Current issues with standards in the measurement and documentation of human skeletal anatomy


Justin Magee, Research Institute for Art and Design, School of Creative Arts, University of Ulster, Northland Road, Derry/Londonderry, Northern Ireland BT48 7JL, UK. T: +44287175355; E:


Digital modeling of human anatomy has become increasingly important and relies on well-documented quantitative anatomy literature. This type of documentation is common for the spine and pelvis; however, significant issues exist due to the lack of standardization in measurement and technique. Existing literature on quantitative anatomy for the spine and pelvis of white adults (aged 18–65 years, separated into decadal categories) was reviewed from the disciplines of anatomy, manipulative therapy, anthropometrics, occupational ergonomics, biomechanics and forensic science. The data were unified into a single normative model of the sub-axial spine. Two-dimensional orthographic drawings were produced from the 590 individual measurements identified, which informed the development of a 3D digital model. A similar review of full range of motion data was conducted as a meta-analysis and the results were applied to the existing model, providing an inter-connected, articulated digital spine. During these data analysis processes several inconsistencies were observed accompanied by an evidential lack of standardization with measurement and recording of data. These have been categorized as: anatomical terminology; scaling of measurements; measurement methodology, dimension and anatomical reference positions; global coordinate systems. There is inconsistency in anatomical terminology where independent researchers use the same terms to describe different aspects of anatomy or different terms for the same anatomy. Published standards exist for measurement methods of the human body regarding spatial interaction, anthropometric databases, automotive applications, clothing industries and for computer manikins, but none exists for skeletal anatomy. Presentation of measurements often lacks formal structure in clinical publications, seldom providing geometric reference points, therefore making digital reconstruction difficult. Published quantitative data does not follow existing international published standards relating to engineering drawing and visual communication. Large variations are also evident in standards or guidelines used for global coordinate systems across biomechanics, ergonomics, software systems and 3D software applications. This paper identifies where established good practice exists and suggests additional recommendations, informing an improved communication protocol, to assist reconstruction of skeletal anatomy using 3D digital modeling.


Measurement of human skeletal anatomy is important to inform a wide range of clinical practice, for instance orthopedics, rheumatology, craniomaxillofacial surgery and midwifery. Measurements are used in the manufacture of medical devices such as artificial hip joints, vertebral discs, pedicle inserts, vascular stents, forensic analysis, gait and other biomechanical analysis. Modeling human anatomy has improved and is used to support a better understanding of form and function. Modeling types include visual models (Ackerman, 1997), mathematical models (Panjabi, 1973), manipulative models (Seo & Magnenat-Thalmann, 2004) and articulated models (Ivancevic & Beagley, 2005). The first digital models of anatomy were developed for flight simulation analysis in 1959 (Fetter, 1982), and since that time they have become more detailed and sophisticated, capable of replicating internal as well as external anatomy using animation, design and engineering software. Digital models may be visually representative or used for quantitative purposes. Development of the latter requires reliable measurement data taken from human anatomy both in vivo and in vitro.

Radiological imaging modalities [e.g. computed radiography, computed tomography (CT), magnetic resonance imaging (MRI)], 3D surface scanning and detailed physical measurement of cadavers using manual calipers have all been employed for anatomical measurement. Imaging has been used for both in vivo and in vitro studies; however, the use of ionizing modalities is regulated in living subjects by the Ionizing Radiation Medical Exposure Regulations (IRMER; Department of Health, 2000). These regulations aim to reduce potentially harmful radiation exposure from imaging in clinical practice. The use of ionizing radiation in imaging research is normally restricted to diagnosis, although these methods have often been used for anatomical measurement research, either retrospectively, following appropriate approval or in early studies, which were not subject to these regulations (Nissan & Gilad, 1984; Amonoo-Kuofi, 1992; Frobin et al. 1997, 2002b; Harrison et al. 1998, 2000a, 2001, 2003a,b; Janik et al. 1998; Genda et al. 2001; Wolf et al. 2001; Marty et al. 2002; Naderi et al. 2003). Images of projected anatomy are acquired using conventional X-ray imaging, and anatomy is subject to variable magnification and geometric distortion. MRI does not employ ionizing radiation, and it is regarded as non-invasive; it provides multi-planar data that require specialist knowledge for image interpretation, analysis and measurement. Measurements of human anatomy using external bony landmarks using calipers are generally less accurate, being subject to palpation error, although common for certain pelvic measurements where the bone is near the skin surface. Palpation error of the spine is reported to be approximately 20 mm (Salisbury & Porter, 1987; Harlick et al. 2007; Robinson et al. 2008) and 10 mm in anthropometrics studies (Pheasant, 1988). It should be noted that in palpated studies in anthropometrics for ergonomic applications, sample sizes are often much larger than clinical studies, as recruitment of subjects requires less stringent research governance.

