Abstract. Automatic segmentation of the liver and its lesion is an important step towards deriving quantitative biomarkers for accurate clinical diagnosis and computer-aided decision support systems. This paper presents a method to automatically segment liver and lesions in CT abdomen images using cascaded fully convolutional neural networks (CFCNs) and dense 3D conditional random fields (CRFs). We train and cascade two FCNs for a combined segmentation of the liver and its lesions. In the first step, we train a FCN to segment the liver as ROI input for a second FCN. The second FCN solely segments lesions from the predicted liver ROIs of step 1. We refine the segmentations of the CFCN using a dense 3D CRF that accounts for both spatial coherence and appearance. CFCN models were trained in a 2-fold cross-validation on the abdominal CT dataset 3DIRCAD comprising 15 hepatic tumor volumes. Our results show that CFCN-based semantic liver and lesion segmentation achieves Dice scores over 94% for liver with computation times below 100s per volume. We experimentally demonstrate the robustness of the proposed method as a decision support system with a high accuracy and speed for usage in daily clinical routine.
Focussed Ion Beam (FIB) milling is a mainstay of nano-scale machining. By manipulating a tightly focussed beam of energetic ions, often gallium (Ga+), FIB can sculpt nanostructures via localised sputtering. This ability to cut solid matter on the nano-scale revolutionised sample preparation across the life, earth and materials sciences. Despite its widespread usage, detailed understanding of the FIB-induced structural damage, intrinsic to the technique, remains elusive. Here we examine the defects caused by FIB in initially pristine objects. Using Bragg Coherent X-ray Diffraction Imaging (BCDI), we are able to spatially-resolve the full lattice strain tensor in FIB-milled gold nano-crystals. We find that every use of FIB causes large lattice distortions. Even very low ion doses, typical of FIB imaging and previously thought negligible, have a dramatic effect. Our results are consistent with a damage microstructure dominated by vacancies, highlighting the importance of free-surfaces in determining which defects are retained. At larger ion fluences, used during FIB-milling, we observe an extended dislocation network that causes stresses far beyond the bulk tensile strength of gold. These observations provide new fundamental insight into the nature of the damage created and the defects that lead to a surprisingly inhomogeneous morphology.
Using X--ray micro--diffraction and surface acoustic wave spectroscopy, we measure lattice swelling and elastic modulus changes in a W--1%Re alloy after implantation with 3110 appm of helium. A fraction of a percent observed lattice expansion gives rise to an order of magnitude larger reduction in the surface acoustic wave velocity. A multiscale elasticity, molecular dynamics, and density functional theory model is applied to the interpretation of observations. The measured lattice swelling is consistent with the
Automatic segmentation of the liver and hepatic lesions is an important step towards deriving quantitative biomarkers for accurate clinical diagnosis and computer-aided decision support systems. This paper presents a method to automatically segment liver and lesions in CT and MRI abdomen images using cascaded fully convolutional neural networks (CFCNs) enabling the segmentation of large-scale medical trials and quantitative image analyses. We train and cascade two FCNs for the combined segmentation of the liver and its lesions. As a first step, we train an FCN to segment the liver as ROI input for a second FCN. The second FCN solely segments lesions within the predicted liver ROIs of step 1. CFCN models were trained on an abdominal CT dataset comprising 100 hepatic tumor volumes. Validation results on further datasets show that CFCN-based semantic liver and lesion segmentation achieves Dice scores over 94% for the liver with computation times below 100s per volume. We further experimentally demonstrate the robustness of the proposed method on 38 MRI liver tumor volumes and the public 3DIRCAD dataset.
Knowledge of mechanical and physical property evolution due to irradiation damage is essential for the development of future fission and fusion reactors. Ion-irradiation provides an excellent proxy for studying irradiation damage, allowing high damage doses without sample activation. Limited ion-penetration-depth means that only few-micron-thick damaged layers are produced. Substantial effort has been devoted to probing the mechanical properties of these thin implanted layers. Yet, whilst key to reactor design, their thermal transport properties remain largely unexplored due to a lack of suitable measurement techniques. Here we demonstrate non-contact thermal diffusivity measurements in ion-implanted tungsten for nuclear fusion armour. Alloying with transmutation elements and the interaction of retained gas with implantation-induced defects both lead to dramatic reductions in thermal diffusivity. These changes are well captured by our modelling approaches. Our observations have important implications for the design of future fusion power plants.
The use of Nye's dislocation tensor for calculating the density of geometrically necessary dislocations (GND) is widely adopted in the study of plastically deformed materials. The "curl" operation involved in finding the Nye tensor, while conceptually straightforward has been marred with inconsistencies and several different definitions are in use. For the three most common definitions, we show that their consistent application leads to the same result. To eliminate frequently encountered confusion, a summary of expressions for Nye's tensor in terms of elastic and plastic deformation gradient, and for both small and large deformations, is presented. A further question when estimating GND density concerns the optimization technique used to solve the under-determined set of equations linking Nye's tensor and GND density. A systematic comparison of the densities obtained by two widely used techniques, L 1 and L 2 minimisation, shows that both methods yield remarkably similar total GND densities. Thus the mathematically simpler, L 2 , may be preferred over L 1 except when information about the distribution of densities on specific slip systems is required. To illustrate this, we compare experimentally measured lattice distortions beneath nano-indents in pure tungsten, probed using 3D-resolved synchrotron X-ray micro-diffraction, with those predicted by 3D straingradient crystal plasticity finite element calculations. The results are in good agreement and show that the volumetric component of the elastic strain field has a surprisingly small effect on the determined Nye tensor. This is important for experimental techniques, such as micro-beam Laue measurements and HR-EBSD, where only the deviatoric strain component is measured.
Lattice defects play a key role in determining the properties of crystalline materials. Probing the 3D lattice strains that govern their interactions remains a challenge. Bragg Coherent Diffraction Imaging (BCDI) allows strain to be measured with nano-scale 3D resolution. However, it is currently limited to materials that form micro-crystals. Here we introduce a new technique that allows the manufacture of BCDI samples from bulk materials. Using tungsten as an example, we show that focussed ion beam (FIB) machining can be used to extract, from macroscopic crystals, micron-sized BCDI samples containing specific preselected defects. To interpret the experimental data, we develop a new displacementgradient-based analysis for multi-reflection BCDI. This allows accurate recovery of the full lattice strain tensor from samples containing multiple dislocations. These new capabilities open the door to BCDI as a microscopy tool for studying complex real-world materials.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.