reading

We thank Ian Francis (Wiley-Blackwell) for initially approaching MDS to solicit this work, and Delia Sandford and Kelvin Matthews (Wiley-Blackwell) for their assistance in technical matters during its production. The following are thanked for providing suggestions to improve our first drafts: Alex Ball, Karl Bates, Robert Bradley, Jen Bright, Martin Dawson, Kate Dobson, Phil Donoghue, Peter Falkingham, Stephan Lautenschlager, Heinrich Mallison, Maria McNamara, Laura Porro, Paul Shearing and Alex Ziegler. Alan Spencer assisted with photography. We also thank the following for permission to re-use figures or for providing previously unpublished images: Karl Bates, Jason Dunlop, Cornelius Faber, Peter Falkingham, Nicolas Goudemand, Joachim Haug, Jason Hilton, Thomas Kleinteich, Heinrich Mallison, Andrew McNeil, Daniel Mietchen, Susanne Mueller, David Penney, Robert Scott, Leyla Seyfullah, David Wacey, Mark Wilson, Philip J. Withers, Florian Witzmann and Alex Ziegler. IR was funded by a NERC Postdoctoral Research Fellowship (NE/H015817/1). RG was funded by an 1851 Royal Commission Research Fellowship. Finally, we wish to thank our families, partners, friends and institutions for their forbearance with us over the long, cold winter of 2012/13, during which this book has taken shape.

About this Enhanced Edition

Registered Office
John Wiley & Sons, Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial Offices
9600 Garsington Road, Oxford, OX4 2DQ, UK
The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK
111 River Street, Hoboken, NJ 07030–5774, USA

For details of our global editorial offices, for customer services and for information about how to apply for permission to reuse the copyright material in this book please see our website at www.wiley.com/wiley-blackwell.

The right of the author to be identified as the author of this work has been asserted in accordance with the UK Copyright, Designs and Patents Act 1988.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher.

Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The publisher is not associated with any product or vendor mentioned in this book.

Limit of Liability/Disclaimer of Warranty: While the publisher and author(s) have used their best efforts in preparing this book, they make no representations or warranties with respect to the accuracy or completeness of the contents of this book and specifically disclaim any implied warranties of merchantability or fitness for a particular purpose. It is sold on the understanding that the publisher is not engaged in rendering professional services and neither the publisher nor the author shall be liable for damages arising herefrom. If professional advice or other expert assistance is required, the services of a competent professional should be sought.

Sutton, M. D. (Mark D.), author.
    Techniques for virtual palaeontology / Mark D. Sutton, Imran A. Rahman, Russell J. Garwood.
        pages cm
    Includes bibliographical references and index.
    ISBN 978-1-118-59113-0 (cloth)
1. Paleontological modeling. 2. Virtual reality in paleontology. 3. Paleontology–Data processing. I. Rahman, Imran A., author. II. Garwood, Russell J., author. III. Title.
    QE721.2.M63S88 2014
    560.285–dc23

Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.

Cover image: “Ventral view of the horseshoe crab Dibasterium durgae Briggs et al. 2012, from the Silurian-aged Herefordshire Lagerstätte, England. Model reconstructed through physical-optical tomography, manual registration, virtual preparation, isosurfacing and ray-tracing, using software packages SPIERS and Blender.” Briggs, D.E.G., Siveter, Derek J., Siveter, David J., Sutton, M.D., Garwood, R.J, & Legg, D. 2012. Silurian horseshoe crab illuminates the evolution of arthropod limbs. P.N.A.S. 109, 15702–15705.

New Analytical Methods in Earth and Environmental Science

Because of the plethora of analytical techniques now available, and the acceleration of technological advance, many earth scientists find it difficult to know where to turn for reliable information on the latest tools at their disposal, and may lack the expertise to assess the relative strengths or potential limitations of a particular technique. This new series addresses these difficulties, and by virtue of its comprehensive and up-to-date coverage, provides a trusted resource for researchers, advanced students and applied earth scientists wishing to familiarise themselves with emerging techniques in their field.

Authors will be encouraged to reach out beyond their immediate speciality to the wider earth science community, and to regularly update their contributions in the light of new developments.

