It may not always be possible to achieve the ideal scan due to marked scoliosis or vertebral collapse. Figure 5.2 illustrates spine scans from four children with varying degrees of spinal curvature. Both Child (A) and Child (B) have relatively mild deformities, such that scan acquisition and patient positioning are minimally affected; adequate analysis of these patients is possible by rotating or slanting the intervertebral markers . Unfortunately, this is not the case for Child (C) and Child (D). Child (C), a young child with cerebral palsy, has a scoliotic and rotated spine, such that the projection of the lower lumbar vertebra appears almost lateral. Although this scan can be analyzed, it is not suitable to compare it to a “normal” reference database since the vertebral areas will not be comparable to normal spine areas. Case (D) also highlights a child with marked scoliosis and vertebral collapse affecting the entire lumbar vertebrae. The degree of collapse and curvature of this child, who has type III osteogenesis imperfecta, makes this scan difficult to analyze and to interpret. In both of these two latter cases, either a whole body and or a radius scan may be more informative.

The spine is a useful site to monitor changes in bone mass. However, to achieve successful follow-up scans the operator must:

If there have been significant weight changes between scans, these ideals may not be possible. For weight changes that result in scan mode variation, the mode change should be recorded so that any necessary corrections and other considerations can be made. When weight changes places the child at the borderline between scan modes, scanning in both standard and pediatric modes is recommended. This creates a comparable scan for the previous measure and a new baseline scan for any future follow up. When making these decisions, it is important to consider the additional radiation exposure from repeat scans.

Use of the auto-low density analysis method (Hologic Discovery / Horizon) will allow the results to be compared to a large pediatric reference database collected using this software [6].

Figure 5.3 illustrates a successful series of measurements over a 2-year period beginning at age 16 in a boy being monitored for the effects of three monthly intravenous bisphosphonate treatments.

The whole body is also a preferred site in children. This scan provides measurements of total and regional bone and body composition parameters, making it a useful site for both clinical and research purposes. Growth and disease may affect both bone and body composition values [22].

With older-generation pencil beam densitometers, whole body scan times were as long as 10–20 min. However, with newer fan beam and narrow fan beam machines, scan times have been reduced to a few minutes, thus making it far more feasible to acquire a whole body scan even on a young or fidgety child [23].

Although analysis of specific skeletal regions can be performed from the whole body scan, the precision is relatively poor compared to the good precision for the whole body scan [24].

When acquiring a whole body scan, it is important that the child not wear any high-attenuating objects. Ideally, the child should be scanned in a hospital gown or in light indoor clothing. The operator should be aware that thick elasticized waist bands and plastic buttons may also cause problems with image artifacts. Additionally if body composition is to be calculated, polyvinyl chloride (PVC) sheets or pillows, as well as sand bags used in positioning, will affect the calculations and should therefore be removed from the scanning table. If any cushions or towels etc. are needed for acquisition due to patient comfort, these must be noted and used again for all follow-up scans.

Whole body scanning can be performed in children with internal high-attenuating objects (e.g., metal rods, pins, or plates) if they are likely to remain in situ for follow-up. However, special attention should be given to the analysis and interpretation of such scans, especially when attempting to compare them with normal data.

In positioning a child for a whole body scan, the following steps should be taken:

Figure 5.4 demonstrates acceptable scans of (a) a 5-year-old child with osteogenesis imperfecta, (b) an 11-year-old child with Diamond Blackfan anemia on chronic steroid therapy, and (c) a 16-year-old with Crohn’s disease.

For small children , the length of the scan field may be adjusted to reduce the scan time. However, this may become problematic when comparing scans at follow-up as the child grows larger. For very tall adolescents, it may not be possible to fit the entire body in the scan field; therefore, it is suggested to position the child with his or her head is just below the top of the table and with the feet flexed upward. If the child is still too long for the scan table, the scan should be acquired by excluding the feet from the scan area.

When performing scans on obese adolescents, it can be difficult to position them so that the entire body is in the scan field. Several techniques can be used in this situation, depending on the fat distribution. With centralized obesity, the elbows may be too close to the edge of the scan field. A folded cotton sheet can be wrapped tightly around the middle portion of the body to hold the elbows close to the body. In this case, care should be taken to keep the palms flat on the DXA table.

When these techniques fail, an alternative approach, if the scanner has the software, is to perform a “hemi” scan (Fig. 5.5d). For a hemi scan, position the patient off-center so that the entire right side of the patient and the shoulder and hip of the left side are within the scan area. The measured values for the right side are then used to estimate the bone and body composition parameters for the left side. To estimate total body values the measured and estimated values are combined. In all cases, scans should be monitored for movement and repeated if necessary.

