Expedited Extraction of 2D Bone Porosity Descriptors from 3D Micro-CT and Serial Section Images

Thumbnail Image



Journal Title

Journal ISSN

Volume Title


Research Projects

Organizational Units

Journal Issue


Anthropologists have extensively studied how physical activity changes the shape of the bones used for those behaviors (Ruff et al., 2006). When mechanical demand changes from an optimal level, bone is formed or resorbed (removed) until it restores the balance of bone meeting mechanical needs. Bone forms on surfaces where demand is increased, and bone is resorbed on surfaces where demand is decreased (Lanyon, 1982). The cross-sectional shapes of limb bones in particular have been broadly tracked across fossil and archaeological populations to detect trends in bone strength associated with the evolution of bipedalism, transition to agriculture, division of labor, roughness of terrain, and tool use (30+ studies reviewed in Ruff and Larsen, 2014). The relationship between mechanical demand and bone shape becomes more complicated when anthropologists examine the internal structures of bone tissue. The cortical "walls" of bone are perforated with hundreds (in the rib) to thousands (in the femur) of canal systems known as vascular pores, which carry the blood vessels that supply bone cells (Agnew and Stout, 2012, Cole, 2014). Vascular pores form through a mechanically-stimulated process of bone turnover called "remodeling", which involves osteoclast cells tunneling into bone tissue, and osteoblast cells filling in all of this tunnel except a central pore (Burr and Akkus 2014). Remodeling is triggered under three circumstances: 1) mechanical demand decreases, so pores form to remove "excess" bone, 2) mechanical demand increases, damaging the bone, so the microscopically cracked tissue is removed through remodeling, or 3) the cellular processes underlying remodeling become de-sensitized and dis-regulated with age, leading to increased porosity beyond what is mechanically stimulated (Agnew and Bolte, 2012, Burr and Akkus, 2014). Since these contradictory mechanical stimuli can all lead to porosity, anthropologists have sought to untangle the relationship between mechanical demand and pore number, shape, and location. Age-associated increases in porosity are also a key marker of bone fragility – decreased bone strength and increased fracture risk – both in archaeological (e.g. Cho and Stout, 2002) and modern populations (e.g. Yeni et al., 1997). At least 76% of the reduction in cortical strength with age results from porosity (McCalden et al., 2003). This is because vascular pores are stress concentrators where microscopic cracks can initiate and then propagate into a spontaneous fracture (Reilly and Currey, 1999, Ebacher et al., 2007). One in three women and one in five men over the age of 50 will experience at least one fracture due to severe bone loss (osteoporosis) (Melton et al., 1992, 1998, Kanis et al., 2000). Traditionally, anthropologists have examined porosity by cutting a ~100 micrometer thick cross-section of a bone. The cross-sections of pore systems on this slice of bone are then counted and measured (Agnew and Stout, 2012) However, pore networks are highly complex in three-dimensions, frequently splitting, merging, and branching at varied angles. In two-dimensional cross-section, pore shapes and numbers are distorted (Stout et al., 1999; Bell et al. 2001). Traditional histological slide preparation also involves grinding away several hundred micrometers of structural information between adjacent cross-sections, making it impossible to accurately reconstruct and follow pore structures as they change along a bone (Cho, 2012). Consequently, previous studies have not agreed on how pores respond in number, size, and orientation to different mechanical demands in different bone types or regions (reviewed in Stout et al. 1999 and Gocha and Agnew 2016). Additionally, manual selection and measurement of hundreds to thousands of individual pores is time-intensive (Agnew and Stout 2012, Cole 2014). This presentation will demonstrate two new automated or near-automated techniques that extract two-dimensional pore measurements from exactly adjacent slices of bone, without a loss of three-dimensional information. The sample consisted of adjacent regions of the midshaft of a cadaveric right-side fourth rib from a 72 year old human female. The first technique, high-resolution micro-computed tomography, involved scanning a 1 cm long rib segment using a Skyscan 1172-D High Resolution Desk-Top Micro-CT. Pore spaces were automatically extracted in three-dimensional space through the Segmentation Editor of AvizoFire 8.1 and exported as series of two-dimensional slices. Each slice represented an adjacent region of the bone length, with thickness 4.88 µm. The second technique, reconstruction from serial sections, involved decalcification in 14% EDTA and freezing cryostat serial sectioning of an immediately adjacent 1 mm long rib segment. Due to the thinness of each serial section (30 µm), pore spaces were empty with distinct boundaries, facilitating automated extraction of the majority of pores through thresholding of pixel intensity, particle size, and particle circularity in ImageJ, and near-automated extraction of section boundaries and the few remaining pores by adjusting wand tool pixel value tolerance. Both microCT and serial section image sequences were loaded into ImageJ, which reported descriptive statistics for pore number, area, and circularity. Average sectional pore mean area and percent porosity were similar between micro-CT (mean area = 0.00560 +/- 0.00262 mm2; % porosity = 5.59 +/- 0.05%) and histological technique (mean area = 0.00444 +/- 0.000771 mm2; % porosity = 5.71 +/- 1.9%). Micro-CT serial sections on average reported fewer, more circular pores (130 +/- 6.58 pores; 0.877 +/- 0.0132 circularity) than histological technique (167 +/- 8.98 pores; 0.709 +/- 0.050 circularity), as micro-CT can better incorporate tissue separating or edging pores. Both techniques preserve three-dimensional structural information and minimize sample processing time. For both methods, adjacent serial sections can also be interpolated to reconstruct the three-dimensional structure of the pore network.


Social and Behavioral Sciences; Social Work; Law: 2nd Place (The Ohio State University Edward F. Hayes Graduate Research Forum)


bone histology, porosity, micro-computed tomography, histological serial sections, biological anthropology