Written by leading international figures, the volumes in the series will typically be 75–200 pages (30,000 to 60,000 words) in length – longer than a typical review article, but shorter than a normal book. Volumes in the series will deal with:

All titles in this series are available in a variety of full-colour, searchable e-book formats. Titles are also available in an enhanced e-book edition which may include additional features such as DOI linking, high resolution graphics and video.

Series Editors

Glossary

Acceleration voltage: The potential difference between cathode and anode in a non-synchrotron X-ray source, which helps define the X-ray energy.

Anaglyph stereo: A visualization approach where images from two closely-spaced viewing positions are coloured red and green/cyan and can hence be viewed stereographically (i.e. three-dimensionally) using red/green or red/cyan coloured glasses.

Analytical reconstruction methods: A group of computed tomography (q.v.) reconstruction algorithms which are computationally efficient, but prone to artefacts.

Artefacts: In any form of scanning, an artefact is a systematic discrepancy between the reconstructed tomogram (q.v.) and the sample’s true attenuation (q.v.) coefficients.

Attenuation: The loss of intensity of incident radiation as it passes through a medium. This typically results from absorption and/or scattering.

Attenuation coefficient: A measure of the strength of attenuation (q.v.) of a material per unit length.

Automatic first-pass registration: Using automated registration (q.v.) prior to manual registration (q.v.) to perform tomographic registration (q.v.).

Automatic registration: Using an automated algorithm to perform tomographic registration (q.v.).

Beam hardening: An artefact (q.v.) found in computed tomography scans with a polychromatic source (q.v.). Beam hardening results from differential attenuation of X-rays at different energies.

Body-size estimation: A method that uses computer models (q.v.) to reconstruct the dimensions of an extinct organism.

Boundary conditions: In a computer model, these are the loads and constraints that are applied to the model.

Bremsstrahlung: A continuous curve between the minimum and maximum X-ray energies in an X-ray source. Punctuated by characteristic radiation (q.v.).

Brilliance: A measure of synchrotron beam quality based on the number of photons emitted per second, the beam’s collimation, the source area, and the spectral distribution.

Calibration images: Also known as flat and dark fields, these are images used to correct projections (q.v.) prior to filtered backprojection (q.v.) in a computed tomography (q.v.) scan.

Characteristic radiation: Energy peaks in the bremsstrahlung (q.v.) found in lab sources, which are unique to any given target metal (q.v.).

Colour CT: An experimental technique in which a specialized detector capable of resolving an X-ray spectrum for each pixel is used to collect projections (q.v.).

Compton scattering: The collision of an X-ray photon and an electron, when the energy of the former greatly exceeds the latter. This results in partial photon energy loss (then scattering or deflection) and a free secondary electron.

Computational fluid dynamics (CFD): A method that uses computer models (q.v.) to simulate flow around an object.

Computed tomography (CT): A form of tomography (q.v.) where tomograms (q.v.) are recovered indirectly via computation from projections (q.v.) acquired with the aid of penetrating radiation, rather than direct imaging.

Computer model: A computer program used to approximate the behaviour of a real-world system of interest.

Confocal laser-scanning microscopy: A form of confocal microscopy (q.v.) that uses a laser beam to image the sample.

Confocal microscopy: A method for serial focusing (q.v.) that images focal planes (q.v.) by eliminating out-of-focus light.

Confocal Raman imagery: The combination of confocal microscopy (q.v.) and Raman spectroscopy (q.v.) to map the chemical structure of a sample in three dimensions.

Cropping: Reducing the size of an image or volume so that it only contains a region of interest (q.v.).

CT revolution: The rapid uptake of virtual palaeontology (q.v.) based on the increasing availability of X-ray computed tomography (q.v.) in the early 21st century.

Decimation: The process of algorithmically reducing the number of triangles in a triangle mesh (q.v.) while minimizing damage to the geometry of the represented object. See also Quadric error metric algorithms.

Dental microwear texture analysis: A technique for quantitatively analysing dental microwear (q.v.) through the measurement of three-dimensional surface texture variables.

