arXiv subjects

Paul Müller

Publications and source records attributed to Paul Müller.

At least 19 recordsLinked to original sources

arXiv2026

Multi-label versus multi-class classification of blood cells and their aggregates in microfluidic channels

Deformability cytometry (DC) is a type of imaging flow cytometry, which uses a camera-equipped device to measure cellular stiffness in addition to other cellular properties at high throughput. Cellular properties such as area and elongation can identify cell types, but this requires prior knowledge of distinguishing properties and cannot be applied to clinically important cell aggregates. Using DC data, we evaluated conventional multi-class (MC) classification and introduced a multi-label (ML) approach for identifying blood cells and their aggregates. In particular, an ML classifier can simultaneously assign multiple cell-type labels to a single imaged event. We show that, unlike MC classification, ML classification can identify cell aggregates not represented in the training data. It also avoids the need for exhaustive, strictly defined aggregate labels, thereby simplifying and speeding up annotation. Since automated blood analyzers do not reliably analyze cell aggregates, our approach may help address this clinical gap.

cs.CV↗

arXiv2022

A New Hyperelastic Lookup Table for RT-DC

Real-time deformability cytometry (RT-DC) is an established method that quantifies features like size, shape, and stiffness for whole cell populations on a single-cell level in real time. To extract the cell stiffness, a lookup table (LUT) disentangles the experimentally derived steady state cell deformation and the projected area, yielding the Young's modulus. So far, two lookup tables exists, but are limited to simple linear material models and cylindrical channel geometries. Here, we present two new lookup tables for RT-DC based on a neo-Hookean hyperelastic material numerically derived by simulations based on the finite element method in square and cylindrical channel geometries. At the same time, we quantify the influence of the shear-thinning behaviour of the surrounding medium on the stationary deformation of cells in RT-DC and discuss the applicability and impact of the proposed LUTs regarding past and future RT-DC data analysis. Additionally, we provide insights about the cell strain and stresses, as well as the influence resulting from the rotational symmetric assumption on the cell deformation and volume estimation. The new lookup tables as well as the numerical cell shapes are made freely available.

q-bio.QM↗

arXiv2022

The Operating System of the Neuromorphic BrainScaleS-1 System

BrainScaleS-1 is a wafer-scale mixed-signal accelerated neuromorphic system targeted for research in the fields of computational neuroscience and beyond-von-Neumann computing. The BrainScaleS Operating System (BrainScaleS OS) is a software stack giving users the possibility to emulate networks described in the high-level network description language PyNN with minimal knowledge of the system. At the same time, expert usage is facilitated by allowing to hook into the system at any depth of the stack. We present operation and development methodologies implemented for the BrainScaleS-1 neuromorphic architecture and walk through the individual components of BrainScaleS OS constituting the software stack for BrainScaleS-1 platform operation.

cs.NE↗

arXiv2020

Accelerated physical emulation of Bayesian inference in spiking neural networks

The massively parallel nature of biological information processing plays an important role for its superiority to human-engineered computing devices. In particular, it may hold the key to overcoming the von Neumann bottleneck that limits contemporary computer architectures. Physical-model neuromorphic devices seek to replicate not only this inherent parallelism, but also aspects of its microscopic dynamics in analog circuits emulating neurons and synapses. However, these machines require network models that are not only adept at solving particular tasks, but that can also cope with the inherent imperfections of analog substrates. We present a spiking network model that performs Bayesian inference through sampling on the BrainScaleS neuromorphic platform, where we use it for generative and discriminative computations on visual data. By illustrating its functionality on this platform, we implicitly demonstrate its robustness to various substrate-specific distortive effects, as well as its accelerated capability for computation. These results showcase the advantages of brain-inspired physical computation and provide important building blocks for large-scale neuromorphic applications.

cs.NE↗

arXiv2018

Scanning Fluorescence Correlation Spectroscopy (SFCS) with a Scan Path Perpendicular to the Membrane Plane

Scanning fluorescence correlation spectroscopy (SFCS) with a scan path perpendicular to the membrane plane was introduced to measure diffusion and interactions of fluorescent components in free standing biomembranes. Using a confocal laser scanning microscope (CLSM) the open detection volume is moved laterally with kHz frequency through the membrane and the photon events are continuously recorded and stored in a file. While the accessory hardware requirements for a conventional CLSM are minimal, data evaluation can pose a bottleneck. The photon events must be assigned to each scan, in which the maximum signal intensities have to be detected, binned, and aligned between the scans, in order to derive the membrane related intensity fluctuations of one spot. Finally, this time-dependent signal must be correlated and evaluated by well known FCS model functions. Here we provide two platform independent, open source software tools (PyScanFCS and PyCorrFit) that allow to perform all of these steps and to establish perpendicular SFCS in its one- or two-focus as well as its single- or dual-colour modality.

