5.3 Fan-Beam Geometry: Source–Object–Detector Distances, Magnification, and Centering

5.3 Fan-Beam Geometry: Source–Object–Detector Distances, Magnification, and Centering

Quick Answer: A fan beam diverges. That single fact drives everything in this section: bone positioned higher above the table sits closer to the source, is projected larger, and — because BMD = BMC ÷ area — reports a lower BMD. Anatomy placed off the scan centerline sits in the diverging edge of the fan and is geometrically distorted. Reproducing table height, patient centering, and positioning on follow-up studies is therefore not neatness; it is precision control.

The Geometry

A fan-beam system uses a slit collimator to produce a wide, thin, diverging beam paired with a linear multi-element detector array. Three distances define the projection:

• SOD — source-to-object distance: from the focal spot to the bone being measured
• SID — source-to-image (detector) distance: from the focal spot to the detector array, a fixed mechanical constant
• OID — object-to-image distance: from the bone to the detector

The magnification factor is:

$M = \frac{\text{SID}}{\text{SOD}}$

SID is fixed by the gantry. SOD is not fixed — it varies with table height and with how far the measured bone lies above the tabletop. In most modern central DXA systems the source is beneath the table and the detector arm passes above the patient, so a bone farther from the tabletop is farther from the source. The exact sign of the effect depends on the manufacturer's source-detector arrangement, but the principle is invariant: as the object moves relative to the source, projected size changes.

Why Magnification Changes BMD

DXA reports areal bone mineral density:

$\text{BMD} \left[\frac{\text{g}}{\text{cm}^2}\right] = \frac{\text{BMC [g]}}{\text{Projected Area [cm}^2]}$

BMC is derived from attenuation and is relatively robust to geometry. Projected area is a geometric measurement and scales with the square of the magnification factor. If magnification increases area by 2% while BMC is unchanged, calculated BMD falls by roughly 2%.

Compare that against a typical facility least significant change of roughly 3% and the practical consequence is immediate: a geometry error of a couple of percent consumes most of the change you are trying to detect. This is why fan-beam systems are more sensitive to positioning height than pencil-beam systems, and why manufacturers specify table-height and patient-thickness handling in their acquisition instructions.

A Worked Illustration

Suppose the fixed SID is 90 cm and a vertebral body sits 5 cm above the source plane, so SOD is 5 cm less than a reference position. A small change in SOD produces a proportional change in $M$, and because area scales as $M^2$, the area error is approximately twice the linear magnification error. A 1% linear magnification difference therefore produces roughly a 2% area difference and roughly a 2% BMD difference in the opposite direction — comfortably within the range that could be mistaken for a therapeutic response.

Estimated BMC and Area

The ARRT outline names "estimated BMC and area" under fan-beam geometry specifically because fan-beam systems must estimate these quantities with a magnification correction rather than measure them directly. Manufacturers apply corrections based on assumed or measured patient thickness and table position. Those corrections are validated for patients positioned according to the manufacturer's instructions. Position the patient outside those assumptions — unusually thick padding, a patient supported on blankets, a table height set arbitrarily — and you have moved outside the correction's validity.

Practical rules that follow:

• Do not add unapproved padding under the region being measured. Radiolucent positioning aids supplied or approved by the manufacturer have known thickness; a folded blanket does not.
• Use the same table height and the same positioning aids on follow-up studies, and record what was used.
• Enter accurate patient weight and height, because some magnification corrections use them.

Object Centering

The fan diverges from a centerline. Anatomy placed on that centerline is projected with minimal distortion; anatomy displaced laterally is imaged by increasingly oblique rays.

Consequences of off-center placement:

• Geometric distortion. Structures at the fan edge are projected obliquely, so a vertebral body may appear subtly sheared and a femoral neck foreshortened or elongated in a way that is not a rotation error.
• Asymmetric edge blurring. Oblique projection widens the penumbra on one side of a bone edge, which degrades the reliability of automatic edge detection.
• Non-reproducibility. Because the magnitude of the effect depends on how far off center the patient was, a follow-up that centers differently introduces a precision error that no analysis step can recover.
• Field truncation. In the worst case, part of the region of interest falls outside the scan field entirely and the acquisition must be repeated.

Centering discipline is therefore concrete: use the laser or light localizer, align the patient's midline to the table centerline for spine acquisitions, and place the femur so the shaft parallels the table long axis with the ROI within the central region of the field. For the lumbar spine, the spine should appear straight and centered in the scan field with approximately equal soft tissue on each side — the same rule that makes the analysis symmetric also keeps the anatomy out of the fan edge.

Why Pencil-Beam and Fan-Beam Values Differ

Pencil-beam systems use a single pinhole-collimated beam and a single detector, raster-scanning in a serpentine path. Because the beam is narrow and effectively perpendicular at every point, magnification and distortion are minimal and largely independent of object height.

That difference in geometry, combined with different detector designs, edge-detection algorithms, and calibration standards, means BMD values are not interchangeable between pencil-beam and fan-beam systems, or between manufacturers. A patient's baseline acquired on one system cannot be compared numerically to a follow-up on another without formal cross-calibration. It is one of the most consequential practical facts in serial densitometry, and it originates in exactly the geometry described here.