Destructive tomography: Encompasses all forms of tomography (q.v.) in which all or part of the specimen is destroyed during the exposure of physical tomographic surfaces.

Direct point-cloud visualization: Techniques for the visualization of a three-dimensional point cloud as a two-dimensional image or sequence of images, without requiring the generation of a triangle mesh (q.v.).

Direct volume rendering: Visualizing a volume (q.v.) directly, that is without generating a triangle-mesh (q.v.).

Downsampling: Reduction in the resolution of a volume, normally by an integral factor.

Fiduciary markings: Markings in a tomographic dataset (q.v.), typically from physical-optical tomography (q.v.), which exist to aid the process of registration (q.v.).

Filament current: The current running through the filament of a non-synchrotron X-ray source, which helps define the X-ray energy.

Filtered back projection: A common algorithm for tomographic reconstruction (q.v.) in which projections are filtered and then superimposed at their acquisition angle over a square grid.

Finite-element analysis (FEA): A method that uses computer models (q.v.) to reconstruct stress, strain and deformation in a structure.

Flat field: A calibration image (q.v.) collected with the beam on but no specimen between source and detector.

Focused ion beam (FIB) tomography: A form of destructive tomography (q.v.) in which tomographic surfaces are physically exposed using a focused beam of ions and then imaged with the ion beam or a coupled-electron beam.

Fresnel Zone Plates: A means of focussing X-rays through a series of rings via diffraction.

Geometric morphometrics: A method that uses landmarks (q.v.) to quantitatively analyse form.

Hard X-rays: X-rays with wavelengths between 0.01 nm (124 keV) and 0.1 nm (12.4 keV).

Island removal: The process of algorithmically removing portions of a triangle mesh (q.v.) that are disconnected from the main mesh.

Isosurface: A mathematically defined surface calculated from a volume (q.v.), following points of a constant value. Isosurfaces are typically calculated using the marching cubes algorithm (q.v.).

Iterative reconstruction algorithms: A group of computed tomography (q.v.) reconstruction algorithms which are computationally inefficient, but less prone to artefacts (q.v.) than analytical reconstruction methods (q.v.).

K-alpha doublet: Characteristic radiation (q.v.) that represents electron transitions from a p-orbital of the L-shell to the vacated K-shell.

K-edge: A jump in the attenuation coefficient (q.v.) of an element when X-ray energy exceeds the binding energy of an atomic electron.

K-edge subtraction: A means of three-dimensional elemental mapping using scans taken just above and below a K-edge (q.v.) which are then subtracted.

Laminography: A form of X-ray tomography (q.v.) for highly anisotropic (i.e. flat) specimens.

Laser scanning: A surface-based technique (q.v.) that uses a laser beam to acquire numerous point co-ordinates for an object or area, which define a three-dimensional point cloud (q.v.).

Magnetic resonance imaging (MRI): A form of non-destructive tomography (q.v.) in which tomograms (q.v.) are produced by using magnetic fields to map the distribution of atomic nuclei in a sample. See also Nuclear magnetic resonance (NMR).

Manual registration: Performing tomographic registration (q.v.) manually, that is judging and adjusting correct registration for each tomogram (q.v.) by eye, normally with the aid of fiduciary markings (q.v.).

Marching cubes: An algorithm for the calculation of an isosurface (q.v.), which generates a surface in the form of a triangle mesh (q.v.).

Masks: Regions of a volume (q.v.) flagged in software as belonging to a particular structure or region; masks are normally specified to enable selective deletion or differential rendering (e.g. colouring) of items in a volume. Also called labels or segments in some software.

Material properties: In a computer model, these are the physical properties (e.g. density and elasticity) that are assigned to the different materials (e.g. bone) in the model.

Mechanical digitization: A surface-based technique (q.v.) that uses the position of a three-dimensional digitization stylus held against a specimen to digitize its form.

Monochromatic: Of an X-ray or other electromagnetic source – comprising a single wavelength, for example a synchrotron (q.v.) beam passed through a monochromator. See also Polychromatic.

Multibody dynamics analysis: A method that uses computer models (q.v.) to simulate the movements of interconnected objects.