Ideal positioning may not always be possible for children with leg or arm contractures. In these circumstances, research has shown that scanning the child in the semi-lateral position with limbs supported had minimal detrimental effect on whole body scan accuracy and precision [25]. Even children with significant contractures can usually be placed comfortably in the semi-lateral position (see Chap. 9). Figure 5.5 demonstrates suboptimal whole body positioning of four children with limb contractures: (a) a child aged 11 years with myotubular myopathy, (b) a child aged 13 years with grand mal epilepsy and learning difficulties, (c) a child aged 18 years with Duchenne muscular dystrophy, and (d) a hemi-scan of a child aged 16 years with Duchenne muscular dystrophy.

The proximal total hip and femoral neck are frequently measured sites in adults. Scanning the proximal hip in children, however, is more difficult because the skeletal landmarks may not be well developed and the femoral neck may be too small for the standard software. These factors contribute to poorer precision in this region. However, there are now more pediatric reference data for this site. The femoral neck region is not recommended in young children since its chan ging shape makes longitudinal studies difficult and unreliable.

Regardless, if a hip scan is warranted, note that the femoral neck box generated by standard DXA software for this region of interest may be too large for the anatomy of smaller subjects. The operator can customize the width and placement of the neck box for a better fit, but this introduces operator-related variability that can complicate subsequent studies and result in data that is no longer comparable to established reference data sets. The advantage of scanning the proximal hip is that it is a predominately cortical site; therefore, it allows the evaluation of an alternative bone element.

In positioning a child for a proximal hip scan, the following steps should be taken:

The acquired scan should include a portion of the femoral shaft, the femoral neck, the whole of the acetabulum, and part of the pelvis. Figure 5.6 illustrates three correctly acquired hip scans. Figure 5.6(a) shows the immature hip of a 6-year-old with osteogenesis imperfecta, Fig. 5.6 (b) shows the hip of a 10-year-old with anorexia nervosa and Fig. 5.6 (c) illustrates the hip of a 17-year-old long distance runner with stress fractures.

Even a developed femur may be problematic to scan and analyze, as illustrated by Fig. 5.7. Figure 5.7(a) shows the shortened femoral neck of a 16-year-old with Charcot- Marie-Tooth disease. The child in Fig. 5.7 (b) is a wheelchair-bound 10-year-old with osteogenesis imperfecta . The unusual load on her femur and femoral neck has resulted in an increased angle between the femoral neck and shaft and, hence, an unusual femoral neck morphometry and an almost absent greater trochanter.

The greatest challenge in the use and interpretation of hip scans in children is in the analysis procedure. Especially in younger and smaller children, the software can fail to properly identify the midline and the border of the greater trochanter. This may require manual placement by the technologist which does introduce error, but if the software does not put regions in the correct place it is not acceptable to just leave them. Comparison to any normative database requires that the regions are correctly placed. Longitudinal comparisons are particularly challenging due to the changes in bone size as children grow. Guidelines for longitudinal analysis of scans are provided by McKay et al [26] and are discussed in detail in Chap. 6.

The importance of identifying vertebral fractures has been highlighted with the most recent definition of osteoporosis; suggesting that osteoporosis can be defined by the identification of one or more vertebral compression fractures, in the absence of local disease or high-energy trauma [27]. Until recently the gold standard technique for identifying these fractures was lateral x-rays. However, VFA by DXA using newer generation bone densitometers has been shown to have excellent sensitivity for vertebral fracture identification, especially for clinically relevant moderate (Grade 2) to severe (Grade 3) vertebral fractures [28, 29].

Although image resolution from a DXA scanner is lower than that of conventional spinal radiographs, VFA has several advantages. DXA systems are capable of acquiring the whole spine in a single projection in both the posterior-anterior and lateral projections; whereas with conventional radiographs the thoracic and lumbar spine requires two separate exposures/films. VFA with parallel beam geometry results in images without image magnification and artifactual concave vertebral endplates (“bean” can affect) due to the parallax effect of the divergent X-ray beam of conventional spinal radiographs. Most significantly for children, the radiation dose from a DXA VFA scan is approximately 10–100 times lower than the radiation dose from lumbar and thoracic spine radiography [28]. Moreover, the vertebral fracture assessment by DXA can be performed at the same time as the routine DXA assessment and negate the need for additional visits to the hospital for further spine imaging. However, the current use of VFA is still machine-dependent. For older DXA scanners such as the GE Lunar Prodigy and the Hologic QDR series, the image resolution is not sufficient to reliably diagnose vertebral fractures in children and as such VFA scan ning on such devices is not recommended [30].