Assessment of the accuracy of dental enamel thickness measurements using microfocal X-ray computed tomography
Article first published online: 6 FEB 2006
Copyright © 2006 Wiley-Liss, Inc.
The Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology
Volume 288A, Issue 3, pages 263–275, March 2006
How to Cite
Olejniczak, A. J. and Grine, F. E. (2006), Assessment of the accuracy of dental enamel thickness measurements using microfocal X-ray computed tomography. Anat. Rec., 288A: 263–275. doi: 10.1002/ar.a.20307
- Issue published online: 20 FEB 2006
- Article first published online: 6 FEB 2006
- Manuscript Accepted: 17 NOV 2005
- Manuscript Received: 25 APR 2005
- microfocal X-ray computed tomography;
- enamel thickness;
- measurement accuracy;
- dental enamel;
Tooth enamel thickness has long been an important character in studies of primate and especially hominin phylogeny, taxonomy, and adaptation. Current methods for accurately assessing enamel thickness involve the physical sectioning of teeth, because measurements of enamel thickness using some radiographic techniques are unreliable. However, because destructive methods limit sample sizes and access to important fossil specimens, it is desirable that they be replaced with nondestructive techniques. Although microfocal X-ray computed tomography (mCT) has been used recently in studies of enamel thickness, the accuracy of this technique has yet to be established. The present research compares physical sections to computer-generated mCT sections of teeth from a variety of primate and nonprimate, recent and fossil taxa to examine whether enamel thickness, tooth size, and diagenetic remineralization (fossilization) impact the ability of mCT to measure enamel thickness accurately. Results indicate that recent teeth of varying size and thickness are clearly and accurately depicted in mCT scans, with measurements from nearly identical planes in physical and mCT sections differing by 3–5%. A fossil papionin molar (ca. 2 Myr) was also accurately measured using mCT scans, although thinner enamel in much older therapsid (ca. 263–241 Myr) teeth could not be distinguished from dentine. mCT is thus an accurate technique for measuring enamel thickness in recent taxa, although heavily mineralized teeth pose an obstacle to the ability of mCT to distinguish dental tissues. Moreover, absolutely thin enamel (less than ∼ 0.10 mm) is difficult to resolve adequately in raw mCT images based on pixel values alone. Therefore, caution must be exercised in the application of mCT to the study of fossilized teeth. © 2006 Wiley-Liss, Inc.
Tooth enamel thickness has long been considered to be of importance in the interpretation of human evolution because of its purported taxonomic and phylogenetic value (Miller,1918; Simons and Pilbeam,1972; Kay,1981; Gantt,1983; Martin,1985; Beynon and Wood,1986; Grine and Martin,1988; Zilberman et al.,1990; Beynon et al.,1991; Macho and Thackeray,1992; Zilberman and Smith,1992a,1992b,1994; Zilberman et al.,1992; Macho and Berner,1993; Molnar et al.,1993; Smith and Zilberman,1994; White et al.,1994; Strait et al.,1997; Shellis et al.,1998; Haile-Selassie,2001; Senut et al.,2001; Strait and Grine,2001,2004; Brunet et al.,2002; Wolpoff et al.,2002; Martin et al.,2003; Haile-Selassie et al.,2004). Enamel thickness is also thought to be of significance for the interpretation of occlusal loading regimens (Schillingburg and Grace,1973; Molnar and Gantt,1977; Kay,1981; Grine and Martin,1988; Khera et al.,1990; Macho and Berner,1993; Dumont,1995; Spears and Macho,1998; Macho and Spears,1999; Schwartz,2000a,2000b,; Martin et al.,2003).
Physical cross-sections of molar teeth have been used to measure enamel thickness by a number of workers (e.g., Molnar and Gantt,1977; Martin,1985; Beynon and Wood,1986; Grine and Martin,1988; Macho and Berner,1993; Dumont,1995; Schwartz,2000b; Grine,2002,2004,2005; Grine et al.,2005). Physical sections provide an accurate portrayal of enamel thickness in a particular plane of section, but problems with specimen orientation (i.e., section obliquity) may render some of these data less than ideal. Moreover, because mechanical sectioning is destructive, even when the cut is as narrow as 70 μm (Grine and Martin,1988), it is difficult to obtain adequate samples for statistical comparison, especially when the taxa under consideration are rare, endangered, or extinct.
Alternative techniques have been developed and employed in response to the problem of accurately reconstructing enamel cap volumes from physical sections, although some of these entail the destruction of the entire enamel cap (Kono-Takeuchi et al.,1997; Kono et al.,2002). Noninvasive methods such as ultrasonic imaging or acoustic microscopy (Baum et al.,1963; Lees,1968,1971; Lees and Barber,1968; Peck et al.,1989; Yang,1991,1996; Mezava et al.,1999; Maev et al.,2000,2002) and radiography have been employed to quantify tooth enamel thickness. Although acoustic microscopy has been used in several clinical studies of enamel and dentine, the accuracy of the quantitative data obtained has yet to be established. Various radiographic techniques have been employed extensively in the measurement of tooth enamel thickness to address clinical as well as evolutionary questions, including flat-plane radiography, computed tomography, and, most recently, microfocal X-ray computed tomography (mCT).
Lateral flat-plane radiographs have been used to measure enamel thickness in living humans. Standard bite-wing radiographs of dental patients have been used to study the effect of sex-linked genes on tooth size (Alvesalo and Tammisalo,1981; Alvesalo,1985; Alvesalo et al.,1985,1987), and they also have been used to assess the effects of sex and ethnic background on tooth crown components (Stroud et al.,1994,1998; Zilberman and Smith,1994; Harris and Hicks,1998; Harris et al.,1999). Lateral flat-plane radiographs also have been used to measure enamel thickness in fossil hominin and hominoid specimens (Sperber,1985,1986; Nagatoshi,1990; Zilberman et al.,1990; Zilberman and Smith,1992b,1992a; Molnar et al.,1993; Faerman et al.,1994; Smith and Zilberman,1994). These studies utilized the parallel-film technique, in which the film is oriented parallel to the major axis of the tooth, and the X-ray beam is directed perpendicular to an imaginary plane that bisects the vertical and mesiodistal axes of the tooth and the plane of the radiographic plate (Gron,1960). Some studies have recorded linear measurements from lateral radiographs, while others have measured the projected area of the enamel cap. It has been shown, however, that measurements of enamel thickness from standard lateral radiographs are unlikely to reflect accurately the true thickness as defined by physical sections through the crown (Grine et al.,2001). Thus, while this type of X-ray may provide a qualitative impression of whether a tooth has relatively thin enamel (e.g., human deciduous molars) or relatively thick enamel (e.g., human permanent molars), measurements derived from standard lateral radiographic images are unlikely to reflect that thickness accurately.