Nanotomography (nano-CT): A form of X-ray computed tomography (q.v.) in which sub-micrometre voxel-sizes are attained.

Neutron tomography: A form of non-destructive tomography (q.v.) in which projections (q.v.) are produced by exposing a sample to a beam of neutrons and recording the resulting neutron attenuation, with subsequent computational analysis to create tomograms (q.v.).

Non-destructive tomography: Encompasses all forms of tomography (q.v.) which do not require physical removal of portions of the specimen. See also Destructive tomography.

Nuclear magnetic resonance (NMR): A physical phenomenon in which nuclei in a magnetic field absorb and re-emit electromagnetic radiation. The basis of Magnetic resonance imaging (q.v.).

Optical tomography: A form of non-destructive tomography (q.v.) in which tomograms (q.v.) are produced by shining light through a sample.

Peels: A method for providing a permanent record of a surface (e.g. a tomographic surface) by chemically impregnating a cellulose peel with material from the surface.

Phase-contrast tomography: A form of X-ray computed tomography (q.v.) which maps the phase shift caused by a sample (and hence its refractive index) rather than a beam’s attenuation.

Phase-shift laser scanning: A form of laser scanning (q.v.) that measures the distance between the scanner and the object by comparing the change in phase between the emitted and reflected laser light.

Photoelectric effect: A form of attenuation (q.v.) where an X-ray photon’s energy slightly exceeds the binding energy of an atomic electron liberating it as a photoelectron (q.v.).

Photogrammetry: A surface-based technique (q.v.) that uses multiple static images of a specimen, captured from different relative positions, to generate a three-dimensional virtual model.

Physical modelling: The production of a physical reproduction of a specimen from a virtual palaeontology dataset, or (historically) from analogue equivalents.

Physical-optical tomography: A form of destructive tomography (q.v.) in which tomograms (q.v.) are produced by physical exposure of surfaces and optical imaging or tracing.

Point cloud: A series of points in three-dimensional space used to define a surface; point clouds in virtual palaeontology (q.v.) are normally generated by surface-based techniques (q.v.), and each point may have a colour associated with it.

Polychromatic: Of an X-ray or other electromagnetic source – comprising multiple wavelengths, for example a lab-based X-ray source. See also Monochromatic.

Polygon mesh: A series of adjacent triangles or higher-order polygons defined by the three-dimensional co-ordinates of their vertices; triangle meshes (q.v.) are the form of polygon mesh normally used in virtual palaeontology (q.v.).

Projections: Two-dimensional images of a three-dimensional object in which incident radiation is differentially absorbed on the path through the sample. Traditional two-dimensional X-ray radiographs are examples of projections. Projections are used to create tomograms (q.v.) in computed tomography (q.v.).

Quadric error metric algorithms: Decimation (q.v.) algorithms that are capable of generating high-fidelity mesh simplification, but are often computationally expensive.

Raman spectroscopy: A technique for analysing the chemical composition of a sample.

Ray tracing: A computationally expensive but optically realistic means of generating a two-dimensional image from a three-dimensional geometry such as a triangle mesh (q.v.).

Region of interest: The portion of a volume that contains the specimen or portion of the specimen to be reconstructed.

Registered tomographic dataset: A tomographic dataset (q.v.) for which tomographic registration (q.v.) has been accomplished. Many scanning techniques (e.g. X-ray computed tomography, q.v.) produce datasets that are pre-registered, and do not require a discrete registration step.

Registration: The process of aligning data. The term is broad, and in virtual palaeontology (q.v.) can mean either (a) the aligning of two or more surface-based datasets or image stacks into a composite dataset, or (b) tomographic registration (q.v.).

Ring artefacts: An artefact (q.v.) which manifests as rings in a tomogram (q.v.), and results from variable sensitivities in detector elements.

Sensitivity analysis: An approach in computer modelling where input parameters are modified to evaluate their influence on the results and, hence, the uncertainty of the model.

Serial focusing: A form of optical tomography (q.v.) in which tomographic surfaces are captured by focusing a microscope at successive depths within a sample.

