Central quantitative computed tomography (QCT) of the spine was first described in the late 1970s [26], and became more widely used during the 1980s [27]. With the introduction of dual energy X-ray absorptiometry (DXA) in 1988 the use of QCT declined. However, there has been renewed interest in QCT as investigators recognise the importance of bone size and geometry in assessing pediatric bone status. QCT is particularly useful in children since it measures volumetric density (g/cm³), which is not as size dependent as areal density (g/cm²) measured by DXA. Recent technical developments (such as spiral and spiral multi-detector MDCT) add to the potential information available from QCT. While QCT for clinical and research purposes may increase in the future, it should be noted that the radiation dose from QCT is substantially higher than that of DXA, but lower than in CT of the torso used for other clinical indications [28–32].

QCT of the spine requires that the patient lie supine on the scanner table with the legs flexed and supported on a pad to flatten out the natural lumbar lordosis (Fig. 2.3a). The height of the scanner table should be kept constant. A bone mineral equivalent phantom is placed under the patient in the site to be scanned (Fig. 2.3b). A water or soft tissue equivalent pad should be placed between the patient and the phantom if there is a significant air gap (Fig. 2.3b). Solid hydroxyapatite calibration phantoms are currently used with regions of varying density that allow the transformation of measurements in Hounsfield units (HU) to bone mineral equivalents in mg/cm³. Some CT manufacturers provide their own software and phantoms (e.g. Siemens Healthcare, Erlangen, Germany); alternatively software and phantoms can be purchased (e.g. Mindways, Austin, Texas, USA). For comparable results in longitudinal studies the same phantom (and scanner) should be used. Similar to DXA instrumentation [33, 34] if scanners or phantoms have to be changed during longitudinal studies then cross-calibration with patients and a phantom, such as the European Spine Phantom (ESP) [35], must be undertaken to make results comparable, although the ionising radiation involved makes duplicate measurements in children problematic (Table 2.2 ).

For two-dimensional (2D) spine measurements , an initial lateral scan projection radiograph is obtained (Fig. 2.3c). A 10 mm section is then performed through the mid plane of the vertebrae to be measured and parallel to the vertebral endplates. The section is confirmed to be in the correct plane when the area of the basi-vertebral vein is identified (Fig. 2.3b). For 2D QCT in adults, generally three vertebrae are scanned (L1 to L3) to ensure that at least 2–3 vertebrae are available for analysis at follow-up scans, should it be necessary to exclude vertebrae that have fractured between measurements. Since vertebral fractures occur less commonly in children and to minimise ionising radiation dose, generally only two adjacent vertebrae (between L1 to L3) are scanned. Vertebrae should be matched to those scanned in the reference database used, because BMD differs between vertebrae. If longitudinal studies are performed it is essential to scan the identical vertebrae examined at baseline.

QCT results are expressed as a mean volumetric bone mineral density (vBMD) in mg/cm³. The trabecular vBMD measured by QCT is a composite of the amount of bone and marrow per voxel. The reason for this is the relatively small size of trabeculae compared to the voxel, resulting in marrow being included in the measurement. Because marrow fat is limited in children, age related marrow changes in fat composition should not confound spinal vBMD measurements in children as significantly as they do in adults [36].

Over the past two decades advances in CT technology and the introduction of multi-detector spiral CT (MDCT) enable rapid scanning so that several vertebral bodies are imaged within less than a minute. This technology has been used for volumetric QCT, which provides 3-D volume analysis of the vertebral bodies (mostly L1-L3). These developments improve precision (coefficients of variation of better than 1 %) and have advantages in children by reducing movement artefacts. In 3D MDCT sagittal midline reformatted images are obtained to define the mid-plane section to be analysed (Fig. 2.3d). Then, on the selected mid-plane cross-sectional image an oval region of interest (ROI) is placed in the central trabecular bone of the vertebral body at the site of the entry posteriorly of the basi-vertebral vein. The ROI should be as large as possible but not include the cortical rim of bone. As quantitative skeletal assessment does not require the optimisation of image quality needed for conventional CT, a low-dose technique can be employed to minimise radiation dose [37, 38] (Table 2.2). The results are expressed as standard deviations (SD) from the mean for appropriate age-, race- and sex-matched reference data (Z-score) . The most frequently used normative data for spinal QCT were reported by Gilsanz and colleagues in 2009 [30, 31, 39, 40] (Table 2.3).

QCT offers several advantages as a densitometric technique . Whereas DXA measures integral (cortical and trabecular) BMD, QCT provides separate measures of cortical and trabecular BMD. As trabecular bone is generally more metabolically active than cortical bone, trabecular vBMD as measured by QCT is more sensitive to change in BMD [41]. The BMD provided is volumetric and less influenced by bone size than DXA, which provides an ‘areal’ density (g/cm²). CT also provides true morphometric dimensions of bones, and in the shafts can measure cross-sectional area of bone, cortical thickness and density, and periosteal and endosteal circumference. These parameters can be used to calculate estimates of biomechanical bone strength including moment of inertia. QCT also has the potential to be applied to peripheral skeletal sites such as radius, tibia and mid femur with lower associated radiation exposure than spinal QCT [30, 42, 43]. In addition QCT studies can be used to perform finite element analysis allowing the calculation of bone strength [44].

