Originally, the scanners used a highly collimated beam of X-rays in conjunction with sequential detectors or a single detector that moved in a raster pattern (i.e., in a series of thin parallel lines) across the patient. This pencil beam system produces the most geometrically correct information, with little or no magnification of the area being scanned.

The current fan beam systems use a slit collimator to generate a beam that diverges in two directions in conjunction with a linear array of solid-state detectors, so bone measurements can be made with a single sweep of the X-ray arm. The fan beam systems use higher energy photon intensities and a greater photon flux, thus producing a better-resolution image considerably faster than the older pencil beam machines. The lumbar spine can be scanned in 30 s or less with the fan beam, as compared with the 3–10 min required for the pencil beam system.

The trade off for improved image resolution with the fan beam is a higher radiation exposure. Additionally, the geometry associated with this technique leads to magnification of the image in one direction [8, 9]. The degree of magnification will depend on the distance of the bone or tissue away from the source: the closer the body part is to the source, the greater the magnification.

Fan beam systems can create either wide or narrow x-ray beam profiles depending on the make and model. The narrowest fan beam systems scan in a rectilinear raster fashion to cover the scan area, much like the original pencil beam machines. However, since the beam is wider than the original pencil beam machine, it can cover the body in a much faster time, typically 30 s. Wide fan beam systems can cover the scan areas of hip, spine, and forearm systems in one pass. Cross-calibration studies demonstrated no detectable magnification effect between the old-generation pencil beam scanner and the new narrow fan beam machine [10] (Fig. 3.2).

The amount of radiation exposure in DXA is extremely low compared to many other X-ray imaging techniques. It has been difficult to estimate the degree of risk of harms associated with these very low levels of radiation except by extrapolation from studies that involved distinctly higher levels of radiation exposure. Presently, studies have not been able to establish a link between health risk and the low levels of radiation exposure that are typical of DXA. According to the Health Physics Society, the risks of health effects for exposures less than 5–10 REM “are either too small to be observed or are nonexistent [11].”

Health effects of radiation have been demonstrated at doses above 5–10 REM (greater than 50,000–100,000 μSv) [11]. The principal risk due to radiation is random X-ray interactions with the body, which can result in carcinogenic or genetic effects. Typically, carcinogenic effects will not manifest in an individual for several decades following an exposure [12]. This is an important consideration when scanning children since they have a longer amount of time for expression of an effect than adults [12]. Because the majority of the children scanned will still be fertile, the potential genetic effects of radiation exposure are a theoretical consideration [13]. However, as shown in Table 3.1, radiation exposures from DXA are approximately 10,000 times less than the radiation doses at which health effects occur.

Estimates of risk from radiation exposure are expressed in terms of effective dose, in units of sieverts or REMs, where 1 milliREM (mREM) equals 10 microsieverts (10 μSv). The effective dose is calculated from the magnitude of exposure, the type of radiation causing the exposure, the organs exposed, and their relative radiosensitivities. The resulting value can be compared to other scanning techniques (Table 3.1), to naturally occurring background radiation (8 μSv/day), or to a round trip transatlantic flight (80 μSv).

The more commonly cited unit of radiation exposure is the entrance surface dose, or ESD, in units of gray (Gy); 10 μGy = 1 mrad (i.e., 1 Gy = 100 rad). ESD is a measure of the radiation on the surface of the patient, before it passes through and is absorbed by the body. It is an easier measure to obtain as it requires only a simple measure of the X-ray output detected at the skin surface. It will be approximately the same for any patient scanned at any one exposure level, irrespective of the region scanned. The ESD will be higher than the effective dose since only a fraction of the X-rays are stopped by the patient. Although ESD gives the operator an indication of the exposure levels, it does not take into account the organs being exposed and the relative radiosensitivities of the irradiated organs.

Table 3.1 lists both the effective and entrance surface doses of ionizing radiation doses associated with the more commonly used densitometers. The doses in the table refer to estimates for either adults, children aged 15, or children aged 5 years old when available.

Movement of the patient during the DXA scan acquisition is a common problem encountered by DXA technologist. If a scan contains movement or some other removable quality issues, it is common practice to attempt the scan again. If an error-free scan is not acquired in three attempts, one should abort that exam to limit the dose exposure to the child.

In summary, the radiation exposure associated with DXA is acceptable for pediatric use. However, efforts should always be made to minimize lifetime radiation exposure through the judicious selection of patients and skeletal sites for DXA scanning and through optimal densitometry technique.

Machine precision is calculated from repeated scanning of a single phantom, without moving the phantom between scans. The usual protocol for the measurement of machine precision requires scanning a phantom ten times on the same day. For newer DXA models, the %CV for this procedure is typically less than 1 %.