Computed tomography (CT) has also been utilized to quantify enamel thickness in recent and fossil hominin teeth (e.g., Zonneveld and Wind,1985; Conroy,1991; Conroy and Vanier,1991; Macho and Thackeray,1992; Conroy et al.,1995; Schwartz et al.,1998). With the exception of the study by Schwartz et al. (1998), however, these analyses have employed standard CT scans, a technique that has been demonstrated to be grossly inaccurate in the measurement of enamel thickness (Grine,1991). Standard CT images generally exaggerate enamel thickness compared to values obtained from mechanical sections (Grine,1991). There are several reasons for this, including beam hardening artifacts, partial volume effect, and the improbability of obtaining a CT slice that traverses the tips of both dentine horns (Grine,1991). Moreover, Spoor et al. (1993) have shown that accurate linear measurements of enamel can be obtained with CT scans only if the half maximum height (HMH) values are calculated between adjacent CT number levels at both the air-enamel and enamel-dentine boundaries. Failure to image at the proper HMH CT values can result in measurement errors in excess of 1.0 mm (Spoor et al.,1993). Even when the correct HMH CT adjustment is used, thin enamel cannot be measured accurately (its thickness will be exaggerated) due to limited spatial resolution of the scanner (Spoor et al.,1993). These factors render suspect the results of enamel thickness measurements reported in a number of studies that did not take them into account (e.g., Zonneveld and Wind,1985; Conroy,1991; Conroy and Vanier,1991; Macho and Thackeray,1992; Conroy et al.,1995).
Technical advances in computer-assisted radiographic techniques have resulted in the application of microfocal X-ray computed tomography as a nondestructive technique for the measurement of enamel thickness (Chaimanee et al.,2003; Gantt et al.,2003; Kono,2004; Olejniczak and Grine,2005). The intricate structure of cancellous bone has been extensively studied using mCT (Kuhn et al.,1990; Bonse et al.,1994; Müller et al.,1996a,b; Rüegsegger et al.,1996; Müller and Rüegsegger,1997; Kappelman,1998; Kinney et al.,1998; Fajardo and Müller,2001; MacLatchy and Müller,2002; Müller,2002,2003), and this technology has even been applied successfully to the visualization of the brain endocast and vestibular apparatus of Mesozoic pterosaurs (Witmer et al.,2003).
A number of studies have examined the repeatability of measurements (and inter- and intraindividual errors in locating landmarks) from mCT images of trabecular bone (e.g., Durand and Rüegsegger,1991; Balto et al.,2000), tooth root canals (Peters et al.,2000), and tooth enamel (Avishai et al.,2004). To date, however, the only studies that have attempted to assess the accuracy of measurements derived from high-resolution X-ray computed tomography have been conducted with reference to trabecular bone configuration (Kuhn et al.,1990; Müller et al.,1996a,b,1998; Hildebrand et al.,1999; Fajardo et al.,2002), periapical and periradicular alveolar bone destruction (Balto et al.,2000; von Stechow et al.,2003), the diameters of tooth root canals (Rhodes et al.,1999), and cortical bone porosity (Cooper et al.,2004). We are unaware of any study that has addressed the issue of measurement accuracy in the mCT assessment of dental hard tissues.
Kuhn et al. (1990), Müller et al. (1996a,b,1998), Hildebrand et al. (1999), and Fajardo et al. (2002) observed negligible differences between measurements derived from mCT images and those obtained using traditional histomorphometric or stereological techniques applied to histological thin sections. Percentage differences between the mCT and histological values reported by these studies ranged between lows of 0.88–1.17 (Müller et al.,1996a,b) and highs of 2.5–6.1 (Müller et al.,1996a,b), depending upon the type of physical section examined. Balto et al. (2000) and von Stechow et al. (2003) also reported negligible differences between measurements obtained from mCT and histological sections of periapical and periradicular bone. Rhodes et al. (1999) similarly reported very high correlations between the values recorded for the diameters of tooth root canals from mCT images and physical sections. Cooper et al. (2004) compared measurements obtained by mCT to those recorded from microradiographs of cortical bone, and while the two methods produced similarities, relatively poor results were obtained for measurements of pore density. Unfortunately, it is not possible to ascertain the source of this discrepancy or its deviation from values derived from physical sections because histological preparations were not measured directly. Fajardo et al. (2002) concluded that mCT can accurately reconstruct the complex architecture of trabecular bone, although slight alterations in threshold values greatly affected the morphometric data.
The only mCT studies that have been directed explicitly at the resolution of dental hard tissues (e.g., Avashai et al.,2004; McErlain et al.,2004) have not involved comparisons of measurements derived from mCT and physical sections. McErlain et al. (2004) examined a single tooth from a 420-year-old Late Iron Age site in Cambodia and simply commented that they were clearly able to differentiate visually the hard tissue components, which were displayed with “exceptional clarity.” The study by Avishai et al. (2004) examined individual error rates in locating landmarks (and resultant measurements) in a series of tooth germs derived from archeological contexts.
Thus, although mCT is a potentially powerful tool by which to obtain information on enamel thickness in a nondestructive manner, its efficacy for this purpose has yet to be demonstrated. It is necessary to establish the validity of enamel thickness measurements obtained using mCT before these data are employed to answer questions of clinical or evolutionary interest. The purpose of the present study was to establish the accuracy of high-resolution mCT for the measurement of enamel thickness in teeth of differing size, enamel thickness, and degree of mineralization. Toward this end, we compared measurements obtained from mCT scans that had not undergone any image artifact corrections (as this may impact measurements) to those obtained from physical sections of the same specimens. In the first instance, we measured mCT and histological sections using standard techniques (i.e., measuring printed images using a digitizing tablet) in order to make our results comparable to those of previously published studies. As mCT studies become more common, direct measurement by means of computer software will also become commonplace. We therefore also compared our measurements obtained by digitization to those taken directly on the three-dimensional models in order to assess the comparability of these protocols.
