Opportunities and challenges for digital morphology

Abstract

Advances in digital data acquisition, analysis, and storage have revolutionized the work in many biological disciplines such as genomics, molecular phylogenetics, and structural biology, but have not yet found satisfactory acceptance in morphology. Improvements in non-invasive imaging and three-dimensional visualization techniques, however, permit high-throughput analyses also of whole biological specimens, including museum material. These developments pave the way towards a digital era in morphology. Using sea urchins (Echinodermata: Echinoidea), we provide examples illustrating the power of these techniques. However, remote visualization, the creation of a specialized database, and the implementation of standardized, world-wide accepted data deposition practices prior to publication are essential to cope with the foreseeable exponential increase in digital morphological data.

Figure 1

Virtual horizontal sections of 2D MRI scans of whole sea urchins (A, C) reveal distinct shapes of protractor muscles in Paracentrotus lividus (B) and Echinometra mathaei (D) (arrows). Magnetic resonance imaging was carried out in Berlin, Germany using a high-field MRI scanner with a 7 T super-conducting electromagnet (Bruker Biospin GmbH, Ettlingen, Germany). Image processing of the ~10 MB large raw image datasets was carried out using ImageJ 1.42q (NIH, Bethesda, USA) and its Volume Viewer plug-in on a standard office PC. The sea urchin (Echinodermata: Echinoidea) species shown here were collected in the wild (Paracentrotus lividus) or taken from a museum collection (Echinometra mathaei, NHM 1969.5.1.61-75).

Figure 2

Virtual horizontal μCT sections through Aristotle's lantern, the sea urchin feeding apparatus - phylogenetically representative selection of 16 out of almost 80 sea urchin species scanned by us using μCT (individual isotropic dataset resolutions ranging from 9 - 24 μm). The arrow depicts an asymmetrical tooth in Plesiodiadema indicum (Aspidodiadematidae). μCT was carried out at the GKSS outstation on the DESY site in Hamburg, Germany using an X-ray tube tomography system (GE Sensing & Inspection Technologies, Wunstorf, Germany). Image processing was carried out using Amira 5.2 (Visage Imaging GmbH, Berlin, Germany). The ~20-25 GB large raw image datasets were analyzed on a standard office PC using a remote visualization software environment (HP Remote Graphics Receiver 5.3.0: Hewlett-Packard Development Company, Palo Alto, USA). The office PC was connected via Internet (100 Mbit/s) to an HP XC/SVA visualization cluster (1× HP ProLiant DL785 with 32 AMD cores/256 GB RAM, and 8× HP xw8600 workstations, all equipped with dual NVIDIA Quadro FX5600 GPUs) at the Zuse Institute Berlin, Germany. Not to scale.

Figure 3

Various traditional and digital morphological visualization techniques, shown in an exemplary fashion using cidaroid sea urchins (Echinoidea: Cidaroida). A Habitus of a museum specimen of Eucidaris metularia (NHM 1969.5.1.15-40), aboral view. B Historical drawing of the internal anatomy of Cidaris cidaris, a closely related species - modified after Stewart [47]. C Volume rendering of the external anatomy of the specimen shown in A, based on a ~25 GB large μCT dataset with 13.91 μm isotropic resolution. D Virtual horizontal section of the μCT dataset at the level of Aristotle's lantern (see also Fig. 2). E Surface rendering of the external and internal anatomy of the specimen shown in A based on a ~100 MB large 3D MRI dataset with 81 μm isotropic resolution. F Virtual horizontal section of the MRI dataset at the level of Aristotle's lantern, digestive tract, and gonads. By clicking anywhere onto this figure, an interactive, partially labeled 3D model of the analyzed species will open (requires Adobe Acrobat Reader 8.0 or higher, see [10,43,44,46] for detailed information regarding 3D modeling and labeling). The museum specimen of Eucidaris metularia was photographed using a digital camera with 7.2 megapixel resolution (Exilim: Casio Computer Co., Tokyo, Japan). 3D visualization was carried out using volume rendering in VG Studio Max 2.0 (C) and threshold-based as well as manual segmentation followed by surface rendering in Amira 5.2 (E).