Serial grinding: A form of physical-optical tomography (q.v.) in which tomographic surfaces are physically exposed by grinding or lapping.

Serial sawing: A form of physical-optical tomography (q.v.) in which tomographic surfaces are physically exposed by saw cuts.

Serial slicing: A form of physical-optical tomography (q.v.) in which physical exposure of tomographic surfaces is accomplished by slicing with a blade.

Soft X-rays: X-rays with wavelengths between 0.1 nm (12.4 keV) and 10 nm (124 eV).

Spiral CT: The most common form of medical X-ray computed tomography (CT, q.v.) scanning in which a source and detector rotate around a sample moving in the z-direction.

Surface-based methods: Approaches to the gathering of data for virtual palaeontology (q.v.) that acquire topographic data on exposed surfaces of a specimen. All digitization techniques covered in this book are either surface based or use tomography (q.v.).

Synchrotron: A particle accelerator that uses charged particles (usually electrons) circulating in a storage ring to produce intense X-rays.

Synchrotron radiation X-ray tomographic microscopy (SRXTM): A high resolution form of synchrotron tomography (q.v.).

Synchrotron tomography: A form of X-ray computed tomography (q.v.) that employs a synchrotron (q.v.) as its X-ray source.

Target metal: The metal used to create X-rays by bombardment with electrons, in a laboratory X-ray source.

Three-dimensional printing: A type of physical modelling where a three-dimensional digital geometry, normally in the form of a triangle mesh (q.v.), is printed as a physical object.

Time-of-flight laser scanning: A form of laser scanning (q.v.) that measures the distance between the scanner and the object by calculating the time taken for the laser beam to return to the scanner.

Tomographic dataset: The dataset produced by tomography (q.v.) of a specimen.

Tomographic reconstruction: The creation of tomograms (q.v.) from projection (q.v.) data through a variety of algorithms such as filtered back projection (q.v.).

Tomographic registration: The process of aligning tomograms (q.v.) with respect to each other. Formally, this involves transforming images such that the vector offset in real space between any point (x,y) in tomograms n and n + 1 is perpendicular to the tomographic plane. See also Registration.

Tomography: The study of three-dimensional structures through a series of two-dimensional parallel slices through a specimen. All digitization techniques covered in this book either use tomography, or are surfaced based (q.v.).

Triangle mesh: A series of adjacent triangles defined by the three-dimensional co-ordinates of their vertices, used to provide a numerical representation of a surface. See also Polygon mesh.

Triangle-mesh rendering: The process of visualizing a three-dimensional triangle mesh (q.v.) as a two-dimensional image, normally using dedicated graphics hardware.

Triangulation-based laser scanning: A form of laser scanning (q.v.) that uses triangulation to determine point co-ordinates for a sample.

Validation: An approach in computer modelling where experimental data are compared with the model results to determine the accuracy of the model.

VAXML: A proposed standard for the storage and dissemination of virtual fossils (q.v.) in triangle-mesh (q.v.) format.

Vector surfacing: A method or group of methods for visualizing tomographic data, in which mathematically defined curves (splines) generated for each tomogram are ‘surfaced’ by some algorithm to produce a three-dimensional digital representation of the specimen.

Virtual fossil: A fossil reconstructed as a three-dimensional digital model.

Virtual palaeontology: The study of three-dimensional fossils through digital visualizations.

Virtual preparation: Manual or semi-automatic ‘tidying’ or marking-up (e.g. masking, q.v.) of a virtual palaeontology (q.v.) dataset so as to improve its utility.

Volume: A three-dimensional dataset in which data is held as an array of voxels (q.v.) representing the values of some property. A volume is the three-dimensional equivalent of a raster (bitmapped) image.

Volume-based reconstruction: A visualization of a tomographic dataset (q.v.) in which the data is treated as a volume (q.v.). Volume-based reconstruction can either use an isosurface (q.v.) or direct volume rendering (q.v.).

Voxel: A volume element, representing a value at a point within a volume (q.v.). A voxel is the three-dimensional equivalent of a pixel.

XANES tomography: An experimental technique allowing three-dimensional mapping of the chemical speciation of an element.