MATERIALS AND METHODS
Study Sample Composition
Twelve isolated teeth that varied in size, absolute enamel thickness, and degree of mineralization were chosen for study (Table 1). One of the modern human molars (Fig. 1) was embedded in a cylinder of methyl methacrylate. Although the absolute sizes of the teeth differed somewhat, all were chosen for their ability to be accommodated within the specimen tubes of the mCT system that was used (μCT 40; Scanco Medical, Switzerland). These tubes range from 13 to 35 mm in diameter. Beyond this, the extant primate and sauropsid specimens were selected on the basis of published values (Dauphin,1987; Sato et al.,1988; Shellis et al.,1998; Martin et al.,2003; Grine,2005; Grine et al.,2005) and personal observations of enamel thickness, such that a range of thin- to thick-enameled teeth was represented. The three fossil teeth were chosen on the basis of absolute enamel thickness and geological age, with the latter expected to exhibit some degree of correlation with degree of diagenetic remineralization.
|Taxon||Age||Tooth||BCD* (mm)||Maximum Lateral Linear Enamel Thickness (mm)|
|Diademodon tetragonus||c. 241 Myr||Maxillary Molariform||8.90||0.05|
|Moschops capensis||c. 263 Myr||Incisor||8.50||0.36|
|Papio robinsoni||c. 2.0 Myr||LM2||10.00||0.90|
The fossil baboon specimen derives from the site of Bolt's Farm, South Africa, and dates to approximately 2.0 Myr ago (Delson,1984,1988). The specimen (University of California Museum of Paleontology, UCMP 56773) is most likely attributable to Papio robinsoni (E. Delson, personal communication). The two therapsid teeth are from the Permo-Triassic sediments of the Karoo Basin of South Africa. The younger of the two specimens (National Museum, Bloemfontein, specimen QR 1222) is a molariform tooth of the cynodont Diademodon tetragonus, the enamel of which is extremely thin (Grine et al.,1979). This fossil dates to the Early-Middle Triassic boundary (Lucas,1995), or the Early Anisian, between some 237 and 245 Myr (Hancox and Rubidge,2001; De Kock et al.,2003; Rubidge, 2005). The older of the two therapsid specimens (South African Museum, Cape Town, specimen K 203) is an isolated tapinocephalid incisor. It derives from the Abrahamskraal Formation of the Beaufort Group and dates to the Middle Permian (Smith and Keyser,1995), or Capitanian, between about 260 and 266 Myr (Hancox and Rubidge,2001; De Kock et al.,2003; Rubidge, 2005). The incisor is compatible with assignment to either Struthiocephalus whaitsi or Moschops capensis; because the latter is more common in these strata (Smith and Keyser,1995), we assign the tooth to it.
Physical Cross-Section and mCT Image Acquisition
In order to generate a three-dimensional model of each tooth, every specimen was scanned in its entirety using a fan-beam-type desktop mCT system (μCT 40; Scanco Medical). The diameter of the specimen tubes used, as well as the voxel dimensions and number of slices produced, are given in Table 2. All scans were output with 2,048 × 2,048 pixels per slice and 8 bits per pixel; the resultant voxels were isometric (having identical length, width, and depth). Each scan used the same settings for voltage (50 kV) and amperage (160 μA), and the acquisition time for each slice was 4–7 sec, with a reconstruction time of approximately 24 sec per slice at an angular increment of 0.36°. After the mCT scanning of each tooth was complete, the specimens were prepared for physical sectioning following established methods (Grine,2005). The tips of the two mesial cusps (protocone and paracone, or metaconid and protoconid) of the primate molars were marked with a spot of permanent ink. An equivalent procedure was employed for the sauropsid and therapsid teeth, taking into account the fact that because they differ morphologically from the primate molars, a homologous plane is not available. In these instances, a series of anatomical landmarks and pseudolandmarks (e.g., cusp tips, cervical margins) was identified in order to define the location of the plane of section. The crown was then embedded in epoxy and sectioned with a 0.15 mm diamond-wafering blade (Buehler Isomet). The blade was positioned immediately distal to the ink marks to ensure that the mesial crown block included both dentine horns. This block face was ground with 400-grade paper and polished with a sequence of diamond pastes to 0.25 μm (Buehler Microcloth) to obtain a topography-free buccolingual (BL) section that traversed the tips of both dentine horns (in the case of the primate molars). The polished surface was lightly etched with 0.5% H3PO4 for 15 sec to remove any smeared enamel, ultrasonicated in distilled H2O, mounted on a stub, and coated with silver for examination by scanning electron microscopy (AMRAY 1810). These specimens were examined at 25 kV in either secondary or, more commonly, back-scattered electron mode. Micrographs were recorded using Polaroid type 55 P/N film at magnifications between 7.5× and 11.0×, depending on the size of the specimen; working distance was held under 27 mm to ensure accurate magnification. Enlargements of the micrographs were used for measurement.
|Taxon||Tooth||Tube Diameter (mm)||Voxel Dimension (mm)||Post-Processing Pixel Resolution (mm)||Scan Length (mm)|
In order to compare measurements recorded from the physical sections to those obtained by mCT, stacked image sequences of each tooth derived from mCT scanning were imported into the VoxBlast software program (Vaytek) to create three-dimensional reconstructions of the crown. The size of the mCT image file at the slice thicknesses used here (3.0–10.0 GB for each image stack in this study) imposes certain hardware limitations, so image stacks were stored at a resolution of 512 × 512 pixels [(the same pixel dimensions used by Kono (2004)] rather than the 2,048 × 2,048 pixels generated by the mCT unit. We noted no visual loss of image quality after the adjustment. Because the slice thickness of the scans was preserved, the pixel size was effectively four times the slice thickness. All VoxBlast procedures described here were performed on a computer equipped with a 2.8 GHz Hyper-Threading Pentium IV processor (Intel) and 2.0 GB of random access memory.
VoxBlast software allows user-defined planes of section to be created, and these can pass through the three-dimensional reconstructions at any orientation. Using two enamel cusp tips as landmarks (in the case of the primate molars), a buccolingual section was made through each tooth model, attempting to capture the same plane as was located in the physical sections. An iterative (slice-by-slice) technique was employed to refine the location of this plane of section based on visual comparison of the SEM micrographs to the mCT model. Anatomical landmarks unique to a particular plane (e.g., cracks in the enamel or dentine, distinct pulp chamber morphology, distinct dentine topography) were identified in order to match the mCT section to the physical section. After locating the desired plane of section in the mCT model, the image of this plane was exported and printed at its resolution based on our postprocessing technique (512 × 512 pixels). Two images representing the same plane of section were thus available for comparison for each specimen: one generated by physical sectioning and one generated by mCT (Fig. 1).