q-bio.BM↗

arXiv2018

A Mixed-Signal Structured AdEx Neuron for Accelerated Neuromorphic Cores

Here we describe a multi-compartment neuron circuit based on the Adaptive-Exponential I&F (AdEx) model, developed for the second-generation BrainScaleS hardware. Based on an existing modular Leaky Integrate-and-Fire (LIF) architecture designed in 65 nm CMOS, the circuit features exponential spike generation, neuronal adaptation, inter-compartmental connections as well as a conductance-based reset. The design reproduces a diverse set of firing patterns observed in cortical pyramidal neurons. Further, it enables the emulation of sodium and calcium spikes, as well as N-Methyl-D-Aspartate (NMDA) plateau potentials known from apical and thin dendrites. We characterize the AdEx circuit extensions and exemplify how the interplay between passive and non-linear active signal processing enhances the computational capabilities of single (but structured) on-chip neurons.

q-bio.NC↗

arXiv2018

Response to Comment on "Cell nuclei have lower refractive index and mass density than cytoplasm"

In a recent study entitled "Cell nuclei have lower refractive index and mass density than cytoplasm", we provided strong evidence indicating that the nuclear refractive index (RI) is lower than the RI of the cytoplasm for several cell lines. In a complementary study in 2017, entitled "Is the nuclear refractive index lower than cytoplasm? Validation of phase measurements and implications for light scattering technologies", Steelman et al. observed a lower nuclear RI also for other cell lines and ruled out methodological error sources such as phase wrapping and scattering effects. Recently, Yurkin composed a comment on these 2 publications, entitled "How a phase image of a cell with nucleus refractive index smaller than that of the cytoplasm should look like?", putting into question the methods used for measuring the cellular and nuclear RI in the aforementioned publications by suggesting that a lower nuclear RI would produce a characteristic dip in the measured phase profile in situ. We point out the difficulty of identifying this dip in the presence of other cell organelles, noise, or blurring due to the imaging point spread function. Furthermore, we mitigate Yurkin's concerns regarding the ability of the simple-transmission approximation to compare cellular and nuclear RI by analyzing a set of phase images with a novel, scattering-based approach. We conclude that the absence of a characteristic dip in the measured phase profiles does not contradict the usage of the simple-transmission approximation for the determination of the average cellular or nuclear RI. Our response can be regarded as an addition to the response by Steelman, Eldridge and Wax. We kindly ask the reader to attend to their thorough ascertainment prior to reading our response.

q-bio.QM↗

arXiv2018

An Accelerated LIF Neuronal Network Array for a Large Scale Mixed-Signal Neuromorphic Architecture

We present an array of leaky integrate-and-fire (LIF) neuron circuits designed for the second-generation BrainScaleS mixed-signal 65-nm CMOS neuromorphic hardware. The neuronal array is embedded in the analog network core of a scaled-down prototype HICANN-DLS chip. Designed as continuous-time circuits, the neurons are highly tunable and reconfigurable elements with accelerated dynamics. Each neuron integrates input current from a multitude of incoming synapses and evokes a digital spike event output. The circuit offers a wide tuning range for synaptic and membrane time constants, as well as for refractory periods to cover a number of computational models. We elucidate our design methodology, underlying circuit design, calibration and measurement results from individual sub-circuits across multiple dies. The circuit dynamics match with the behavior of the LIF mathematical model. We further demonstrate a winner-take-all network on the prototype chip as a typical element of cortical processing.

q-bio.NC↗

arXiv2018

Statistical-Raman-Microscopy and Atomic-Force-Microscopy on Heterogeneous Graphene Obtained after Reduction of Graphene Oxide

Graphene oxide can be used as a precursor to graphene but the quality of graphene flakes is highly heterogeneous. Scanning-Raman-Microscopy (SRM) is used to characterize films of graphene derived from flakes of graphene oxide with an almost intact carbon framework (ai-GO). The defect density of these flakes is visualized in detail by analyzing the intensity and full-width at half-maximum of the most pronounced Raman peaks. In addition, we superimpose the SRM results with AFM images and correlate the spectroscopic results with the morphology. Furthermore, we use SRM technique to display the amount of defects in a film of graphene. Thus, an area of 250 x 250 {\my}m2 of graphene is probed with a step-size increment of 1 μm. We are able to visualize the position of graphene flakes, edges and the substrate. Finally, we alter parameters of measurement to analyze the quality of graphene fast and reliable. The described method can be used to probe and visualize the quality of graphene films.

physics.chem-ph↗

arXiv2018

Accurate evaluation of size and refractive index for spherical objects in quantitative phase imaging