X-ray computed tomography: A form of non-destructive tomography (q.v.) in which projections (q.v.) are produced by exposing a sample to an X-ray beam and recording the resulting X-ray attenuation, with subsequent computational analysis to create tomograms (q.v.).

X-ray microtomography: A form of high-resolution X-ray computed tomography (q.v.) in which voxel-sizes smaller than ~50 μm are attained. See also Nanotomography (nano-CT).

Index

tomographic energy-dispersive diffraction imaging, see synchrotron energy-dispersive x-ray diffraction tomography

1 Introduction and History

Abstract: We define virtual palaeontology as the study of three-dimensional fossils through digital visualizations. This approach can be the only practical means of studying certain fossils, and also brings benefits of convenience, ease of dissemination, and amenability to dissection and mark-up. Associated techniques fundamentally divide into surface-based and tomographic; the latter is a more diverse category, sub-divided primarily into destructive and non-destructive approaches. The history of the techniques is outlined. A long history of physical-optical studies throughout the 20th century predates the true origin of virtual palaeontology in the 1980s. Subsequent development was driven primarily by advances in X-ray computed tomography and computational resources, but has also been supplemented by a range of other technologies.

1.1 Introduction

Virtual palaeontology is the study of fossils through interactive digital visualizations, or virtual fossils. This approach involves the use of cutting-edge imaging and computer technologies in order to gain new insights into fossils, thereby enhancing our understanding of the history of life. While virtual palaeontological techniques do exist for handling two-dimensional data (e.g. the virtual lighting approach of Hammer et al. 2002), for most palaeontologists the field is synonymous with the study of three-dimensionally preserved material, and the term is used in this context throughout this book. Note also that the manual construction of idealized virtual models of taxa (e.g. Haug et al. 2012, Fig. 11), while very much a worthwhile undertaking, is not included in the concept of virtual palaeontology followed herein.

The majority of fossils are three-dimensional objects. While compression of fossils onto a genuinely two-dimensional plane does of course occur (Figure 1.1a), it is the exception, and in most preservational scenarios at least an element of the original three-dimensionality is retained (Figure 1.1b). Three-dimensional preservation retains more morphological information than true two-dimensional modes, but typically this information is problematic to extract. Isolation methods, of which several exist, are one solution. Fossils may simply ‘drop out’ or be naturally washed out of rocks; wet-sieving of poorly consolidated sediments mimics this process. Specimens may also be extracted chemically, for example, by dissolving the matrix (e.g. Aldridge 1990). These approaches are effective where applicable, but are prone to losing associations between disarticulated or weakly connected parts of fossils, and to damaging delicate structures. Specimens can also be physically ‘prepared’ out using needles, drills or gas-jet powder abrasive tools (e.g. Whybrow and Lindsay 1990); while usually preserving associations, this approach may also damage delicate structures, scales poorly to small specimens, and cannot always expose all of a specimen. Finally, isolation of a fossil only provides access to its surface.

Figure 1.1 Dimensionality in fossils: (a) Completely two-dimensional graptolite fossils; genuinely two-dimensional fossils such as this are the exception. (b) A three-dimensionally preserved trilobite cephalon; most fossils exhibit at least partial three-dimensional preservation. Scale bars are 10 mm. Both specimens are from Lower Ordovician, Wales.

Correctly chosen, virtual palaeontological techniques can overcome many of the disadvantages of physical isolation methods, and bring many novel advantages too. Virtual specimens are typically more convenient to work with, requiring only a computer rather than expensive and lab-bound microscopes. They allow for virtual dissection and sectioning, where parts of the specimen can be isolated for clarity without fear of damage. They allow for mark-up, typically in the form of colour applied to discrete anatomical elements, which can greatly increase the ease of interpretation. They can be used as the basis for quantitative studies of functional morphology, such as finite-element analysis of stress and strain (e.g. Rayfield 2007), or hydrodynamic flow modelling (e.g. Shiino et al. 2009). Finally, as virtual specimens are simply computer files, they can be easily copied and disseminated to interested parties, facilitating collaborative analysis and publication.

