Software Tools for Handling Magnetically Collected Ultra-thin Sections for Microscopy

Jelena Banjac
2019
Student project

Abstract

This report presents the software tools that will be available online to help users accurately segment ultra-thin sections of brain tissue in a large image to determine the section’s coordinates. In order to predict these coordinates, the goal was to use a state-of-art object instance segmentation framework called Masked Region-based Convolutional Neural Network (Masked R-CNN) on the dataset containing sections of brain tissue in the light microscopy images. The predicted coordinates of the sections will later be used for automated image acquisition in the high-resolution electron microscope. We will demonstrate that Masked R-CNN can be used to perform highly effective and efficient automatic segmentation of microscopy images containing the sections. The machine learning pipeline used will be explained in detail. In addition, we will show the results of how different parameters and settings of this pipeline were changing the performance. This semester's project was done in collaboration with the Center for Interdisciplinary Electron Microscopy (CIME) lab and Machine Learning and Optimization (MLO) lab at EPFL.

Official source

https://infoscience.epfl.ch/record/288188?ln=en

About this result

This page is automatically generated and may contain information that is not correct, complete, up-to-date, or relevant to your search query. The same applies to every other page on this website. Please make sure to verify the information with EPFL's official sources.

Related concepts

Related publications

Related publications (92)

Microstructure, properties and biocompatibility of the nitrided Ti-6Al-4V alloy for medical application

Philippe Buffat

Investigation of the microstructure, properties and biocompatibility of the Ti-6Al-4V alloy nitrided under glow discharge was performed. The microstructural analyses were carried out using light microscopy, X-ray diffraction, analytical scanning and transmission electron microscopy. Phase identifications and chemical composition of the layer and bulk material (substrate) were determined by electron diffraction and energy dispersive X-ray spectrometry. Atomic force microscopy was applied for layer surface topography measurements. Microhardness and Young's modulus measurements as well as frictional wear resistance and corrosion resistance tests were performed. The investigation revealed a clear correlation between the micro/nano structure and surface topography of the layer with its micromechanical, tribological and corrosion properties. In-vitro examinations of biofilm and cell behaviour show that the nitrided Ti-6Al-4V alloy exhibits good biocompatibility.

2006

This thesis contributes to the field of biomechanics of vascular wall. The focus is particularly on the microstructure of vascular elastin and collagen constituents and their contribution to the macroscopic mechanical behavior of the wall. The analysis is done in the framework of continuum mechanics. The work characterizes structural features of elastin and collagen fibers using microscopy techniques and introduces these features to constituent-based constitutive models. The models are applied to the experimental data, derived from inflation-extension tests, to predict the gross mechanical behavior of the tissue. The developed constitutive models could be further used to study in detail the mechanics of vascular tissue in health and disease. The thesis is presented in form of an introduction, four chapters (corresponding to four papers) and a conclusion. The introduction provides the motivation for this thesis as well as the background on vascular wall structure and mechanics. A brief description of the imaging techniques used is presented. Also, structural constitutive modeling of vascular wall is briefly discussed with particular attention to the modeling of elastin and collagen constituents. The first paper focuses on anisotropic properties of elastin in veins. We show that earlier constituent-based strain energy functions (SEFs), where elastin is modeled as an isotropic material, fail in describing accurately the tissue response to inflation-extension loading. We hypothesize that these shortcomings are partly due to unaccounted anisotropic properties of elastin. We extend the previously developed biomechanical model in our laboratory (Zulliger et al., 2004, J. Biomech., 37(7): 989-1000 (2004)) to account for elastin anisotropy and present an anisotropic strain energy function for elastin with one family of fibers in the longitudinal direction. The model is validated using experimental data from inflation-extension tests on rabbit facial veins. The tissue is tested under a fully relaxed smooth muscle state, for longitudinal stretch ratios ranging from 100% to 130% of the in vivo length. The model with the anisotropic elastin fits well the data for a wide range of longitudinal stretch ratios. The main finding of this paper is that the anisotropic description of elastin is required for a full 3-D characterization of the biomechanics of the venous wall. The second paper addresses the role of elastin in anisotropic properties of arteries with particular attention to the structural organization of elastin. A constituent-based model including an anisotropic elastin, with one family of fibers in the circumferential direction, is presented. Micro-structural imaging, based on electron microscopy techniques, is used to support this anisotropy. Inflationˆextension tests, on intact and elastase-treated arteries, provide a data set to validate the model and to study the effect of elastin removal. We show that the SEF, with an anisotropic elastin part, characterizes more accurately the mechanical properties of the arterial wall as compared to models with simply an isotropic elastin. Transmission electron microscopy (TEM) and serial block-face scanning electron microscopy (SBF-SEM) show interlamellar elastin fibers in the circumferential direction and therefore support the nature of the assumed anisotropy. The model predicts an earlier engagement of collagen in elastase-treated arteries compared to the intact arteries and thus suggests a clear functional interaction between the elastin and collagen component that is often neglected in constituent-based SEFs. The third paper presents a structural constitutive model of the vascular wall which integrates both waviness and orientational dispersion of collagen fibers. We extend the model of Zulliger et al., which already accounts for collagen waviness, to include orientational distribution of collagen. We study the effect of parameters related to the orientational distribution on macro-mechanical behavior of the tissue during inflation-extension tests. The model is further applied to the experimental data from rabbit facial veins. The model accurately fits the experimental data of veins, but it does not improve the quality of the fit compared to the one without dispersion. We show that the orientational dispersion of collagen fibers can be compensated by a less abrupt and shifted to higher strain collagen engagement pattern. This should be taken into consideration when the model is used to fit experimental data and model parameters are used to study structural modifications of collagen fiber network in physiology and disease. In the fourth paper, we measure and quantify the waviness and orientational distributions of collagen in rabbit carotids. Quantification of collagen orientation distributions at the zero stress state of arteries is needed to develop realistic and precise biomechanical models. Using the fluorescence collagen marker CNA38-OG488 and confocal laser scanning microscopy, we visualize collagen fibers in adventitia of rabbit common carotids ex vivo. To get the properties related to the zero stress sate, the arteries are cut open along their longitudinal axes. We use semi-automatic and automatic techniques to measure parameters related to the waviness and fiber orientation. We show that the straightness parameter (i.e. the ratio between the distances of endpoints of a fiber to the fiber length) is distributed with a beta distribution. The shape of the probability density distribution does not depend on the mean angle orientation of fibers. In addition, our measurements reveal four axially symmetric families of fibers with mean orientations of 0°, 90°, 43° and -43° and circular standard deviations of 40°, 47°, 37° and 37°, with respect to the axial direction, respectively. To the best of our knowledge, this is the first study focusing on structural properties of collagen in the zero stress state and quantifying fiber waviness. The results of this study can be used to develop more precise structural models of the adventitia including waviness and orientational dispersion of fibers. The conclusion section summarizes the main results of the thesis, presents the improvements made over previous studies, and proposes future perspectives of this work.

