An interactive three dimensional approach to anatomical description—the jaw musculature of the Australian laughing kookaburra (Dacelo novaeguineae)

Avian and kingfisher cranial anatomy

A review of the literature indicates that descriptions of avian jaw anatomy are limited, with studies of only a few species, use of two-dimensional drawings, and complicated written descriptions. A major obstacle is the use of differing terminology (Holliday & Witmer, 2007). Papers that incorporate higher levels of detail include Lautenschlager, Bright and Rayfield’s description of the common buzzard jaw musculature (2013), Holliday and Witmer’s description and illustration of the albatross jaw musculature (2007), Elzanowski’s description of the tinamous jaw (1987), Zusi and Storer’s pied-billed grebe jaw study (1969), Richards and Bock’s comparison of Hawaiian honeycreeper feeding apparatus (1973), and Zweers’ description of mallard myography (1974). From these descriptions it is evident that avian jaw muscle anatomy differs in relative and absolute size of the various muscles, and in the position and extent of attachments. Consequently, we determined that the jaw musculature in kingfishers should not be reconstructed on the basis of anatomical descriptions of other families.

The jaw muscles of several species of kingfisher (though not Dacelo novaeguineae) have been described by Burton (1984); however his illustrations are difficult to interpret, with little information identifying which part of the skull he is referring to in the figures. Burton’s description did, however, reveal several variations in muscle origin and insertion areas between the kingfisher species he studied. This observation reinforced the need to document the kookaburra anatomy independently.

The lack of detailed published data on the kookaburra jaw musculature meant that firsthand anatomical information was required to accurately model the jaw system. This information was obtained through the dissection of the head of an adult laughing kookaburra (Dacelo novaeguineae). The kookaburra was CT scanned prior to dissection, and this scan data was used in conjunction with the dissection information to create a three-dimensional computer model of the skull and jaw muscles of D. novaeguineae.

A 3D approach to anatomical description

Anatomy is inherently three-dimensional, but traditional descriptions of anatomy use a combination of text and 2D picture modes. New technologies in data collection, assembly, visualisation and communication, are offering significant new opportunities for improving the study, documentation and publication of anatomy, particularly given the proliferation of 3D imaging capabilities.

The following description adopts and exemplifies a three-dimensional approach to displaying anatomical systems through the use of (1) three-dimensional Adobe Portable Document Format (3D PDF) figures, and (2) 3D-enabled web pages. The 3D PDF medium allows an interactive 3D model to be inserted into an electronic document in PDF format, meaning that specialised software is not required to view the model other than the freely available Adobe Reader (desktop version), and permanently embedding the figure directly in the document file. This method of interactive visualisation is becoming increasingly popular, with publications in biomedical science (Kumar et al., 2009; Selvam, Vasilyev & Wang, 2009), astrophysics (Barnes & Fluke, 2008), invertebrate (Baeumler, Haszprunar & Ruthensteiner, 2008; Neusser, Heß & Schrödl, 2009; Ziegler et al., 2008) and vertebrate anatomy (de Bakker et al., 2012; Holliday et al., 2013; Ruthensteiner & Heß, 2008; Lautenschlager, Bright & Rayfield, 2013) and paleontology (Knoll et al., 2012; Lautenschlager, 2013). In addition, free, open source software that provides a pathway for the generation of 3D PDF figures, especially for use in science, is now available (Barnes et al., 2013).

The 3D meshes produced for the 3D PDF were also used to generate a model available for viewing in web pages. Web-based 3D visualisation is becoming more common with the growing number of browsers supporting accelerated 2D and 3D graphics, especially via the WebGL standard (http://www.khronos.org/webgl); recent examples include applications in bioinformatics (Pettit & Marioni, 2013), neuroinformatics (Ginsburg et al., 2011), and surgical teaching (Birr et al., 2013). Like 3D PDF, web-based 3D visualisation enables the interactive manipulation of a 3D model without requiring specialised software other than a modern web browser.

Here we present a detailed description of the 3D osteology and jaw musculature of D. novaeguineae, and provide a summary of the avian jaw muscle synonymies. We aimed to demonstrate and exploit the potential of 3D viewing technology in the anatomical sciences by producing a 3D model that shows the position of the kookaburra jaw muscles on the skull relative to one another; the model can be rotated and superficial muscles hidden or made transparent to allow better visualisation of the deeper anatomy. While the 3D model allows for a detailed descriptive model in a single figure-frame, it is only available for viewing in digital format. For this reason, still images suitable for 2D print media are also included.

Specimen

A fresh, deceased adult laughing kookaburra (Dacelo novaeguineae), obtained from Australia Museum’s Ornithological collection (specimen no. AM BF2162) was used in the following description. This specimen had died in the care of wildlife rescuers as a result of an infection on the right elbow, although the head and skull did not show any obvious pathologies and thus were deemed to illustrate normative anatomy.

