Search

The quality of microscopy images often suffers from optical aberrations. These aberrations and their associated point spread functions have to be quantitatively estimated to restore aberrated images. The recent state-of-the-art method PhaseNet, based on a convolutional neural network, can quantify aberrations accurately but is limited to images of point light sources, e.g. fluorescent beads. In this research, we describe an extension of PhaseNet enabling its use on 3D images of biological samples. To this end, our method incorporates object-specific information into the simulated images used for training the network. Further, we add a Python-based restoration of images via Richardson-Lucy deconvolution. We demonstrate that the deconvolution with the predicted PSF can not only remove the simulated aberrations but also improve the quality of the real raw microscopic images with unknown residual PSF. We provide code for fast and convenient prediction and correction of aberrations., Comment: 7 pages, 9 figures, presented at ICFSP on 9 Sept 2022 in Paris, France, to be published in ICFSP conference proceedings in IEEE Xplore digital library

The KArlsruhe TRItium Neutrino (KATRIN) experiment, which aims to make a direct and model-independent determination of the absolute neutrino mass scale, is a complex experiment with many components. More than 15 years ago, we published a technical design report (TDR) [https://publikationen.bibliothek.kit.edu/270060419] to describe the hardware design and requirements to achieve our sensitivity goal of 0.2 eV at 90% C.L. on the neutrino mass. Since then there has been considerable progress, culminating in the publication of first neutrino mass results with the entire beamline operating [arXiv:1909.06048]. In this paper, we document the current state of all completed beamline components (as of the first neutrino mass measurement campaign), demonstrate our ability to reliably and stably control them over long times, and present details on their respective commissioning campaigns., Comment: Added missing acknowledgement, corrected performance statement in chapter 4.2.5, updated author list and references

We explore the response of the insensitive high explosive (IHE) 1,3,5-Triamino-2,4,6- trinitrobenzene (TATB) under detonation-induced shock conditions using in-situ synchrotron X-ray diffraction in the 100 ns time scale using either a conventional or a colliding detonation drive. In all of the detonation experiments on various sizes and morphologies of TATB, we observe an extended stability of the TATB triclinic crystal structure. As the detonation front passes through the TATB, X-ray diffraction indicates a portion of the TATB exhibits a compression up to 30+ GPa, followed subsequently by a pressure release and continued decomposition over a few hundred nanoseconds. Likewise, for colliding detonation-driven shock compression of single crystals of TATB, a significant portion of the triclinic crystal structure appears to be stable up to 60+ GPa. Conversely, in similar detonations of a LLM-105 PBX, X-ray diffraction is simply indicative of decomposition without the apparent compression and slow decomposition seen in TATB. The results indicate the surprising resilience of TATB under these high-pressure, temperature and shock conditions, providing a baseline for understanding the insensitivity of TATB that is considered the industry standard for an insensitive high explosive. The results also provide intriguing information for the extended reaction zone in TATB, and the hot-spot mechanisms for initiating and propagating detonation in this uniquely insensitive explosive., Comment: Accepted in Physical Review Materials, https://journals.aps.org/prmaterials/accepted/a6076KafM7c1c00414b62063da29797895775b4b6

Estimation of optical aberrations from volumetric intensity images is a key step in sensorless adaptive optics for 3D microscopy. Recent approaches based on deep learning promise accurate results at fast processing speeds. However, collecting ground truth microscopy data for training the network is typically very difficult or even impossible thereby limiting this approach in practice. Here, we demonstrate that neural networks trained only on simulated data yield accurate predictions for real experimental images. We validate our approach on simulated and experimental datasets acquired with two different microscopy modalities, and also compare the results to non-learned methods. Additionally, we study the predictability of individual aberrations with respect to their data requirements and find that the symmetry of the wavefront plays a crucial role. Finally, we make our implementation freely available as open source software in Python.

The motion of electrons in or near solids, liquids and gases can be tracked by forcing their ejection with attosecond x-ray pulses, derived from femtosecond lasers. The momentum of these emitted electrons carries the imprint of the electronic state. Aberration corrected transmission electron microscopes have observed individual atoms, and have sufficient energy sensitivity to quantify atom bonding and electronic configurations. Recent developments in ultrafast electron microscopy and diffraction indicate that spatial and temporal information can be collected simultaneously. In the present work, we push the capability of femtosecond transmission electron microscopy (fs-TEM) towards that of the state of the art in ultrafast lasers and electron microscopes. This is anticipated to facilitate unprecedented elucidation of physical, chemical and biological structural dynamics on electronic time and length scales. The fs-TEM numerically studied employs a nanotip source, electrostatic acceleration to 70 keV, magnetic lens beam transport and focusing, a condenser-objective around the sample and a terahertz temporal compressor, including space charge effects during propagation. With electron emission equivalent to a 20 fs laser pulse, we find a spatial resolution below 10 nm and a temporal resolution of below 10 fs will be feasible for pulses comprised of on average 20 electrons. The influence of a transverse electric field at the sample is modelled, indicating that a field of 1 V/$\mu$m can be resolved., Comment: 13 pages, 9 figures, 49 references

Sample-induced image-degradation remains an intricate wave-optical problem in light-sheet microscopy. Here we present biobeam, an open-source software package that enables to simulate operational light-sheet microscopes by combining data from 10^5 - 10^6 multiplexed and GPU-accelerated point-spread-function calculations. The wave-optical nature of these simulations leads to the faithful reproduction of spatially varying aberrations, diffraction artifacts, geometric image distortions, adaptive optics, and emergent wave optical phenomena, and renders image-formation in light-sheet microscopy computationally tractable., Comment: changed video codec to h264, no other changes