Long-term precision is measured by repeatedly scanning a phantom daily or weekly over months to years to monitor any temporal changes in the machine. These measurements can be used to assess the long-term stability of a scanner; since the measurements from a phantom should theoretically be the same each day, any drift or change would therefore be due to the machine.

In vivo short-term precision is calculated by repeated scanning of subjects a minimum of two times on the same day or within a short time interval. To achieve statistical power, BMD testing must be done three times in each of 15 individuals or twice in each of 30 subjects. The standard deviation for each patient is calculated, and then the root mean square standard deviation for the group is calculated. A good explanation for these calculations can be found on the website of the International Society for Clinical Densitometry (http://​www.​iscd.​org). The ISCD also provides an online calculator for quantifying your precision values. Because this procedure requires two scans and twice the radiation exposure, in vivo precision testing is considered by some to be clinical research. Regardless of interpretation, all participants should provide written informed consent.

The precision estimates reflect both machine precision and operator precision. For this reason, in vivo precision is worse (greater %CV) than the machine precision alone, but in vivo precision is representative of the real scanning situation. The best precision will be achieved if the patients are scanned and analyzed by an operator that has been trained in the positioning and analysis of patients. An excellent metric for training is certification from a program specializing in DXA operation.

Precision studies should be performed in the population to be scanned most often since precision can vary as a function of body size [14]. However, precision measured in mature individuals may differ from that measured in children due to the latter’s smaller size and variable ability to cooperate. The ability of the software to detect the edges of smaller bones may also affect precision in children. Ideally, pediatric data should be gathered when possible. One multicenter study of DXA precision in 155 children, ages 6–15 years, demonstrated coefficient of variation values of 0.64–1.03 for spine and 0.66–1.20 for whole body BMD, depending on the age range [15].

This measure is obtained by repeat scanning of a group of patients over a period of time. It is harder to evaluate since, unlike a phantom, which maintains stable bone density over time, the patient’s bone density may increase or decrease. For children, this is particularly difficult to estimate due to the expected changes in bone measures in growing children.

The least significant change (LSC) is the smallest percent difference that can be detected by the technique from repeat measurement of a patient. The value usually expressed for LSC is 2.8 × precision where the precision can be in the form of the %CV or SD. However, there are a few caveats. First, both measurements should be acquired on the same DXA system. Second, The DXA system should be known to be stable in its calibration between the two measurements. Third, the LSC defines how much change has to occur to have 95 % confidence that any real change in the bone density has occurred at all. The correct way to report change in bone density would be to report the measured difference ± the %LSC. For example, with a measured decrease of 4 % between two DXA scans and a precision of 1 %, the LSC would be 2.8 %. The report should state there was a decrease 4 ± 2.8 % in bone density. If the decrease had only been 2 %, then it would be appropriate to report that no significant change in bone density has occurred.

Although availability of DXA may vary from country to country, this technique is now widely available in both general hospitals and academic medical centers. In some areas, mobile units are also available, reducing the need for the patient to travel long distances to the nearest machine.

Although any radiation exposure results in a degree of risk to the patient, DXA has one of the lowest effective doses of all the ionizing radiation imaging techniques, being equivalent to approximately less than one day’s naturally occurring radiation in most cases. The low dose of DXA is a strength, since there are no alternatives to measure bone density with lower dose. Quantitative CT is used in children but has at least six times or higher dose.

Much work has been done by the manufacturers of DXA machines to produce a stable X-ray source and an efficient detector system, thereby making DXA a precise technique for measuring bone. The average coefficient of variation for a spine DXA scan is 1.5 % or less, compared to as much as 5 % for an average calcaneus ultrasound scan [20]. Additionally, sophisticated analysis software packages are used, which, for a large proportion of DXA scans, require little or no operator intervention, thus further improving precision.

Current generation DXA fan-beam systems and hardware have drastically shortened scan times and offer a definite advantage to older pencil beam systems. Whole body DXA scans can be completed in from 3 to 10 min, and spine scans, in as short a time as 30 s, which minimizes the possibility of movement artifacts in young children.

As a result of the wide availability and relatively low radiation dose, DXA data have been collected on samples of healthy infants, children, and adolescents in several countries. Table 3.2 is a survey of selected reference data for a variety of makes and countries [21]. With respect to data available by the manufacturers on their systems, reference data for spine and total body BMD is available for GE and Hologic down to 5 years for GE and 3 years for Hologic. These data have been used clinically as reference values to identify children with “normal” versus “abnormal” bone density.