As mCT analysis of teeth becomes commonplace, the measurement of dental tissues using software alone (rather than digitizing a printed image) is likely also to become widespread. Therefore, we compared measurements that were recorded using both techniques to see whether they differ. Measuring the mCT sections and the SEM images of the histological sections using a digitizing tablet served to control potential sources of measurement error by ensuring that identical techniques were used to measure both sets of images. Digitization also made the method of analysis used here comparable to that of previous studies. We then compared digitized measurements with on-screen measurements in order to establish whether the data derived from these two techniques are equivalent. Certain measurements that are based on anatomical landmarks (e.g., the thickness of enamel between the dentin horn tip and the enamel cusp tip) were recorded on the mCT images for three of the specimens using VoxBlast software. These values were compared to measurements obtained by digitizing the printed images.
Measurements describing the thickness of enamel relative to other dental dimensions were recorded for each primate molar (Fig. 2). These include the area of coronal dentine and pulp (b), the area of the enamel cap (c), and the bicervical diameter (BCD). Eleven measurements describing the linear thickness of enamel at various locations across the molar crown were also recorded for each primate tooth. Because different absolute enamel thicknesses are accompanied by differential rates of error using standard CT methods (Grine,1991; Spoor et al.,1993), the linear measures were devised to examine whether areas of thick and thin enamel in an individual tooth yield comparable rates of measurement error using mCT. In the case of the fossil cercopithecoid molar, the plane of section on which we took measurements coursed through the distal cusps due to postmortem breakage of the mesial half of the tooth. The morphology of extant sauropsid and fossil therapsid teeth is not comparable to that of primates, so a unique set of linear enamel thickness measurements was defined for comparing the two planes of section for each of these teeth (Fig. 3).
All measurements were recorded on enlargements of physical section micrographs and the printed mCT sections using SigmaScan software (Systat Software) interfaced with a SummaSketch III digitizing tablet (CalComp). Values were recorded to the nearest 0.1 mm (or 0.1 mm2). Measurements derived from scanning electron microscopy were calibrated using the measurement scale printed on the micrographs. Measurements derived from mCT models were calibrated by the bicervical diameter of the image. This was determined using VoxBlast, as the number of pixels spanning the bicervical diameter related to the pixel size in microns. In order to minimize intraobserver error, each measurement was recorded three times, and the mean of these three trials was recorded as the value. In order to eliminate the potentially confounding effect of image manipulation on the resulting measurements, we did not attempt to enhance any image by eliminating scan artifacts (e.g., the ring artifacts evident in the Papio ursinus scan in Fig. 1), or by sharpening borders through the use of automated image filtering techniques.
Evaluation of Fossilized/Mineralized Enamel
The relative difference in density of enamel and dentine impacts directly the ability of mCT (or any X-ray–based radiographic technique) to distinguish these tissues. The relative densities of dentine and enamel were measured with OsiriX software (v1.6) (Rosset et al.,2004) on an Apple Macintosh computer. A single mCT cross-section of each fossil tooth, the two sauropsid teeth, and an extant primate molar were analyzed using OsiriX by tracing a region of interest (ROI) line over the cross-section, ensuring that the line traversed both the enamel and the dentine. The value of each pixel crossed by the line (0 for black pixels, 255 for white pixels, and between 0 and 255 for shades of gray) was exported into a spreadsheet and plotted against its location on the line. The resulting charts were superimposed onto the mCT images of the teeth (Fig. 4). Substantial differences in the pixel values for enamel and dentine imply that they are heterogeneous in their density and therefore discernable for measurement, whereas like-valued pixels for enamel and dentine are not distinguishable by the mCT scanner.
Among the recent primate and sauropsid teeth, the enamel-dentine junction and outer enamel surface (hence the thickness of enamel) were clearly visible (e.g., Figs. 1 and 4). Scans of the modern human tooth that had been embedded in methyl methracylate had more image noise than the other specimens, although measurements did not seem to be affected. Comparisons between the measurements derived from mCT images and physical sections are presented in Tables 3 and 4. The percentage difference between the mCT-derived and the physical section-derived measurements, expressed as an average of the 130 total measurements recorded, was 3.52% across the extant sample. As can be seen from the tables, there was no tendency for smaller teeth (e.g., Ateles paniscus mean difference = 2.89%) to have more or less percentage difference between measurements than larger teeth (e.g., Papio ursinus mean difference = 2.96%). There was no tendency for thick-enameled teeth (e.g., Homo sapiens M3 mean difference = 4.66%) to have less measurement difference than thin-enameled teeth (e.g., Chiropotes satanus M1 mean difference = 4.71%). Moreover, there was no consistent pattern of mCT measurements being either smaller or larger than those taken on physical sections.
|Measurement||Ateles paniscus||Cebus apella||Chiropotes satanus||Homo sapiens M3||Homo sapiens M3||Homo sapiens dm2||Papio ursinus|
|mCT||Physical||% Diff.||mCT||Physical||% Diff.||mCT||Physical||% Diff.||mCT||Physical||% Diff.||mCT||Physical||% Diff.||mCT||Physical||% Diff.||mCT||Physical||% Diff.|
|Average (Absolute Value)||2.89%||2.92%||4.71%||2.80%||4.66%||3.69%||2.96%|
|Measurement||Alligator mississipiensis||Crocodylus palustris|
|mCT||Physical||% Diff.||mCT||Physical||% Diff.|
|Average (Absolute Value)||4.80%||5.52%|
Across the different measurements taken, there is no obvious trend toward thinner regions of enamel being more prone to differences in measurements than thicker areas of the same tooth among the recent primate specimens. Measurements L1, L2, and L3, which represent the linear thickness of lingual enamel from nearer to the cervix to nearer to the cusp, respectively, showed an overall decrease in percent difference, although the same was not true for buccal measurements. Moreover, the increase from L1 to L3 was not substantial, although samples are not large enough to facilitate statistical comparisons. Area measurements (measurements b and c) tend to have lower percent differences than linear enamel thickness measurements.