The British Standards Institute (BSI) ( is the National Standards Body (NSB). Standardization exists for measuring subjects for spatial interaction (BS 7250 1998, BS 6385 2004), anthropometric databases (BS 15535 2003), within automotive design (BS 173 1979, BS 6549 1999, BS 2631-4 2001), for clothing industries (BS 5511 1977, BS 13402-1 2001), and for computer manikins (BS 15536-2 2007, BS 15536-1 2008); however, no such standards exist for the measurement of human anatomy, which has led to significant problems. The aim of this paper is to discuss the issues created by the lack of standardization in measurement and documentation of skeletal anatomy research. We discuss these issues in the context of recent research into ‘3D digital modeling of skeletal posture’ (Magee, 2010). This paper suggests recommendations for a new standardized anatomical measurement protocol, in consideration of digital modeling reconstruction.

Materials and methods

When investigating the original research ‘3D digital modeling of spinal posture’, methods were adopted in developing the digital models that identified the need for a new measurement and documentation protocol. Measurement of the sub-axial spine was extracted from the literature in the date range 1893–2006. Literature was retrieved from PubMed, ScienceDirect, Google scholar, and a manual search of each article reference list was conducted. Literature was included for white adults, as ethno-geographic categorization is common in scientific research. The age category of 18–65 years was used. In the context of spinal research, the term adulthood occurs when people reach their maximum stature, at 20.2 years for boys and 17.3 years for girls (Roche & Davila, 1972). Earlier, postural and morphological variation is prevalent (Mac-Thiong et al. 2004; Poussa et al. 2005), and at 50 years deformation of the spine may begin (Milne & Lauder, 1974), widely recognized as more acute beyond 65 years. In anthropometrics the typical adult age categories are 19–65 years (Pheasant, 1988) or 18–64 years (Open Ergonomics 2008), although categorization in 10-year intervals is recommended (Durnin & Womersley, 1974; Department of Health, 1998, 2002).

Sex categories were used to differentiate characteristics in anatomy, published in literature (Hanna & Washburn, 1953, Singh & Potturi, 1978; Luo, 1995; Patriquin et al. 2005; Walker, 2005) associated with the pelvis, although spinal research often combined male and female data together so both male and female data were analyzed.

From the literature reviewed, 60 documents directly related to the quantitative measurement of human skeletal anatomy of the sub-axial spine. From these, 43 met the inclusion/exclusion criteria. These included the sub-axial spine data for white subjects, aged between 18 and 65 years, and excluded studies absent of documented ethno-geographic data; single figure-type data (e.g. endomorphic samples only); studies that recorded damaged or shrunken cadavers and living samples with spinal degeneration, disease or trauma; studies that did not record average stature or other scalable references; studies that did not document time of research, making it difficult to determine position of data along a time line and subsequently difficult to normalize.