Despite all these advantages, virtual palaeontology is not as widely used as it might be; one possible reason is that the techniques involved are perceived as ‘difficult’, and while there is no lack of technical detail available on individual techniques, no in-depth treatment and comparison of all available techniques exists, which can make the field intimidating to those entering it for the first time. This book aims to overcome this issue. It is intended to provide those interested in doing palaeontology through virtual methods, or in interpreting virtual data provided by other workers, with background theoretical knowledge and practical grounding. In particular, it aims to provide palaeontologists with the information they need to select an appropriate methodology for any particular study, to understand the pitfalls and limitations of each technique, and to provide suggestions for carrying out work with maximal efficiency. Theoretical concepts are covered with the intention of providing scientists with sufficient depth of understanding to develop and modify techniques, where appropriate.

Figure 1.2 Tomography. Three parallel and evenly spaced serial tomograms (1–3) through an idealized gastropod fossil, and the resultant tomographic dataset. Modified from Sutton (2008, Fig. 1). Reproduced with permission of The Royal Society of London.

Virtual palaeontological data-capture techniques can be divided most fundamentally into (a) tomographic (slice-based) approaches, and (b) surface-based approaches. Tomography is the study of three-dimensional structures through a series of two-dimensional parallel ‘slices’ through a specimen (Figure 1.2). In tomography, an individual slice-image is termed a tomogram, and a complete set of tomograms is (herein) termed a tomographic dataset. Any device capable of producing tomograms is a tomograph. Note that while the definition of tomography given above is the original one (derivation is from the Greek tomos – section, cut, slice and graphein – writing, imaging, study), in recent years this term has often been restricted to techniques where virtual tomograms are computed indirectly from projections, rather than imaged in a direct way. However, we consider our broader definition to be both more historically accurate and more useful, with all such techniques sharing much in common, especially with regards to reconstruction methodology. The term we prefer for tomographic techniques based on computation of virtual tomograms is computed tomography. Tomography can be divided into (a) destructive and (b) non-destructive (scanning) methodologies. The former include the long-established techniques of serial grinding, sawing, slicing, etc. (here grouped together as physical-optical tomography, Section 2.2), together with focused ion-beam tomography (Section 2.3). Non-destructive tomographic techniques are diverse, and include the many variants of X-ray computed tomography or CT (Section 3.2), neutron tomography (Section 3.3), magnetic resonance imaging (Section 3.4), and optical tomography (serial focusing – Section 3.5). Surface-based techniques are those where the geometry of an external surface is digitized in some fashion; they include laser scanning (Section 4.2), photogrammetry (Section 4.3) and mechanical digitization (Section 4.4). This book concludes with an examination of the techniques and software available for specimen reconstruction and study (Chapter 5), a review of the applications of virtual models beyond simple visualization (Chapter 6), and a final overview and consideration of possible future developments (Chapter 7).

1.2 Historical Development

Virtual Palaeontology, in the sense used in this book, began in the early 1980s when the emerging medical technology of X-ray computed tomography was first applied to vertebrate fossils. The power of tomography to document and reconstruct three-dimensionally preserved material has, however, long been recognized, and modern techniques have a lengthy prehistory of physical-optical tomography (sensu Section 2.2), combined in some cases with non-computerized visualization techniques.

1.2.1 Physical-Optical Tomography in the 20th Century

Palaeontological tomography was introduced in the first years of the 20th century by the eccentric Oxford polymath William J. Sollas, who noted the utility of serial sectioning in biology and realized that serial grinding could provide similar datasets from palaeontological material. His method (Sollas 1903) utilized a custom-made serial-grinding tomograph capable of operating at 25 µm intervals, photography of exposed surfaces, and manual tracing from glass photographic plates. Sollas applied this approach with considerable zeal to a wide range of fossil material, and was able to demonstrate the fundamental utility and resolving power of tomography to a broad audience. He also described (Sollas 1903) a physical-model visualization technique in which tomograms were traced onto thin layers of beeswax which could then be cut to reproduce the original slice, stacked together and weakly heated to fuse them into a cohesive model. A quick-and-dirty approach to model-making, using glued cardboard slices rather than fused wax, was also in early use; while documentation is lacking, this appears also to be traceable back to Sollas.