Measuring the average refractive index (RI) of spherical objects, such as suspended cells, in quantitative phase imaging (QPI) requires a decoupling of RI and size from the QPI data. This has been commonly achieved by determining the object's radius with geometrical approaches, neglecting light-scattering. Here, we present a novel QPI fitting algorithm that reliably uncouples the RI using Mie theory and a semi-analytical, corrected Rytov approach. We assess the range of validity of this algorithm in silico and experimentally investigate various objects (oil and protein droplets, microgel beads, cells) and noise conditions. In addition, we provide important practical cues for future studies in cell biology.

q-bio.QM↗

arXiv2018

Refractive index measurements of single, spherical cells using digital holographic microscopy

In this chapter, we introduce digital holographic microscopy (DHM) as a marker-free method to determine the refractive index of single, spherical cells in suspension. The refractive index is a conclusive measure in a biological context. Cell conditions, such as differentiation or infection, are known to yield significant changes in the refractive index. Furthermore, the refractive index of biological tissue determines the way it interacts with light. Besides the biological relevance of this interaction in the retina, a lot of methods used in biology, including microscopy, rely on light-tissue or light-cell interactions. Hence, determining the refractive index of cells using DHM is valuable in many biological applications. This chapter covers the main topics which are important for the implementation of DHM: setup, sample preparation and analysis. First, the optical setup is described in detail including notes and suggestions for the implementation. Following that, a protocol for the sample and measurement preparation is explained. In the analysis section, an algorithm for the determination of the quantitative phase map is described. Subsequently, all intermediate steps for the calculation of the refractive index of suspended cells are presented, exploiting their spherical shape. In the last section, a discussion of possible extensions to the setup, further measurement configurations and additional analysis methods are given. Throughout this chapter, we describe a simple, robust, and thus easily reproducible implementation of DHM. The different possibilities for extensions show the diverse fields of application for this technique.

q-bio.QM↗

arXiv2018

Single-cell diffraction tomography with optofluidic rotation about a tilted axis

Optical diffraction tomography (ODT) is a tomographic technique that can be used to measure the three-dimensional (3D) refractive index distribution within living cells without the requirement of any marker. In principle, ODT can be regarded as a generalization of optical projection tomography which is equivalent to computerized tomography (CT). Both optical tomographic techniques require projection-phase images of cells measured at multiple angles. However, the reconstruction of the 3D refractive index distribution post-measurement differs for the two techniques. It is known that ODT yields better results than projection tomography, because it takes into account diffraction of the imaging light due to the refractive index structure of the sample. Here, we apply ODT to biological cells in a microfluidic chip which combines optical trapping and microfluidic flow to achieve an optofluidic single-cell rotation. In particular, we address the problem that arises when the trapped cell is not rotating about an axis perpendicular to the imaging plane, but instead about an arbitrarily tilted axis. In this paper we show that the 3D reconstruction can be improved by taking into account such a tilted rotational axis in the reconstruction process.

q-bio.QM↗

arXiv2017

3D correlative single-cell imaging utilizing fluorescence and refractive index tomography

Cells alter the path of light, a fact that leads to well-known aberrations in single cell or tissue imaging. Optical diffraction tomography (ODT) measures the biophysical property that causes these aberrations, the refractive index (RI). ODT is complementary to fluorescence imaging and does not require any markers. The present study introduces RI and fluorescence tomography with optofluidic rotation (RAFTOR) of suspended cells, quantifying the intracellular RI distribution and colocalizing it with fluorescence in 3D. The technique is validated with cell phantoms and used to confirm a lower nuclear RI for HL60 cells. Furthermore, the nuclear inversion of adult mouse photoreceptor cells is observed in the RI distribution. The applications shown confirm predictions of previous studies and illustrate the potential of RAFTOR to improve our understanding of cells and tissues.

q-bio.QM↗

arXiv2017

Pattern representation and recognition with accelerated analog neuromorphic systems

Despite being originally inspired by the central nervous system, artificial neural networks have diverged from their biological archetypes as they have been remodeled to fit particular tasks. In this paper, we review several possibilites to reverse map these architectures to biologically more realistic spiking networks with the aim of emulating them on fast, low-power neuromorphic hardware. Since many of these devices employ analog components, which cannot be perfectly controlled, finding ways to compensate for the resulting effects represents a key challenge. Here, we discuss three different strategies to address this problem: the addition of auxiliary network components for stabilizing activity, the utilization of inherently robust architectures and a training method for hardware-emulated networks that functions without perfect knowledge of the system's dynamics and parameters. For all three scenarios, we corroborate our theoretical considerations with experimental results on accelerated analog neuromorphic platforms.

q-bio.NC↗

arXiv2017

An Accelerated Analog Neuromorphic Hardware System Emulating NMDA- and Calcium-Based Non-Linear Dendrites