The International System of Units (Metric), which is preferred in European practice, was used for engineering drawing and dimensioning of the digital models. It was noted that although the imperial method is used in American architecture and industrial design, clinical publications use the metric system. Drawings in third angle projection were produced following international standards (BS 8888 2011), for the front (anterior), back (posterior) and side views (sagittal). Orthographic CAD drawings were generated using ashlar vellum graphite v7.1 with its 3D parametric modeling counterpart cobalt v6.2 ( Software had 1e-06º (degree) or mm precision, providing CAD construction accuracy, which was more than sufficient for tolerances published in the literature, of ± 10.0 (palpated anatomy errors), ± 1.0 (some pelvic anatomy), ± 0.1 (all vertebrae and most other anatomy), and only occasionally ± 0.01 º or mm (vertebral anatomy). Tolerances of ± 0.1 mm for linear and ± 0.1 º for angular dimensions were used in the digital model as they were the most common tolerances published in detailed studies. The 3D vertebrae models were constructed as separate CAD drawings, including top (transverse), side (sagittal) and front (anterior) elevations. Three-dimensional digital models (Fig. 1) were generated from the 2D CAD data using 3dsmax software ( commonly used for object visualization and animation. Additional reference material was used to help inform how 3D geometry morphed between measured anatomy (Goldfinger, 1991; Kappelman et al. 2000; Ellis et al. 2001).

Figure 1.

 Skeletal Joe sub-axial spine model.

A normalization rationale was used, scaling and combining measurement data from static anatomy studies with various physical sized subjects. The measurement of stature was used for the normalization scaling factor, defined by the ratio between the average value of each original study and the average stature of the UK population. The resulting anthropometrics were then combined and overall averages calculated increasing the likelihood that normal distribution was represented within the context of this digital reference model.


Summary of ‘skeletal Joe’, a normative digital model of the sub-axial spine

From the selected data, one study included females only and four studies included males only. Thirty-six studies included both sexes, with 16 documenting male and female results separately. From the 20 studies that combined data without distinction, 17 related to the spine, one to the femur neck angles, one to the occiput and one to pelvic dimensions. Literature was sourced from the disciplines of anatomy, manipulative therapy, anthropometrics, occupational ergonomics, biomechanics and forensic science. In addition, the much cited Hammon–Todd collection was closely inspected by access to their original museum collection spreadsheets (see Acknowledgements) to acquire quantitative data, not published in some literature (that had unique measurements), allowing their data to be included (Caldwell & Molloy, 1933; Francis, 1955). There were a total of 590 measurements used to produce the model as 2D CAD drawings. They included 47 pelvic measurements, 484 relating to the spine and two dimensions for skin fold thickness, required in the context of palpable bony landmarks, 27 spinal defining its curvature, alignment and vertebral orientation, 11 for the atlas, 27 for the axis, 23 in each of the five remaining cervical vertebrae, 17 in each of the 12 thoracic vertebrae and 20 in each of the lumbar vertebrae. 57 general anthropometrics measurements were included, providing a spatial context for anatomy relative to the midpoint between the feet while in the anatomical position. This position has Cartesian coordinates of 0, 0, 0, and is referred to as the origin (Fig. 2).

Figure 2.

 Two-dimensional drawings of skeletal Joe (a – upper and lower limbs; b – lumbar spine and pelvis).

The age categories of 18–29, 30–39, 40–49, 50–59 and 60–65 years old were selected for data recording following consideration of best practice (Durnin & Womersley, 1974; Department of Health, 1998, 2002). The average age of subjects for each of the spinal regions was 32.7 years for the cervical spine, 31.6 years for the thoracic spine and 27.9 years for the lumbar spine. The pelvic data age was not possible to calculate due to lack of published information. The normalized digital model measured 1758.2 mm in stature. The original publication of these results (Magee, 2005, 2010) was superseded by a refined model (Magee, 2010). The latter model included a range of motion (ROM) meta-analysis leading to the development of a normative articulated digital model, for the entire interconnected spine in respect of maximum, average (Fig. 3) and minimum values published for ‘full’ ROM (Magee et al. 2009, 2010).

Figure 3.

 Average accumulative flexion extension ROM.

The maximum articulation for the entire spine for flexion (+) and extension (−) was +193.7 ° and −225.6 °, respectively, with 45.9, 21.5 and 32.6% occurring in each of the cervical (C0–T1), thoracic (T1–T12) and lumber (T12–S1) regions. The percentages of flexion and extension were different in the cervical (+48.4 and −52.6%), thoracic (+69.0 and −31.0%) and lumbar spines (+28.2 and −71.8%). During lateral bending the maximum total rotation was 208 ° (each side), with 29.9, 39.4 and 30.7% occurring in each of the spinal regions. During axial rotation the maximum total rotation was 255.7 ° (50% distributed to each side), with 38.5, 38.7 and 22.8% occurring in each of the spinal regions.