Lineage tracing, the joint segmentation and tracking of living cells as they move and divide in a sequence of light microscopy images, is a challenging task. Jug et al. have proposed a mathematical abstraction of this task, the moral lineage tracing problem (MLTP), whose feasible solutions define both a segmentation of every image and a lineage forest of cells. Their branch-and-cut algorithm, however, is prone to many cuts and slow convergence for large instances. To address this problem, we make three contributions: (i) we devise the first efficient primal feasible local search algorithms for the MLTP, (ii) we improve the branch-and-cut algorithm by separating tighter cutting planes and by incorporating our primal algorithms, (iii) we show in experiments that our algorithms find accurate solutions on the problem instances of Jug et al. and scale to larger instances, leveraging moral lineage tracing to practical significance., Comment: Accepted at ICCV 2017

Argon-37 is an environmental signature of an underground nuclear explosion. Producing and quantifying low-level Ar-37 standards is an important step in the development of sensitive field measurement instruments. This paper describes progress at Pacific Northwest National Laboratory in developing a process to generate and quantify low-level Ar-37 standards, which can be used to calibrate sensitive field systems at activities consistent with soil background levels. This paper presents a discussion of the measurement analysis, along with assumptions and uncertainty estimates., Comment: Proceedings of the 20th International Conference on Radionuclide Metrology and its Applications 8-11 June 2015, Vienna, Austria

Assay methods for measuring 238U, 232Th, and 210Pb concentrations in refined lead are presented. The 238U and 232Th concentrations are determined using isotope dilution inductively coupled plasma mass spectrometry (ID-ICP-MS) after anion exchange column separation of dissolved lead samples. The 210Pb concentration is inferred through {\alpha}-spectroscopy of a daughter isotope, 210Po, after chemical precipitation separation on dissolved lead samples. Subsequent to the 210Po {\alpha}-spectroscopy assay, a method for evaluating 210Pb concentrations in solid lead samples was developed via measurement of bremsstrahlung radiation from \b{eta}-decay of a daughter isotope, 210Bi, by employing a 14-crystal array of high purity germanium (HPGe) detectors. Ten sources of refined lead were assayed. The 238U concentrations were <34 microBq/kg and the 232Th concentrations ranged <0.6-15 microBq/kg, as determined by the ICP-MS assay method. The 210Pb concentrations ranged from ~0.1-75 Bq/kg, as inferred by the 210Po {\alpha}-spectroscopy assay method., Comment: Submitted to Journal of Radioanalytical and Nuclear Chemistry

Femtosecond electron microscopy produces real-space images of matter in a series of ultrafast snapshots. Pulses of electrons self-disperse under space-charge broadening, so without compression, the ideal operation mode is a single electron per pulse. Here, we demonstrate for the first time femtosecond single-electron point projection microscopy (fs-ePPM) in a laser-pump fs-e-probe configuration. The electrons have an energy of only 150 eV and take tens of picoseconds to propagate to the object under study. Nonetheless, we achieve a temporal resolution with a standard deviation of 120 fs, combined with a spatial resolution of 100 nm, applied to a localized region of charge at the apex of a nanoscale metal tip induced by 30 fs 800 nm laser pulses at 50 kHz. These observations demonstrate real-space imaging of reversible processes such as tracking charge distributions is feasible whilst maintaining femtosecond resolution. Our findings could find application as a characterization method, which, depending on geometry could resolve tens of femtoseconds and tens of nanometres. Dynamically imaging electric and magnetic fields and charge distributions on sub-micron length scales opens new avenues of ultrafast dynamics. Furthermore, through the use of active compression, such pulses are an ideal seed for few-femtosecond to attosecond imaging applications which will access sub-optical cycle processes in nanoplasmonics., Comment: 14 pages, 8 figures, submitted to Structural Dynamics

Lineage tracing, the tracking of living cells as they move and divide, is a central problem in biological image analysis. Solutions, called lineage forests, are key to understanding how the structure of multicellular organisms emerges. We propose an integer linear program (ILP) whose feasible solutions define a decomposition of each image in a sequence into cells (segmentation), and a lineage forest of cells across images (tracing). Unlike previous formulations, we do not constrain the set of decompositions, except by contracting pixels to superpixels. The main challenge, as we show, is to enforce the morality of lineages, i.e., the constraint that cells do not merge. To enforce morality, we introduce path-cut inequalities. To find feasible solutions of the NP-hard ILP, with certified bounds to the global optimum, we define efficient separation procedures and apply these as part of a branch-and-cut algorithm. We show the effectiveness of this approach by analyzing feasible solutions for real microscopy data in terms of bounds and run-time, and by their weighted edit distance to ground truth lineage forests traced by humans., Comment: 11 pages

We consider the task of pixel-wise semantic segmentation given a small set of labeled training images. Among two of the most popular techniques to address this task are Decision Forests (DF) and Neural Networks (NN). In this work, we explore the relationship between two special forms of these techniques: stacked DFs (namely Auto-context) and deep Convolutional Neural Networks (ConvNet). Our main contribution is to show that Auto-context can be mapped to a deep ConvNet with novel architecture, and thereby trained end-to-end. This mapping can be used as an initialization of a deep ConvNet, enabling training even in the face of very limited amounts of training data. We also demonstrate an approximate mapping back from the refined ConvNet to a second stacked DF, with improved performance over the original. We experimentally verify that these mappings outperform stacked DFs for two different applications in computer vision and biology: Kinect-based body part labeling from depth images, and somite segmentation in microscopy images of developing zebrafish. Finally, we revisit the core mapping from a Decision Tree (DT) to a NN, and show that it is also possible to map a fuzzy DT, with sigmoidal split decisions, to a NN. This addresses multiple limitations of the previous mapping, and yields new insights into the popular Rectified Linear Unit (ReLU), and more recently proposed concatenated ReLU (CReLU), activation functions.