The limitations of QCT include the substantially greater dose of ionising radiation than DXA for spine scans (Table 2.2). For single slice (2D) QCT low-dose protocols using 80 kVp (or 120 kVp) and 120 mAs (or 150–200 mAs) result in effective doses of less than 200 μSv [32]. Felsenberg et al. described a low-energy, low-dose protocol with 80 kVp and 146 mAs resulting in effective doses down to 50–60 μSv, including the preliminary scan projection (scoutview) radiograph [45]. In comparison DXA has radiation doses in the range of 5–13 μSv for the lumbar spine and 9 μSv for the hip while an AP lumbar spine radiograph has a dose of 700 μSv and a standard abdominal CT has an exposure dose of approximately 8000 μSv [46]. For volumetric (3D) MDCT QCT radiation doses are higher and have been estimated to be as high as 1.5 mSv for the spine and 2.5–3 mSv for the hip [47]. Access to QCT may be problematic because many radiology departments lack the appropriate phantoms and software to perform bone densitometry studies. Furthermore, CT equipment is in great demand for other clinical diagnostic purposes. Currently there is a dearth of commercial analysis packages for QCT that can be purchased which require little set-up (for example Mindways, Austin, Texas, USA), Therefore, some centres have resorted to developing their own analysis software [48]. As with other bone densitometry techniques, QCT requires skilled and dedicated technical staff to perform the scans and optimise precision. Finally, there are far fewer published pediatric reference data for QCT than for DXA (Table 2.3) [30, 39].

More recently Habashy et al. investigated the feasibility and potential limitations of estimating BMD from standard diagnostic CT studies. They compared BMD measurements obtained with and without an external calibration phantom and found that phantomless QCT adds clinically useful BMD information to standard diagnostic CT studies [49]. The preliminary scout view should always be scrutinised for the presence of vertebral fractures (Fig. 2.3e).

In summary, and based on the general consensus of the ISCD QCT Task Force and working group [30]: the most common sites scanned on whole body CT scanners are the lumbar spine (L1 to L3), mid femoral shaft and the tibia. In the long bones, a metaphyseal and diaphyseal site should be scanned, and ideally in longitudinal studies the whole bone using 3D MDCT QCT should be scanned. This will enable results from similar anatomical sites to be compared which may be problematic in growing bones in 2D pQCT. Also, lying on the table of a general purpose scanner may afford easier positioning of children with disability (e.g. cerebral palsy, Duchenne muscular dystrophy) and rapid scanning (20 s) reduces movement artefact. The proximal femur (hip) should not be scanned as it is in adults as the ionising radiation dose is high (3 mSv). Currently, adult scan protocols are being used in children, and there is an urgent need for development of pediatric protocols to minimise radiation dose. More robust reference data and analysis tools are required for the wider applicability of the technique.

Peripheral quantitative computed tomography (pQCT) first became commercially available in the early 1990’s [50–52]. The most commonly used pQCT scanner (XCT 2000, Stratec, Pforzheim, Germany) utilises the original rotate-translate CT technology, which generates only single 2D slices (1–2 mm thick) and requires a long acquisition time of about 1 min to obtain a single slice.

The sites of measurement are the radius, tibia and femur. For clinical assessment of a child’s bone , the most commonly used site is the distal 4 % of the forearm or tibia length proximal to the distal growth plate. Peripheral QCT is most commonly applied to the non-dominant forearm . The forearm length is measured as the distance between the tip of the ulnar styloid and the olecranon . The forearm is placed pronated in the pQCT gantry with the elbow resting on a block and the hand gripping the hand fixture and the arm is secured with Velcro straps to prevent movement (Fig. 2.4a and b). In children it is important to avoid the section including the growth plate, which produces falsely high measures due to the zone of provisional calcification. To locate the appropriate scan slice, a scanogram is performed (Fig. 2.4c and d). A reference line is placed to bisect the medial border of the end of the distal radius in adults. In children if the growth plate is visible, the reference line is positioned to bisect the medial border of the distal dense metaphysis (Fig. 2.4c). If the distal radial growth plate has fused, the reference line is placed to bisect the medial border of the distal articular surface of the radius (Fig. 2.4d). For the tibia the reference line location varies but is usually placed on the metaphysis and again the scanner moves to the measurement site from this point (Fig. 2.5) . Upon closure of the growth plate at skeletal maturity the distal surface of the epiphysis joint margin is used for placement of the reference line. Radial abnormalities such as Madelung’s deformity (disturbance of growth of the medial component of the distal end of the radius which causes reduction in the carpal angle) may cause difficulties in positioning the reference line. In children treated with bisphosphonates, the reference line should be positioned to try to ensure that the growth arrest lines (residue of the provisional zone of calcification) are avoided in the measurement which will be artefactually elevated [53, 54].