This work set out to develop a normalized digital 3D model of the human spine and pelvis using a wide range of scientific literature. The model has been presented; however, the process was challenging due to lack of standardization and/or inconsistency with:

  •  anatomical terminology;
  •  scaling of measurements;
  •  measurement methodology, dimensioning and anatomical reference positions;
  •  global coordinate systems.

Anatomical terminology

Data to create digital anatomical models for CAD reconstruction must have clear and replicable geometric references. However, there is a wide variety of terminology describing anatomical landmarks in the literature. This was particularly pertinent when describing the vertebrae. For the purpose of this study, to provide consistency and clarity, abbreviations for each dimension were modified in line with the majority of the literature (Tables 1 and 2). Relating to these redefined terms, an illustration of the measurements was developed (Fig. 4) modified from the methods of Panjabi et al. (1991a,b, 1992). In other circumstances, similar terminology was used for entirely different anatomical measurements. For example, the pelvic tilt normally referred to the angle between the horizontal plane and the arcuate line, which connects the inferior, anterior corner of the S1 endplate to the superior midpoint of the acetabulum (Twomey, 1979; Bogduk & Twomey, 1987; Ellis & Dussek, 1995; Troyanovich et al. 1997; Harrison et al. 2001). However, some measured an angle between the vertical and a line connecting the S01 endplate midpoint to the hip joint center (HJC; Marty et al. 2002), which, although a useful measurement, was inconsistent with others (Fig. 5). Ferguson’s angle is also referred to as the sacral slope. When describing the location of a bony landmark, consistency was equally important, but this was not always the case. For example, although most literature used the terms superior and inferior, some (Doherty & Heggeness, 1995) used ‘upper’ and ‘lower’. Another problem occurred when comparing measurements of the posterior vertebral body height (VBHp). Some literature measured this dimension on the sagittal midline using cadavers (Panjabi et al. 1991a,b, 1992; Tan et al. 2004), while others measured the extremity of the vertebral corner points on an X-ray image (Nissan & Gilad, 1984). Therefore, the use of the suffix ‘c’ for central position was introduced, providing a distinction in height that exists due to the concave nature of the vertebral endplates, which is most predominant in the cervical spine (Fig. 6). Standardization of terminology would prevent misinterpretation of information.

Table 1. Vertebrae dimension descriptions.
ASHArticular surface height
C1FacC01 Facet
DIDens instance
DB-Disc body
DHDens height
DDDens diameter
EPLHEndplate lip height
FDForamen diameter
LMLateral mass
SP-Spinous process
TFTransverse foramen
TDHTotal dens height
TPTransverse process
VB-Vertebral body
V-indVertebral indentation
VR-Vertebral ring
Table 2. Vertebrae measurement descriptions.
1ASHArticular surface heightC2
2C1Fac-DC01 facet depthC1
3DBHaDisc body height anteriorC2–L5
4DBHpDisc body height posteriorC2–L5
5DDDiDens diameter depth inferiorC2
6DDDsDens diameter depth superiorC2
7DDWiDens diameter width inferiorC2
8DDWsDens diameter width superiorC2
9DHDens heightC2
10DIDens instanceC2
11EPDiEndplate depth superiorC2–L5
12EPDsEndplate depth superiorC2–L5
13EPI-tiEndplate instance- tilt inferiorC2–L5
14EPI-tsEndplate instance- tilt superiorC2–L5
15EPLHEndplate lip heightC2–L5
16EPWiEndplate width inferiorC2–L5
17EPWsEndplate width superiorC2–L5
18FDDForamen diameter-depthC1–L5
19FDWForamen diameter-widthC1–L5
21LAa-VRaLateral mass anterior to vertebral anteriorC1
22PDIPedicle diameter instanceC3–C7
23PDWPedicle diameter widthC3–C7
24SPISpinous process instanceC2–L5
25SPLSpinous process lengthC2–L5
26SPL-piSpinous process length posterior inferiorC2–L5
27SPL-psSpinous process length posterior superiorC2–L5
28SPWSpinous process widthC2–L5
29TDHTotal dens heightC2
30TPWTransverse process widthC1–L5
31TP-LMTrans process to lateral massC1
32VBHaVertebral body height anteriorC2–L5
33VBHpVertebral body height posteriorC2–L5
34VBHpcVertebral body height posterior centerC2–L5
35V-indVertebral indentionC2–L5
36VRDVertebral ring depthC1
37VRHaVertebral ring height anteriorC1
38VRHpVertebral ring height posteriorC1
39VRTaVertebral ring thickness anteriorC1
40VRTpVertebral ring thickness posteriorC1
Figure 4.