Data acquisition and processing

The kookaburra was scanned in a clinical computed tomography (CT) scanner (Toshiba Aquilion 64) at the Newcastle Calvary Mater Hospital, at 120 kV, 75 mA and using a slice thickness of 0.5 mm. The resulting DICOM images were imported into MIMICS v11 (MATERIALISE, Belgium) segmentation software, wherein the skull of the kookaburra was digitally separated from the soft tissue by combining a scalar threshold with manual selection tools; the high density of the bone compared to the surrounding soft tissue allowed for quick and simple segmentation of the skull by applying a threshold based on the grey scale pixel values representative of the bone. A 3D model of the cranium and mandible was then created from this segmentation. This digital skull model provided a scaffold to illustrate the muscle attachment points and muscle geometry determined in the dissection (Fig. 1).

Segmentation of the jaw muscles proved more difficult due to the small size of the muscles, the low resolution of the scanner and the use of CT (without staining for contrast) rather than MRI. While the soft tissue making up the kookaburra jaw muscles was visible in the CT scan data, and the bellies of the larger muscles could be differentiated and separated using MIMICS segmentation tools, the resolution of the scan was too low to accurately identify the smaller jaw muscles and the muscle attachment sites. For this reason, the soft tissue CT scan data was used alongside the dissection information to identify placement of the jaw muscles (Figs. 1 and 2).

The jaw muscles were dissected out beginning with the superficial layer. Each individual muscle was described using drawings, photographs and notes to illustrate its position, direction and attachment sites. Muscles of the hyoid and eye were not required for the biomechanical analysis and were not included in this study. The completed notes on kookaburra muscle anatomy were compared with previously published data on avian jaw musculature (Baumel et al., 1993; Burton, 1984; Elzanowski, 1987; Holliday & Witmer, 2007; Zusi & Storer, 1969) to determine the individual names of the jaw muscles observed. Muscle nomenclature used was based on Holliday & Witmer (2007) and Burton (1984). Osteological nomenclature (Fig. 3) was based on Baumel et al. (1993), Zusi & Storer (1969) and Ghetie et al. (1976).

In conjunction with the CT scan data of the specimen, the illustrated dissection notes were used to digitally recreate a three dimensional representation of the kookaburra jaw musculature on the specimen’s skull model (Fig. 2). This model was created using the CT editing software MIMICS to manually select areas of the CT image slices where the muscles were determined to occur. The CT slices were then joined in MIMICS to create a 3D model of each muscle which were individually exported as stereolithography (STL) format files.

3D image production

Web-based figure

Our web-based figure is based on the JavaScript three.js (http://threejs.org), public domain extensions to three.js (via http://learningthreejs.com), and E. Graether’s TrackballControls.js (http://egraether.com). Three.js provides a simple entry point to visualisation of 3D scientific data on the Web by supporting the loading of several common 3D model data formats, and rendering the 3D model using WebGL (http://khronos.org/webgl) where available, and otherwise the 2D Javascript canvas API for devices with no WebGL support. The OBJLoader.js module provides a ready-made importer for the ubiquitous Alias Wavefront OBJ format. Using these components, we developed a custom JavaScript-based web page that reads the identical model description file used for assembling the 3D PDF figure, and constructs a web-based 3D visualisation, coloured as for the 3D PDF version, and offering the same capability to interactively rotate and zoom the model, and to toggle the model components.

Musculoskeletal anatomy

Depressor muscles

1. M. depressor mandibulae (mDM) has a fleshy origin on the squamosal region on the back of the braincase, just posterior to the opening of the ear. It becomes thicker as it extends around the jaw joint to insert over most of the posterior surface of the mandible articular bone (Figs. 4G and 4H).

Methodological issues

With the increasing use of CT as a primary means of collecting anatomical data for use in 3D modelling, it is important to note the limitations of this imaging modality. Whilst it is possible to collect the data required for the 3D modelling of the bones from CT alone, our experience with the specimen used in this study is that incorporating detailed muscle anatomy into a 3D model requires a combination of CT imaging and dissection. In CT, contrast between bone and soft tissue is high, which maximises the accuracy of segmentation protocols (both automatic algorithms and manual image processing) for segmenting bones. However, in the absence of fixing protocols (e.g., iodine potassium iodide—‘Lugol’s iodine’), contrast between muscles and other soft tissues is low and it is difficult to identify individual muscle bellies on the basis of CT data alone; both the perimysium and intermuscular septa are inconsistently visible. An additional limitation is the comparatively low resolution of medical CT in imaging small animals; for the scans used in this study, maximum pixel resolution in each slice was 0.337 mm, which is a relatively large proportion of the maximum cranial diameter (44 mm), even for a large kingfisher species such as the kookaburra. Without the opportunity to use dissection to augment the 3D imaging of muscle anatomy we could not have produced the level of anatomical data presented here.

