ILL), Grenoble. The experiment incorporated a cohabiting atomic-mercury magnetometer in order to reduce spurious signals from magnetic field fluctuations. The result has been published in an earlier letter; here, the methods and equipment use…
▽ More
A description is presented of apparatus used to carry out an experimental search for an electric dipole moment of the neutron, at the Institut Laue-Langevin (ILL), Grenoble. The experiment incorporated a cohabiting atomic-mercury magnetometer in order to reduce spurious signals from magnetic field fluctuations. The result has been published in an earlier letter; here, the methods and equipment used are discussed in detail.
△ Less
Submitted 5 June, 2013; v1 submitted 31 May, 2013;
originally announced May 2013.
-
Measurement of Linear Stark Interference in 199Hg
Authors:
T. H. Loftus,
M. D. Swallows,
W. C. Griffith,
M. V. Romalis,
B. R. Heckel,
E. N. Fortson
Abstract:
We present measurements of Stark interference in the 6$^1S_0$ $\rightarrow$ 6$^3P_1$ transition in $^{199}$Hg, a process whereby a static electric field $E$ mixes magnetic dipole and electric quadrupole couplings into an electric dipole transition, leading to $E$-linear energy shifts similar to those produced by a permanent atomic electric dipole moment (EDM). The measured interference amplitude,…
▽ More
We present measurements of Stark interference in the 6$^1S_0$ $\rightarrow$ 6$^3P_1$ transition in $^{199}$Hg, a process whereby a static electric field $E$ mixes magnetic dipole and electric quadrupole couplings into an electric dipole transition, leading to $E$-linear energy shifts similar to those produced by a permanent atomic electric dipole moment (EDM). The measured interference amplitude, $a_{SI}$ = $(a_{M1} + a_{E2})$ = (5.8 $\pm$ 1.5)$\times 10^{-9}$ (kV/cm)$^{-1}$, agrees with relativistic, many-body predictions and confirms that earlier central-field estimates are a factor of 10 too large. More importantly, this study validates the capability of the $^{199}$Hg EDM search apparatus to resolve non-trivial, controlled, and sub-nHz Larmor frequency shifts with EDM-like characteristics.
△ Less
Submitted 24 June, 2011; v1 submitted 24 November, 2010;
originally announced November 2010.
-
Improved limit on the permanent electric dipole moment of 199Hg
Authors:
W. C. Griffith,
M. D. Swallows,
T. H. Loftus,
M. V. Romalis,
B. R. Heckel,
E. N. Fortson
Abstract:
We report the results of a new experimental search for a permanent electric dipole moment of 199Hg utilizing a stack of four vapor cells. We find d(199Hg) = (0.49 \pm 1.29_stat \pm 0.76_syst) x 10^{-29} e cm, and interpret this as a new upper bound, |d(199Hg)| < 3.1 x 10^{-29} e cm (95% C.L.). This result improves our previous 199Hg limit by a factor of 7, and can be used to set new constraints…
▽ More
We report the results of a new experimental search for a permanent electric dipole moment of 199Hg utilizing a stack of four vapor cells. We find d(199Hg) = (0.49 \pm 1.29_stat \pm 0.76_syst) x 10^{-29} e cm, and interpret this as a new upper bound, |d(199Hg)| < 3.1 x 10^{-29} e cm (95% C.L.). This result improves our previous 199Hg limit by a factor of 7, and can be used to set new constraints on CP violation in physics beyond the standard model.
△ Less
Submitted 10 March, 2009; v1 submitted 15 January, 2009;
originally announced January 2009.