 Illustrated terminology of the vertebrae, modified from Panjabi et al., integrating other investigator measurement dimensions (a – C1 vertebral measurements; b – C2 vertebral measurements; c – C3 to C7 vertebral measurements; d – T1 to L5 vertebral measurements). Articular facet dimensions are not included, due to gaps in the literature data relating to specific geometric location, but should be considered in future standards.

Figure 5.

 Pelvic measurement inconsistency [A – pelvic tilt example (Harrison et al. 1998); B – pelvic tilt example (Marty et al. 2002); C – Ferguson’s angle/sacral slope).

Figure 6.

 Two types of posterior height dimensions in lumbar vertebrae.

Scaling of measurements

There are large variations in measurement of the VBHp in the literature. Several researchers have compared their measurements with others, whilst not including the effect of human scale (Krag et al. 1988; Scoles et al. 1988; Panjabi et al. 1991a,b, 1992; Doherty & Heggeness, 1995; Harrison et al. 2003a). We compared the average measurements of vertebrae for a range of subject groups with published statures, which were discovered to range from 1678 to 1754 mm. This was a 4.5% variation and resulted in a variation of 2 mm in vertebral height. A similar observation was identified by other researchers as an issue across clinical research, specifically for vertebral measurement. ‘Dimensions of cervical vertebrae and discs have been documented, but results were quoted in millimeters and thus cannot be compared among radiographs taken in different settings and subjects of different stature’ (Frobin et al. 2002a). This issue is compounded by the limited documentation of scaling factors within individual studies. They recommend comparing vertebral height with depth ratios, rather than Standard International (SI) units of measurement. This proposed method is not common practice, and measurement using SI units such as millimeters remains important for most real-life applications and was the practice in all of the reviewed literature. Scaling of anatomy relative to other body measurements was not common in anatomical studies, but was evident elsewhere and provides a similar rationale to Frobin’s protocol. Within ergonomics, stature is used as a scaling reference (Pheasant, 1988; Peebles & Norris, 1998). In gait analysis to predict hip centers, both stature and leg length are used (Hof, 1996; Pierrynowski & Galea, 2001; Moisio et al. 2003), and in forensic science spine lengths of partial corpses are used to predict stature (Jason & Taylor, 1995). Recently, the forensic science methods for estimating stature from corpses have been debated comparing the reliability of three methods. These include ‘anatomical reconstruction, regression based on long bone lengths, and measuring skeletal vertex–talus length in the grave for individuals buried in a supine position’. These methods were observed to have inaccuracies requiring an additional 25 mm in addition to skeletal length reaching a more representative stature estimation (Petersen, 2011). Research has also indicated that stature and vertebral size were proportionally scalable (Ruhli et al. 2005). Several publications that provided quantifiable measurements of selected vertebrae did not have scaling references and could not be considered in the present research (Cotterill et al. 1986; Berry et al. 1987; Krag et al. 1988; Scoles et al. 1988). However, all studies reviewed, including these, showed larger VBH dimensions to Panjabi in the region of 1–2 mm, and were more closely related to using X-ray methods (Harrison et al. 2004). Again this supports the rationale for Panjabi’s dimensions being measured at a central point rather than extremities.