Sollas was primarily a vertebrate palaeontologist, and it was in this field that his methods first became widely accepted, most notably in the seminal studies of Stensiö (1927) on the cranial anatomy of Devonian fish. From the mid-20th century, however, serial grinding became a well-established palaeontological technique, and was applied to a very wide range of fossil vertebrates, invertebrates, and plants. These applications are far too numerous to cite, but an excellent example of a group whose students embraced it with some degree of fervour is the Brachiopoda. Brachiopods are often preserved three-dimensionally and articulated with valves firmly closed, concealing taxonomically and palaeobiologically informative internal structures such as lophophore supports; following the pioneering work of Muir-Wood (1934), the use of manually traced serial sections to document these structures has become almost ubiquitous.

A range of serial-grinding tomographs, for the most part custom-built devices, have been used since Sollas’s work (e.g. Simpson 1933; Croft 1950; Ager 1965; Sutton et al. 2001b); these have varied substantially in complexity, degree of automation, maximum specimen size and minimum grind-interval, although none have substantially improved on the original machine in the latter respect. Two major variants on the technique have also been important, both responses to the destructive nature of serial grinding. Firstly, acetate peels (Walton 1928, see Galtier and Phillips 1999 for a more modern treatment) have been widely adopted as a means of data capture, especially but not exclusively in palaeobotany. Peels provide a permanent record of mineralogy and can be combined with staining techniques to increase contrast between certain types of material; they have thus been viewed as superior to mere photography of surfaces. Peels do, however, bring a peculiar set of problems of their own (see Section 2.2.2.3), and their use has unfortunately rendered many historical datasets ill-suited to modern visualization methods. Secondly, serial sawing using fine annular or diamond-wire saws (Kermack 1970) became popular for larger fossils such as vertebrates in the latter quarter of the 20th century, as it allowed retention of original material (albeit at the cost of an increase in minimum tomogram spacing).

While physical-optical tomography was commonplace in the 20th century, physical model-making noticeably fell out of favour, considered perhaps to be too laborious and of doubtful scientific utility. Students of particular groups (e.g. brachiopods) became sufficiently familiar with tomograms to be able to integrate them into mentally conceived three-dimensional representations, and the potential benefits of being able to directly communicate these visualizations beyond the cognoscenti were arguably overlooked. Reconstructions from tomographic data, where published, typically took the form of idealized pictorial or diagrammatic representations from such mentally assembled models; while aesthetically pleasing and often gratifyingly simplified (for an example from palaeobotany see the cupule reconstructions of Long 1960), this form of reconstruction lacked objectivity. That said, physical models were undoubtedly difficult to assemble, fragile, difficult to transport and hard to work with; while some workers continued to use them (e.g. Jefferies and Lewis 1978), truly effective visualization was not eventually achieved until the advent of interactive virtual fossils at the start of the 21st century.

1.2.2 The CT Revolution

Tomography in palaeontology has seen an enormous rise in uptake in recent years – Figure 1.3 provides a graphical representation of the use of the term ‘tomography’ in the palaeontological literature. It shows a fairly steady rise for the 30 years between 1975 and 2005 (the drop in 1996 is probably a methodological artefact of the way the literature was indexed), followed by an upswing that is, to say the least, eye-catching. This phenomenon is, for the most part, a result of the increasing availability and popularity of X-ray CT, and we refer to it herein as the CT revolution. X-ray computed axial tomography (CT or CAT scanning) is a technology that arose as an advanced form of medical radiography in the early 1970s, taking advantage of the increasing availability of computing power together with technical and algorithmic advances. CT, its history and its derivatives are described in more detail in Section 3.2non-destructive tomographyArchaeopteryxX-ray microtomographywww.digimorph.org

Acknowledgements

About this Enhanced Edition

New Analytical Methods in Earth and Environmental Science

Series Editors

Glossary

Index

1 Introduction and History

1.1 Introduction

1.2 Historical Development

1.2.1 Physical-Optical Tomography in the 20th Century

1.2.2 The CT Revolution