Whilst small-animal CT scanners (maximum field of view ∼100 mm) or micro-CT (∼20 mm) can deliver significantly higher resolution than clinical CT, these scanners are not as readily available as clinical CT scanners and we were not able to utilise one at the time of this study. These higher resolution CT scanners also cannot produce images that clearly differentiate muscle bellies without first preparing the specimen with an iodine stain, and require much longer scan times than a clinical CT.

It is possible that magnetic resonance imaging (MRI) can provide an alternative to dissection by producing high resolution images of the soft tissue, however MRI is not able to image bone well. A specimen can be both CT and MRI scanned and the resulting images co-registered to fit to the same 3D space co-ordinates, so that CT can be used to model the skeletal elements, while MRI can provide muscle detail. At the time of this study an MRI scanner and co-registration code was not available to us.

Nomenclature

Determining the correct names of particular muscles identified in the kookaburra based on previous avian jaw anatomy literature was a complicated process. For many taxonomic groups, anatomical descriptions have ranged across decades, languages, and authors; the issues we discovered with respect to non-standard nomenclature are not restricted to kingfishers. Muscle groups, particularly the M. adductor mandibulae externus and M. pterygoideus dorsalis muscles were separated into several sections in some studies (Burton, 1984; Richards & Bock, 1973), or treated as a whole in others (Holliday & Witmer, 2007). Within this study, questions remain regarding the extent to which some muscle bellies should be regarded as individual muscles, or be considered as a part of a larger muscle. Specifically, is M. adductor mandibulae externus rostralis temporalis post-orbital lobe part of one of the other M. adductor mandibulae externus muscles (with the only reference found of a separate muscle in this area recorded by Burton (1984); and should M. pterygoideus ventralis lateralis Part 2 and Part 3 be considered as separate muscles to M. pterygoideus ventralis lateralis Part 1?

Nomenclature used in this study was based mainly on Burton’s (1984) study as it includes kingfisher anatomy and uses similar terminology to Baumel et al.’s Nomina Anatomica Avium (1993). However, temporal muscle nomenclature was based upon Holliday & Witmer (2007). Baumel et al. (1993) and Burton (1984) each separate M. adductor mandibulae externus into three parts of mAME rostralis, mAME ventralis and mAME caudalis, but we found no clear evidence of 5 separate heads for this muscle in D. novaeguineae. Holliday and Witmer’s nomenclature for the external adductor complex was more similar to the anatomy visible in our specimen.

There is a lack of consistency in the way in which muscle names are written and abbreviated in text of previous literature. In an attempt to develop a standardised format for writing muscle names and abbreviations, we use the following format:

1. When writing the muscle name in text, we use an uppercase ‘M.’ (contraction for Musculus) at the start of the name, following Baumel et al.’s Nomina Anatomica Avium (1993).

2. When writing the muscle name in full within text, we italicise the name; hence, M. pseudotemporalis superficialis. The muscle name is more easily distinguished visually from the surrounding text, making the text easier to read.

3. When abbreviating the muscle (e.g., for figure labels), we use lower case for the ‘m’ and upper case for the abbreviation of the muscle name; hence, mPST. This format visually highlights the informative part of the label.

4. Parts of the muscle are written in lower case, hence mPSTs and mPSTp. This makes it easier to identify the muscle under discussion, avoiding the potential confusion that can arise between muscles in the external adductor complex if the abbreviation for the part is written in upper case—Holliday & Witmer (2007) use mAMEP and mAMP for the M. adductor mandibulae externus profundus and the M. adductor externus posterior respectively. We abbreviate these muscles as mAMEp and mAMP.

Table 1 lists the avian jaw muscle nomenclature that we have used in this paper, the respective abbreviations for use in labels, and the synonymies in selected literature. Note that some of these published nomenclatures were presented as anatomical illustrations rather than written descriptions, making it difficult to accurately determine exact muscle synonymy.

The role of 3D images in communicating anatomical information

Descriptive anatomy is the foundation for understanding functional anatomy. The recent surge in biomechanical studies of form-function questions in a range of taxa means descriptive anatomy remains as relevant as ever (Holliday et al., 2013). The current emphasis on 3D modelling approaches indicates that anatomical data should be collated and communicated in 3D, as much as is practically possible. It is certainly possible to communicate anatomical information to the level required for 3D models using text or 2D diagrams; the issue is simply that a great deal of text and/or a large number of 2D images are required. Conversely, a single 3D PDF figure or web visualisation can communicate an equivalent level of information with the user being able to interact with the model to gain finer understanding of structure.