Measurement methodology, dimensioning and anatomical reference positions

For spinal measurements, some investigators clearly described their measurement protocol and the anatomy measured (Dempster, 1955; Panjabi et al. 1991a,b, 1992). Panjabi also specified a reference plane for alignment when measuring vertebrae, with the posterior body height positioned vertically. Others used a diagonal line connecting the opposite corners of a quadrilateral, representing the sagittal view of the vertebrae, to provide geometric structure. In each of these models the measurement of the spinous process also differed (Fig. 7). Some researchers did not provide enough detail on the exact location of measurements. For example, a study of the C1 lateral mass had no reference anatomy identifying its relative position to other anatomy, but provided an isolated set of measurements (Dong et al. 2003). It was not possible to include such data in this digital model without the positional context. In the absence of a standardized list of recommended anatomical measurements, some anatomy of clinical importance, such as intervertebral foramen width, has been ‘rarely documented’, as observed in back pain research (Ruhli et al. 2006). In the pelvis in particular, variations of anatomical reference points existed, for example, some investigators used the anterior superior iliac spine (ASIS) position, while others used the HJC as the primary reference point for sacrum location. In some cases pioneering pelvic research (Martin, 1957 cited in Patriquin et al., 2005, p120) was referred to as using standard measurement techniques (Patriquin et al. 2005); however, across the four measurements they publish, two additional techniques were required from separate authors. A gap in measurement technique for the posterior border of the hipbone was identified (Issac, 2002), referring to Martin (1957, 1959) and 18 other publications. Issues in measurement of the sciatic notch were discussed (Singh & Potturi, 1978), comparing the methods of Martin (1957) with five other authors, with two different measurement techniques being identified. Some referred to the well-established definitions of vertebral geometry by Martin (1928), but also identified measurement gaps relating to the various vertebral process geometries (Ruhli et al. 2005). Graphical reconstruction of data was often difficult because auxiliary dimensioning (one without positional reference) was commonly used for anatomical measurements on cadavers (Panjabi et al. 1991a,b, 1992; Bailey et al. 1995; Doherty & Heggeness, 1995) or on CT images (Zhou et al. 2000; Naderi et al. 2003), and sometimes oblique (two connecting points in 3D space) and orthographic measurements (projected measurements to X, Y, Z planes) were combined together (Caldwell & Molloy, 1933, 1940; Issac, 2002), which, although useful in 3D construction, are more difficult to understand in orthographic projection. However, this is presently the normal method of viewing or comparing measurements of the anatomy. Several of these studies randomly used ‘chain dimensioning’ (adjacent dimensions, each starting where the previous one ended), which produced cumulative tolerance errors. In design, the preferred dimensioning method to avoid tolerance issues is ‘parallel dimensioning’, where all dimensions originate from a standard reference point (BS 8888 2011). If a combination of chain and parallel dimensioning cannot be avoided, it may be necessary to include dimensioning by coordinates and a table, so that 3D world spatial context is fully understood. These methods are practiced in industrial design, engineering and, to an extent, in ergonomics, but not for anatomical measurements. In the present research oblique measurements were calculated using trigonometry to find their orthographic values for the X, Y, Z planes, before normalization and construction using CAD. Some of the pelvic anatomy was known to have large variation between individuals, including the iliac crest, the sacro-iliac notch, the Ischia, and some of the pubis anatomy. They were more difficult to construct in CAD as the information provided was not quantitative, but rather descriptive locations and therefore guidelines only.

Figure 7.

 Variation in geometric representation of vertebrae measurements (a: Nissan and Gilad, 1984; b: Panjabi et al. 1992).