Recording distance measurements using known landmarks on the computer screen (i.e., directly on mCT volume models in VoxBlast) did not result in substantially different measurements from those obtained by digitizing printed images (Table 5). There is, on average, from 2.60% to 3.50% difference between measurements taken on-screen and those using a digitizing tablet. These differences accord with those between the mCT prints and physical sections, as noted above.
|Measurement||Homo sapiens RM3||Papio ursinus LM2||Cebus apella LM2|
|On-Screen||Digitizer||% Diff.||On-Screen||Digitizer||% Diff.||On-Screen||Digitizer||% Diff.|
|Average (Absolute Value)||2.60%||3.19%||3.50%|
There is a substantial difference between the ability of computer software (e.g., OsiriX, VoxBlast) and the human eye to distinguish enamel from dentine if scans are not first treated with some segmentation routine. Figure 5 depicts the physical and mCT-derived sections of the Alligator tooth, in which one can readily differentiate enamel from the underlying dentine. Also depicted is an ROI line through the same tooth, demonstrating that although the very thin enamel (as thin as 40 μm; Table 4) on the lingual surface is visible to the eye, the software is unable to distinguish the value of the enamel pixels from those of the dentine. This problem becomes important if computer software alone is used to record measurements, such as the volume of the enamel cap. A computer-rendering of the Alligator tooth is depicted in Figure 5; this demonstrates that although enamel is visible to the human eye on the lingual surface of the tooth, software alone (without any tissue segmentation) is incapable of differentiating it from the underlying dentine.
With regard to the mCT images of the fossil teeth (Table 6), while the extant primate molars evince a strong contrast between dentine and enamel, there is less contrast along the ROI line in the 2.0-Myr-old Papio robinsoni molar (Fig. 4). Nevertheless, the enamel cap was clearly visible on this fossil baboon crown in comparison to the substantially older Karoo fossils. Enamel could barely be distinguished from the underlying dentine in the Moschops capensis incisor, and this distinction was accentuated somewhat by their partial physical separation along a postmortem fissure (Fig. 4). The Diademodon tetragonus tooth crown was a homogeneous gray across the entire scan; in this case, mCT was incapable of clearly distinguishing enamel from dentine (Fig. 4).
|Measurement||Moschops capensis||Papio robinsoni|
|mCT||Physical||% Diff.||mCT||Physical||% Diff.|
|Average (Absolute Value)||3.95%||3.47%|
The 2.0-Myr-old Papio robinsoni molar demonstrates that enamel and dentine can be differentiated by mCT in a fossil when the enamel is of moderate thickness (0.5–1.0 mm). The relative inability of mCT to differentiate between enamel and dentine in the Moschops tooth, and its total inability to do so in the Diademodon tooth, may reflect the comparatively greater degree of fossilization (i.e., remineralization) of the enamel (and especially dentine) in these Karoo specimens.
However, geochronological age does not account entirely for differences in the ability of mCT to visualize enamel. The Diademodon tooth is approximately 22 Myr younger than the Moschops incisor. Nevertheless, it is possible to visualize the enamel cap of the Moschops specimen, whereas it is totally invisible to mCT in Diademodon. It is possible that diagenetic alteration, which may be (to some degree) independent of geological age, may have differentially affected the Moschops and Diademodon teeth such that the remineralized dentine and enamel are of a more homogeneous structural composition in the former. In order to investigate this possibility, these two teeth were subjected to energy-dispersive X-ray microanalysis to investigate the composition of the dentine and enamel. The physical section faces of each specimen were examined in a LEO 1550 (LEO/Zeiss) SEM equipped with an EDAX Sapphire Si(Li) detector and Phoenix analyzer (EDAX) at 15 kV, zero tilt, a 4.1 takeoff, and a working distance of 8.0 mm. The results of these analyses are depicted in Figure 6. In the Moschops incisor, the dentine and enamel have essentially identical proportions of calcium and phosphorous (i.e., typical calcium phosphate), and they differ only in the amount of silicon present. In the Diademodon tooth, while calcium and phosphorous predominate the mineral chemistry of the dentine, the enamel is wholly lacking in phosphorous (it also exhibits some manganese). Thus, the dentine and enamel are dissimilar to one another in chemical composition in both specimens, although diagenetic alteration has not been the same in each. The introduction of silicon into the dentine compartment of the Moschops incisor (presumably in the form of quartz) might be expected to affect the mCT X-ray beam attenuation as much as the diagenetic loss of phosphorous from the enamel compartment of the Diademodon crown, which suggests an increase in porosity.
This appears to suggest that neither geological age per se nor the differential diagenetic alteration of the chemical composition of the enamel and dentine adequately explains the differences in the ability of mCT to visualize the enamel caps of the two therapsid teeth. Rather, the difference between them in the detection of enamel via mCT seems to be related principally to its absolute thickness. Enamel in Moschops is approximately six times thicker than in Diademodon (∼ 0.30 mm in Moschops vs. ∼ 0.05 mm in Diademodon). However, the inability of mCT to resolve the enamel cap in Moschops as clearly as in recent and substantially younger fossil teeth may be related to the diagenetic alterations of the dentine in the former.
The percent differences (mean = 3.52%) reported here between enamel thickness measurements derived from mCT scans and physical sections of recent teeth are significantly improved over standard CT procedures (7–34%) (Grine,1991) and especially over lateral dental radiographs (15–115%) (Grine et al.,2001). The percent differences obtained here are comparable to those recorded in studies comparing mCT scans and physical sections of trabecular bone (Müller et al.,1996a,b,1998). This indicates that mCT is a valuable technique for the nondestructive assessment of enamel thickness in extant primate teeth. Figure 1 demonstrates the similarity between mCT and physical sections measured in this study, where only slight differences in the shapes of these sections are apparent.
The percent difference for the measurements reported here apparently does not relate to the thickness of the enamel or the size of a given specimen (Tables 3 and 4). We suggest that the differences are due to human error in the process of measuring the photographs and printed images of the sections, rather than any inherent differences in the comparability of the two sections measured for each tooth or to any other factor having to do with specimen size or thickness.
Although distances measured on the computer screen by way of drawing lines did not differ substantially from those taken by hand using a digitizing tablet (Table 5), the mCT technology used here does appear to be limited in its ability to distinguish enamel from dentine when the enamel is very thin and/or when these two dental tissues have undergone diagenetic alteration. Thus, the extant sauropsid teeth and the Ateles and Chiropotes molars examined here demonstrate that mCT is capable of differentiating between enamel and dentine in recent specimens with thin enamel (as thin as 0.04 mm in the Alligator tooth). However, as is evident from the ROI line across the Alligator tooth (Fig. 5), whereas the pixel values clearly differentiate the thicker enamel, the extremely thin lingual enamel (ca. 0.04 mm) is not clearly distinguished from the adjacent dentine. Thus, it would appear that absolutely thin enamel (less than ∼ 0.10 mm) is difficult to resolve adequately with mCT based on pixel values alone. Nevertheless, while it is not distinguished from dentine based on pixel values, the thin lingual enamel of the Alligator tooth is visible to the human eye in the mCT scans (e.g., Fig. 5). Since the comparably thin enamel of the Diademodon is not visible in terms of pixel values or to the human eye, this problem appears to be exacerbated by diagenetic processes involved in fossilization. It is possible that other mCT systems, such as industrial mCT (e.g., Bush et al.,2004) and synchrotron mCT (Tafforeau,2004; Tafforeau et al.,2006), may be better suited to imaging thinly enameled fossils.