This paper presents an extension of the BrainScaleS accelerated analog neuromorphic hardware model. The scalable neuromorphic architecture is extended by the support for multi-compartment models and non-linear dendrites. These features are part of a \SI{65}{\nano\meter} prototype ASIC. It allows to emulate different spike types observed in cortical pyramidal neurons: NMDA plateau potentials, calcium and sodium spikes. By replicating some of the structures of these cells, they can be configured to perform coincidence detection within a single neuron. Built-in plasticity mechanisms can modify not only the synaptic weights, but also the dendritic synaptic composition to efficiently train large multi-compartment neurons. Transistor-level simulations demonstrate the functionality of the analog implementation and illustrate analogies to biological measurements.

cs.NE↗

arXiv2016

Lattice thermal expansion and anisotropic displacements in α-sulfur from diffraction experiments and first-principles theory

Thermal properties of solid-state materials are a fundamental topic of study with important practical implications. For example, anisotropic displacement parameters (ADPs) are routinely used in physics, chemistry, and crystallography to quantify the thermal motion of atoms in crystals. ADPs are commonly derived from diffraction experiments, but recent developments have also enabled their first-principles prediction using periodic density functional theory (DFT). Here, we combine experiments and dispersion-corrected DFT to quantify lattice thermal expansion and ADPs in crystalline α-sulfur (S8), a prototypical elemental solid that is controlled by the interplay of covalent and van der Waals interactions. We first report on single-crystal and powder X-ray diffraction (XRD) measurements that provide new and improved reference data from 10 K up to room temperature. We then use several popular dispersion-corrected DFT methods to predict vibrational and thermal properties of α-sulfur, including the anisotropic lattice thermal expansion. Hereafter, ADPs are derived in the commonly used harmonic approximation (in the computed zero-Kelvin structure) and also in the quasi-harmonic approximation (QHA) which takes the predicted lattice thermal expansion into account. At the PBE+D3(BJ) level, the latter leads to excellent agreement with experiments. Finally, more general implications of this study for realistic materials modeling at finite temperature are discussed.

cond-mat.mtrl-sci↗

arXiv2016

The Theory of Diffraction Tomography

Tomography is the three-dimensional reconstruction of an object from images taken at different angles. The term classical tomography is used, when the imaging beam travels in straight lines through the object. This assumption is valid for light with short wavelengths, for example in x-ray tomography. For classical tomography, a commonly used reconstruction method is the filtered back-projection algorithm which yields fast and stable object reconstructions. In the context of single-cell imaging, the back-projection algorithm has been used to investigate the cell structure or to quantify the refractive index distribution within single cells using light from the visible spectrum. Nevertheless, these approaches, commonly summarized as optical projection tomography, do not take into account diffraction. Diffraction tomography with the Rytov approximation resolves this issue. The explicit incorporation of the wave nature of light results in an enhanced reconstruction of the object's refractive index distribution. Here, we present a full literature review of diffraction tomography. We derive the theory starting from the wave equation and discuss its validity with the focus on applications for refractive index tomography. Furthermore, we derive the back-propagation algorithm, the diffraction-tomographic pendant to the back-projection algorithm, and describe its implementation in three dimensions. Finally, we showcase the application of the back-propagation algorithm to computer-generated scattering data. This review unifies the different notations in literature and gives a detailed description of the back-propagation algorithm, serving as a reliable basis for future work in the field of diffraction tomography.

q-bio.QM↗

arXiv2015

Characterization and Compensation of Network-Level Anomalies in Mixed-Signal Neuromorphic Modeling Platforms

Advancing the size and complexity of neural network models leads to an ever increasing demand for computational resources for their simulation. Neuromorphic devices offer a number of advantages over conventional computing architectures, such as high emulation speed or low power consumption, but this usually comes at the price of reduced configurability and precision. In this article, we investigate the consequences of several such factors that are common to neuromorphic devices, more specifically limited hardware resources, limited parameter configurability and parameter variations. Our final aim is to provide an array of methods for coping with such inevitable distortion mechanisms. As a platform for testing our proposed strategies, we use an executable system specification (ESS) of the BrainScaleS neuromorphic system, which has been designed as a universal emulation back-end for neuroscientific modeling. We address the most essential limitations of this device in detail and study their effects on three prototypical benchmark network models within a well-defined, systematic workflow. For each network model, we start by defining quantifiable functionality measures by which we then assess the effects of typical hardware-specific distortion mechanisms, both in idealized software simulations and on the ESS. For those effects that cause unacceptable deviations from the original network dynamics, we suggest generic compensation mechanisms and demonstrate their effectiveness. Both the suggested workflow and the investigated compensation mechanisms are largely back-end independent and do not require additional hardware configurability beyond the one required to emulate the benchmark networks in the first place. We hereby provide a generic methodological environment for configurable neuromorphic devices that are targeted at emulating large-scale, functional neural networks.

q-bio.NC↗