Global coordinate systems

Several industrial and clinical protocols for global coordinate systems were reviewed from the physical perspective of the measured object (Fig. 8; Table 3). International standards exist within automotive design (BS 173 1979, BS 2631-4 2001) referring to the orientation of the vehicle and the occupants within them. No standardization has been firmly established for clinical applications, with researchers using very different methods (White & Panjabi, 1978, 1990; Stokes & Gardner-Morse, 1999). However, for the study of biomechanics, the International Society of Biomechanics (Wu et al. 2005) and Information Society Technologies (Hilhal et al. 1999) proposed standards that have been integrated within gait software technologies. These followed the right-handed coordinate systems. In some medical visualization and analysis software difficulties in dealing with the various coordinate systems has been identified as a common issue (3D Slicer 2011). These include world, anatomical (or patient) and image coordinate systems. Although the world coordinate system uses a Cartesian reference system, the anatomical system uses anatomical planes (axial, coronal and transverse) and the image system uses voxels (i, j, k). The Digital Imaging and Communications in Medicine (DICOM) standard (NEMA 2012), which relates to distributing and viewing medical, is similar to 3D Slicer in that a Z-up anatomical system is used; however, the X and Y directions are inverted when compared. Within software applications either right-handed or left-handed systems were used, respectively, for software such as SGI’s OpenGL (Silicon Graphics 2008) or microsoft window’s directx (Microsoft 2007). The specific orientation of the coordinate system within 3D modeling software tended to be either Y-up (developed for those traditionally from 2D image sources, such as animators, graphic designers) or Z-up (developed for those with 3D practice experience, such as automotive designers, architects and engineers). The literature suggests that ‘most 3D graphics languages and all maths texts’ use the right-handed system (Fosner, 1997). Some software manufacturers have managed this by controlling the OpenGL structure under a Z-Up orientation. The H-Anim standard (ISO/IEC 19774 2006) currently specifies guidelines to allow a humanoid character constructed in one VRML 97-compliant software environment to be read in another, and recommended the right-handed Y-up orientation. A standard for digital models exists (BS 15536-2 2007, BS 15536-1 2008); it recommends a Z-up coordinate system with X forward (from the perspective of the object). This is similar to several other methods. It recommends specification of 3D rotation values for spinal flexion (+), extension (−), lateral bending to the right (+) and left (−), although axial rotation of the spine is not weighted with directional values – this could be a future consideration. Importantly, recommendations were also made to provide consistency with the operational software within which the manikin exists. In the case of the present research 3D Studio Max, which prefers the Z-up system, was used (Woods et al. 2001). The Z-up X forward orientation is counterintuitive in this environment, for example in plan view the +X axis points downwards in the negative quadrant. Its default setting is Z-up, with Y backwards, and was preferred for the present research.

Figure 8.

 Right-hand rule (X-up orientation).

Table 3. Reviewed global co-ordinate systems, observed from the position of the object.
  1. *A positive value donates an increase in the voxel intensity.

BS EN ISO 15536 1/2+Z+Y+X
BS ISO 2631+Z+Y+X
ISB 2002+YZ+X
Stokes, 1999+Z+Y+X
VICON system+Z+Y+X
White & Panjabi, 1978+Y+X+Z
HBE Ivancevic, 2008+ZX+Y
Fonar upright MRI+Y+X+Z
3D Studio Max+Z+XY
3D Slicer: WorldYX+Z
3D Slicer: Anatomical - right anterior, superior (RAS)Inferior to superior (+Z)Left to right (−Y)Posterior to anterior (+X)
DICOM: Anatomical - left, posterior, posterior (LPS)Inferior to superior (+Z)Right to left (+Y)Anterior to posterior (−X)
Image system*−j−i−k

A limitation of the model is that some measurements within the pelvic region could have had different interpretations in CAD, due to the descriptive nature of anatomy rather than referenced to a specific location; nevertheless, such interpretations had no relationship to anatomy that was regarded to be more clinically relevant. Descriptive anatomy is useful, assisting comprehension of geometry; however, for the purpose of digital modeling a quantitative and positional context is also required.


This research proposes that there are important requirements regarding collaborative understanding of methodologies, standards and for the effective completion of connected or interdisciplinary research. Several areas of suggested standardization were identified that are based on the integration of industrial design and engineering methodologies earlier in scientific research, as published by other researchers (Driver et al. 2011).

Recommendations include the following.