EPFL2010

Toxicity study of nanostructures

Lenke Horváth

The great achievement of nanotechnology is the controlled synthesis of a large variety of nanometer-size materials like needle-formed nanotubes, nanowires and two-dimensional graphene flakes. Due to their unique physico-chemical properties, these nanostructures are considered to be of great benefit for many applications in engineering, electronics, alternative energies and nanomedicine. Since the expectations for the improvement of our everyday life through engineered nanomaterials are high, there is rapid expansion of their manufacturing, which makes it likely that intentional and unintentional human and environmental exposure will increase in the near future. Consequently, the concern grows, related to their possible health hazards, as some of them strongly resemble asbestos. Motivated by this issue, we have investigated the in vitro acute cellular toxicity associated with four model nanomaterials: carbon nanotubes, boron nitride nanotubes, titanium dioxide nanofilaments and graphene oxide. This study focused on the toxic effect these nanomaterials had on cell types found in the respiratory system, where exposure to these materials is most prominent. These are lung epithelial cells, macrophages, and fibroblasts, but also other cell types like kidney cells were tested. The cytotoxicity was assessed by using MTT, DNA and FMCA assays, which measure different endpoints such as metabolic activity, cell proliferation and viable cell number. The cell death was determined by the Annexin V assay. We employed various microscopic techniques: light microscopy to reveal the morphological alterations associated with the nanomaterial toxicity at the cellular level; scanning electron microscopy to study the cell-nanomaterial interactions on the surface of the cell membrane; transmission electron microscopy to examine the uptake and subsequent localization of the nanomaterials within the cytosol and cell organelles. In addition, the generation of intracellular reactive oxygen species induced by graphene oxide was detected by the DCF assay. Last but not least, the pro-inflammatory potential and the biochemical perturbations in cells exposed to boron nitride nanotubes were investigated by Western blot and Synchrotron Infrared Microspectroscopy (SIRMS), respectively. All these techniques point to the adverse effects of the investigated nanostructures: i) The toxic effect of carbon nanotubes and carbon nanoparticles showed a time- and dose-dependent impairment in the metabolic activity of the cells characteristic for each cell type. Moreover, distinct morphological alterations typical for cell death were particularly apparent in macrophages. ii) The toxic potential of boron nitride nanotubes exhibited more pronounced adverse effects than carbon nanotubes. This was demonstrated by cell viability assays combined with cytopathological and biochemical analyses, showing induction of serious morphological changes, particularly in macrophages and fibroblasts, and biochemical processes characteristic for cell death. A higher acute toxicity was determined for boron nitride nanotubes when compared to crocidolite asbestos and their pro-inflammatory potential was demonstrated by the secretion of mature IL-1β cytokine in macrophages. iii) Titanium dioxide nanofilaments were also shown to impair the metabolic activity of the studied cells and induce morphological changes pointing to cell insult. iv) Graphene oxide exhibited a mild cytotoxic action in comparison to carbon nanotubes on epithelial cells and macrophages. The interaction of the nanomaterial with the cell surface generated reactive oxygen species during the initial phase of the exposure and transmission electron microscopy showed that graphene oxide flakes are taken up via the endocytic pathway. In summary, our findings highlight important physico-chemical parameters, which are important in relation to the toxic effect of nanomaterials. These are: i) the chemical composition; ii) the surface modification including functionalized groups and structural defects; iii) the geometry: length, diameter and tortuosity. In addition to the identified nanomaterial characteristics, we pinpoint that the target cells ́ response depends on their type, which is likely to be linked to their physiological function.

EPFL2012

Toxicity study of nanostructures

Lenke Horváth

EPFL2012

EPFL2010