The poor visualization of enamel by mCT in the ca. 250-Myr-old therapsids compared to the ca. 2.0-Myr-old papionin, and the somewhat lower contrast in pixels along the ROI in this specimen compared to recent teeth, suggest that diagenetic alteration, which may differ substantially with geochronological age, may have an impact on visualization based on mCT scanning. On the other hand, the differences in detection by mCT of the enamel caps of the two therapsid specimens suggest that the absolute thickness of enamel (0.30 mm in Moschops vs. 0.05 mm in Diademodon) is a more problematic issue. Finally, the ability of segmentation software to distinguish enamel and dentine differs in some cases from the ability of the human eye to detect the same two tissues. The thin enamel of the Alligator tooth (e.g., Fig. 5) demonstrates that while the human eye may detect and measure thin enamel on mCT scans, computer-automated tissue segmentation (based on pixel shade values) may fail to do so. Thus, the measurement of enamel thickness based on mCT scans by means of computer software alone may exacerbate the two problems of thin enamel and remineralization via diagenesis.
The authors are grateful to R. Monk and J. Spence (Department of Mammalogy, American Museum of Natural History) for permission to mCT and section a molar of Cebus apella. Specimens were graciously loaned by R. Smith (South African Museum, Cape Town), J. Botha (National Museum, Bloemfontein), and P. Holroyd (University of California Museum of Paleontology, Berkeley). Information pertaining to these specimens was provided by E. Delson, B. Rubidge, and M. Raath. S. Judex and S. Xu provided access to and assistance with mCT facilities. J. Sipla, A. Kaufman, T. Smith, and S. Frank provided advice and assistance. We thank T. Radsbury for the discussion and we are grateful to L. Betti-Nash for the masterful execution of the illustrations. This manuscript benefited greatly from the comments and suggestions of J. Laitman and two anonymous reviewers.
- 1981. Enamel thickness in 45 X females permanent teeth. Am J Hum Genet 33: 464–469. , .
- 1985. Dental growth in 47 XYY males and in conditions with other sex-chromosome anomolies. In: SandbergAA, editor. The Y chromosome. New York: Alan R. Liss. p 277–300. .
- 1985. Enamel thickness in 47 XYY males' permanent teeth. Ann Hum Biol 12: 421–427. , , .
- 1987. 47 XYY females, sex chromosomes, and tooth crown structures. Hum Genet 7: 345–348. , , .
- 2004. New approach to quantifying developmental variation in the dentition using serial microtomographic imaging. Microsc Res Tech 65: 263–269. , , , , , .
- 2000. Quantification of periapical bone destruction in mice by micro-computed tomography. J Dent Res 79: 35–40. , , , , .
- 1963. Observation of internal structures of tooth by ultrasonography. Science 139: 495–496. , , .
- 1986. Variations in enamel thickness and structure in East African hominids. Am J Phys Anthropol 70: 177–193. , .
- 1991. On thick and thin enamel in hominoids. Am J Phys Anthropol 86: 295–309. , , .
- 1994. 3D computed X-ray tomography of human cancellous bone at 8 μm spatial and 10−4 energy resolution. Bone Miner 25: 25–38. , , , , , , , .
- 2002. A new hominid from the Upper Miocene of Chad, Central Africa. Nature 418: 145–151. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , .
- 2004. High-resolution computed tomography study of the cranium of a fossil anthropoid primate, Parapithecus grangeri: new insights into the evolutionary history of primate sensory systems. Anat Rec A 281A: 1083–1087. , , .
- 2003. A Middle Miocene hominoid from Thailand and orangutan origins. Nature 422: 61–65. , , , , , , .
- 1991. Enamel thickness in South African australopithecines: noninvasive evaluation by computed tomography. Paleont Afr 28: 53–59. .
- 1991. Noninvasive evaluation of enamel thickness and volume in South African australopithecines by computed tomography. Am J Phys Anthropol 12(Suppl): 60. , .
- 1995. Some observations on enamel thickness and enamel prism packing in the Miocene hominoid Otavipithecus namibiensis. Am J Phys Anthropol 98: 595–600. , , .
- 2004. Comparison of microcomputed tomographic and microradiographic measurements of cortical bone porosity. Calcif Tiss Int 74: 437–447. , , , .
- 1987. Some results on the dental enamel structure in fossil and recent reptiles. CR Acad Sci II 305: 1217–1219. .
- 2003. Permian-Triassic megnetostratigraphy in the Karoo Basin of South Africa. Geochem Cosmochim Acta 67: A77–A78. , , , , , .
- 1984. Cercopithecid biochronology of the African Plio-Pleistocene: correlation among eastern and southern hominid-bearing localities. Cour Forsch Inst Senckenberg 69: 199–218. .
- 1988. Chronology of South African australopith site units. In: GrineFE, editor. Evolutionary history of the “robust” Australopithecines. New York: Aldine de Gruyter. p 317–324. .
- 1995. Enamel thickness and dietary adaptation among extant primates and chiropterans. J Mammal 76: 1127–1136. .
- 1991. Cancellous bone structure: analysis of high-resolution CT images with run-length method. J Comp Assist Tomogr 15: 133–139. , .
- 1994. A Neanderthal infant from the Barakai Cave, western Caucasus. J Hum Evol 27: 405–415. , , , , .
- 2001. Three-dimensional analysis of non-human primate trabecular architecture using micro-computed tomography. Am J Phys Anthropol 115: 327–336. , .
- 2002. Assessing the accuracy of high-resolution X-ray computed tomography of primate trabecular bone by comparison with histological sections. Am J Phys Anthropol 118: 1–10. , , .
- 1983. The enamel of Neogene hominoids: structual and phyletic implications. In: CiochonRL, CorrucciniRS, editors. New interpretations of ape and human ancestry. New York: Plenum Press. p 249–298. .