  •  Anatomical terminology requires standardized agreement and consistency in the context of specific measurements between different anatomical locations.
  •  Establishing a scaling reference such as stature to enable comparative analysis.
  •  Measurement protocol should include the identification of agreed reference points and spatial positioning information. Dimensioning protocol – an improved method to current practice for documenting anatomical measurement – would be in line with the Technical Product Specification and Documentation standards (BS 8888 2011), specifically adopting parallel dimensioning relative to an origin or standard anatomical reference position. If it is necessary to use chain dimensioning, it should relate to a parallel dimension for spatial context. A standard reference position for each part of the anatomy is suggested. When auxiliary measurements are dimensioned, the spatial position should be contextualized using Cartesian coordinates and a table. This provides a measure of orientation that was not always evident in the literature.
  •  Global coordinate systems both for conducting measurement studies and digital modeling should be considered in line with British Standards for Computer Manikins (BS 15536-2 2007).

Standardization of anatomical reference anatomy is important and at least one should be included in any quantitative anatomy study. These are identified in the sagittal and coronal planes. In the sagittal plane, different reference positions relate to specific regions of anatomy.

  • S1 anterior, superior corner. The primary reference for the sacrum, ASIS, HJC and the entire spine locations, relating to spinal alignment (Harrison et al. 2000b). The S1 endplate should be orientated at a set angle (horizontal, vertical or at 45 º), which allows for clearer comparison between independent researchers’ results.
  • C7 spinous process tip: to identify the cervico-thoracic junction.
  • Individual vertebrae – the inferior (primary) and superior (secondary) posterior points at the vertebral body center (or on an X-ray projected image). These locations provide references for the vertebral endplates that are used for Cobb angle measurements (Yochum & Rowe, 1987), and their angles determine disc height (Frobin et al. 2002b). They relate to George’s line, which is required for the Harrison posterior tangent measurement method (Harrison et al. 2000a) and has other clinical relevance (Yochum & Rowe, 1987). The vertebral posterior height is commonly used as an indicator of vertebral size, having the least variable morphology. This reference measurement contrasts the ‘sagittal and transverse diameters of the vertebral bodies, as well as the pedicle height’, which increase in size with age (Ruhli et al. 2005). Furthermore, as illustrated in Fig. 7, the interconnecting line between these points provides a vertical reference plane for spatial orientation, which is most useful during digital reconstruction.

In the coronal view for both the pelvic and vertebral anatomy, the center line is suggested as the primary reference plane, at the level of the aforementioned sagittal references at each part of the anatomy. As absolute symmetry of anatomy is highly unlikely, any measurement lateral to the center line should be quantified in both positive and negative values in line with the current related standards (BS 15536-2 2007).


Anatomical geometry is unique between individual people, and some measurements have little clinical relevance. Certain applications require only specific isolated measurements, and may have rationale without additional spatial context. However, for the development of reference or normative models, for which much research exists, consideration is required regarding the ability to reconstruct data. Presently, the measurement techniques used in the majority of clinical literature may include elements of bad practice for the reasons outlined. As digital modeling of anatomy becomes increasingly important to simulating and understanding the human body, so too do requirements to enable reliable geometric reconstruction. Standardization regarding skeletal anatomy measurement involving inter-disciplinary dialogue is essential. In order to formalize standardization, the support of the BSI Engineering Standards Committee ( would be appropriate. However, as this standard has limited commercial benefit, being more research relevant, a publically available specification would be appropriate. Research Council funding would be required, establishing a multidisciplinary working group, to agree best practice, illustrated terminology and to commission the standard.


Thanks to: Lyman Jellema (Collections Manager, Hamann-Todd collection, Cleveland Museum of Natural History, USA) for providing original museum spreadsheets with essential stature and ethno-geographic data not included in published literature; Dr Deed Harrison of Chiropractic BioPhysics (CBP, USA) for discussions on their elliptical modeling and ideal spine theories, and for providing specific publications on spinal geometry; Karen Bergkamp (ErgoTech, South Africa) for assistance with South African ergonomics data and further information on the demographic of the Terry-Dart collection; and Dr Gary Heathcote (Anthropology Research Centre, University of Guam, USA) for providing unique sacral index data from literature (Martin, 1928), which was difficult to source.