- 2003. 3D approach to interpret enamel thickness and volume. Am J Phys Anthropol 36(Suppl): 99. , , , , .
- 1979. Enamel prisms and diphydonty: linked apomorphies of Mammalia. S Afr J Sci 75: 114–120. , , .
- 1988. Enamel thickness and development in Australopithecus and Paranthropus. In: GrineFE, editor. Evolutionary history of the “robust” Australopithecines. New York: Aldine de Gruyter. p 3–42. , .
- 1991. Computed tomography and the measurement of enamel thickness in extant hominoids: implications for its paleontological application. Palaeont Afr 28: 61–69. .
- 2001. An evaluation of dental radiograph accuracy in the measurement of enamel thickness. Archs Oral Biol 46: 1117–1125. , , .
- 2002. Scaling of tooth enamel thickness, and molar crown size reduction in modern humans. S Afr J Sci 98: 503–509. .
- 2004. Geographic variation in tooth enamel thickness does not support Neandertal involvement in the ancestry of modern Europeans. S Afr J Sci 100: 389–394. .
- 2005. Enamel thickness of deciduous and permanent molars in modern Homo sapiens. Am J Phys Anthropol 126: 14–31. .
- 2005. Molar enamel thickness in the Chacma baboon, Papio ursinus. Am J Phys Anthropol 128: 812–822. , , , , , .
- 1960. A geometrical evaluation of image size in dental radiography. J Dent Res 39: 289–301. .
- 2001. Late Miocene hominids from the Middle Awash, Ethiopia. Nature 412: 178–181. .
- 2004. Late Miocene teeth from Middle Awash, Ethiopia, and early hominid dental evolution. Science 303: 1503–1505. , , .
- 2001. Breakthroughs in the biodiversity, biogeography, biostratigraphy, and basin analysis of the Beaufort group. J Afr Earth Sci 33: 563–577. , .
- 1998. Enamel thickness in maxillary human incisors: a radiographic assessment. Archs Oral Biol 43: 825–831. , .
- 1999. Absence of sexual dimorphism in enamel thickness of human deciduous molars. In: MayhallJT, HeikkinenT, editors. Dental morphology. Oulu, Finland: Oulu University Press. p 338–349. , , .
- 1999. Direct three-dimensional morphometric analysis of human cancellous bone: microstructural data from spine, femur, iliac crest, and calcaneus. J Bone Miner Res 14: 1167–1174. , , , , .
- 1998. Advances in three-dimensional data acquisition and analysis. In: RosenbergerAL, FleagleJG, McHenryHM, StrasserE, editors. Primate locomotion. New York: Plenum. p 205–222. .
- 1981. The nut-crackers: a new theory of the adaptations of the Ramapithecinae. Am J Phys Anthropol 55: 141–151. .
- 1990. Anatomy of cusps of posterior teeth and their fracture potential. J Prosthet Dent 64: 139–147. , , , .
- 1998. Three-dimensional in vivo morphometry of trabecular bone in the OVX rat model of osteoporosis. Technol Health Care 6: 339–350. , , , .
- 2002. A three-dimensional analysis of enamel distribution patterns in human permanent first molars. Archs Oral Biol 47: 867–875. , , .
- 2004. Molar enamel thickness and distribution patterns in extant great apes and humans: new insights based on a 3-dimensional whole crown perspective. Anthropol Sci 112: 121–146. .
- 1997. A new method of evaluating enamel thickness based on a three-dimensional measuring system. Anthropol Sci 105: 217–229. , , , .
- 1990. Evaluation of a microcomputed tomography system to study trabecular bone structure. J Orthoped Res 8: 833–842. , , , , .
- 1968. Specific acoustic impedence of enamel and dentine. Archs Oral Biol 13: 1491–1500. .
- 1968. Looking into teeth with ultrasound. Science 161: 477. , .
- 1971. Ultrasonics in hard dental tissues. Int Dent J 21: 403–417. .
- 1995. Towards dicynodont biochronology. Albertina 16: 33–40. .
- 1992. Computed-tomography and enamel thickness of maxillary molars of Plio-Pleistocene hominids from Sterkfontein, Swartkrans, and Kromdraai (South-Africa): an exploratory study. Am J Phys Anthropol 89: 133–143. , .
- 1993. Enamel thickness of human maxillary molars reconsidered. Am J Phys Anthropol 92: 189–200. , .
- 1999. Effects of loading on the biochemical behavior of molars of Homo, Pan, and Pongo. Am J Phys Anthropol 109: 211–227. , .
- 2002. A comparison of the femoral head and neck trabecular architecture of Galago and Perodicticus using micro-computed tomography (μCT). J Hum Evol 43: 89–105. , .
- 2000. Acoustic microscopy: a new method for investigation of dental tissues. Stomatology 79: 14–19. , , , , , , .
- 2002. New data on histology and physico-mechanical properties of human tooth tissue obtained with acoustic microscopy. Ultrasound Med Biol 28: 131–136. , , , .
- 1985. Significance of enamel thickness in hominoid evolution. Nature 314: 260–263. .
- 2003. Enamel thickness and microstructure in pitheciin primates, with comments on dietary adaptations of the middle Miocene hominoid Kenyapithecus. J Hum Evol 45: 351–367. , , .
- 2004. Micro-computed tomography of a 500-year-old tooth: technical note. J Assoc Canad Radiol 55: 242–245. , , , .
- 1999. Evaluation of human tooth structure with the ultrasonic imaging technique. J Oral Sci 41: 191–197. , , , , , , .
- 1918. The Piltdown jaw. Am J Phys Anthropol 1: 25–52. .
- 1977. Functional implications of primate enamel thickness. Am J Phys Anthropol 46: 447–454. , .
- 1993. Hominid enamel thickness: I, the Krapina Neandertals. Am J Phys Anthropol 92: 131–138. , , , , .
- 2002. The Zürich experience: one decade of three-dimensional high-resolution computed tomography. Topics Magnet Resonance Imag 13: 307–322. .
- 2003. Bone microarcitecture assessment: current and future trends. Osteoporos Int 14: 89–99. .
- 1996a. Resolution dependency of microstructural properties of cancellous bone based on three-dimensional μ-tomography. Technol Health Care 4: 113–119. , , , , , .
- 1996b. Morphometric analysis of noninvasively assessed bone biopsies: comparison of high-resolution computed tomography and histologic sections. Bone 18: 215–220. , , , , .
- 1997. Micro-tomographic imaging for the nondestructive evaluation of trabecular bone architecture. In: MeunierA, editor. Bone research in biomechanics. Amsterdam: IOS Press. p 61–79. , .
- 1998. Morphometric analysis of human bone biopsies: a quantitative structural comparison of histological sections and micro-computed tomography. Bone 23: 59–66. , , , , , , .
- 1990. Molar enamel thickness in European Miocene and extant Hominoidea. Int J Primatol 11: 283–295. .
- 2005. High resolution measurement of Neandertal tooth enamel thickness by micro-focal computed tomography. S Afr J Sci 101: 219–220. , .
- 1989. Studies on sound and carious enamel with the quantitative acoustic microscope. J Dent Res 68: 107–112. , , .
- 2000. Three-dimensional analysis of root canal geometry by high-resolution computed tomography. J Dent Res 79: 1405–1409. , , , .
- 1999. Micro-computed tomography: a new tool for experimental endodontology. Int Endodont J 32: 165–170. , , , , .
- 2004. OsiriX: an open-source software for navigating in multidimensional DICOM images. J Digital Imag 17: 205–216. , , .
- 2006. Re-uniting lost continents: fossil reptiles from the ancient Karoo and their wanderlust. S Afr J Geol (in press). .
- 1996. A microtomographic system for the non-destructive evaluation of bone architecture. Calcif Tissue Int 58: 24–29. , , .
- 1988. Morphology of the American alligator (Alligator mississipiensis) tooth: I, the fine-structure and analysis of the enamel. Am Zool 28: A78. , , , , .
- 1973. Thickness of enamel and dentine. J S Calif Dent Assoc 41: 33–52. , .
- 1998. Enamel thickness and the topography of the enamel-dentine junction in South African Plio-Pleistocene hominids with special reference to the Carabelli trait. J Hum Evol 35: 523–542. , , , .
- 2000a. Enamel thickness and the helicoidal wear plane in modern human mandibular molars. Archs Oral Biol 45: 401–409. .
- 2000b. Taxonomic and functional aspects of the patterning of enamel thickness distribution in extant large-bodied hominoids. Am J Phys Anthropol 111: 221–244. .
- 2001. First hominid from the Miocene (Lukeino Formation, Kenya). CR Acad Sci Ser II A 332: 137–144. , , , , , .
- 1998. Variations in molar enamel thickness among primates. J Hum Evol 35: 507–522. , , , .
- 1972. Hominiod paleoprimatology. In: TuttleR, editor. The functional and evolutionary biology of primates. Chicago: Aldine. p 36–62. , .
- 1994. Thin enamel and other tooth components in Neanderthals and other hominids. Am J Phys Anthropol 95: 85–87. , .
- 1995. Biostratigraphy of the Tapinocephalus assemblage zone. In: RubidgeBS, editor. SACS biostratigraphic series no 1. Pretoria: Council for Geoscience. p 8–17. , .
- 1998. Biomechanical behaviour of modern human molars: implications for interpreting the fossil record. Am J Phys Anthropol 106: 467–482. , .
- 1985. Comparative primate dental enamel thickness. In: TobiasPV, editor. Hominid evolution: past, present, and future. New York: Alan R. Liss. p 443–454. .
- 1986. Palaeodontology: radiographic revelations of the Australopithecinae. In: SingerR, LundyJK, editors. Variation, culture, and evolution in African populations. Johannesburg, South Africa: Witwatersrand University Press. p 195–207. .
- 1993. Linear measurements of cortical bone and dental enamel by computed-tomography: applications and problems. Am J Phys Anthropol 91: 469–484. , , .
- 1997. A reappraisal of early hominid phylogeny. J Hum Evol 32: 17–82. , , .
- 2001. The systematics of Australopithecus garhi. Ludus Vitalis 9: 109–135. , .
- 2004. Inferring hominoid and early hominid phylogeny using craniodental characters: the role of fossil taxa. J Hum Evol 47: 399–452. , .
- 1994. Sexual dimorphism in mesiodistal dentine and enamel thickness. Dent-Max-Fac Radiol 23: 169–171. , , .
- 1998. Enamel thickness of the posterior dentition: its implications for nonextraction treatment. Angle Orthod 68: 141–146. , , .
- 2004. Aspects phylogénétiques et fonctionnels de la microstructure de l'émail dentaire et de la structure tridimensionnelle des molaires chez les primates fossils et actuels: apports de la microtomographie é rayonnement X synchrotron. PhD thesis. Montpellier: Université de Montpellier II. .
- 2006. Applications of X-Ray synchotron microtomography for non-destructive studies of paleontological specimens. Appl Phys A (in press). , , , , , , , , , , , , , , , , , .
- 2003. Three-dimensional quantitation of periradicular bone destruction by micro-computed tomography. J Endodont 29: 252–256. , , , .
- 1994. Australopithecus ramidus, a new species of early hominid from Aramis, Ethiopia. Nature 371: 306–312. , , .
- 2003. Neuroanatomy of flying reptiles and implications for flight, posture, and behavior. Nature 425: 950–953. , , , .
- 2002. Sahelanthropus or “Sahelpithecus”? Nature 419: 581–582. , , , .
- 1991. Ultrasound surface imaging and the measurement of tooth enamel thickness. PhD dissertation. Manchester: University of Manchester. .
- 1996. Tooth enamel surface imaging measurement by ultrasound imaging method. J Acoust Soc Am 100: 2710. .
- 1990. Components of australopithecine teeth, a radiographic study. J Hum Evol 5: 515–529. , , .
- 1992a. A comparison of tooth structure in Neanderthals and early Homo sapiens sapiens: a radiographic study. J Anat 180: 387–393. , .
- 1992b. Evolutionary trends in hominid tooth components: a radiographic study. In: SmithP, TchernovE, editors. Structure, function, and evolution of teeth. London: Freund. p 349–359. , .
- 1992. Tooth components of mandibular deciduous molars of Homo sapiens sapiens and Homo sapiens neanderthalensis: a radiographic study. Am J Phys Anthropol 87: 255–262. , , .
- 1994. Sex discrimination using tooth components measured from bite-wing radiographs. Am J Phys Anthropol S18: 214. , .
- 1985. High resolution computed tomography of fossil hominid skulls: a new method and some results. In: TobiasPV, editor. Hominid evolution: past, present, and future. New York: Alan R. Liss. p 427–436. , .