A mammogram is a radiograph of the breast. It is used mainly for screening for breast cancer. Women without symptoms are invited at regular intervals so that cancers are found early. It is also used to examine a lump or other symptom. A standard examination has four images, two views of each breast. Fat appears dark, fibroglandular tissue and tumors appear bright, and calcifications appear as small bright specks. The images are large, about 3300 × 4100 pixels.
Three mammograms. Left: craniocaudal (CC) view of a right breast on film; fat is dark and fibroglandular tissue bright, the lead marker reads CRANIO CAUDAL R, and the label at the bottom is the film number (National Cancer Institute, public domain). Middle: digital MLO view of a left breast with an irregular, spiculated mass in the upper breast, a typical appearance of invasive cancer (Bahman Rasuli, Radiopaedia.org, rID 91430, CC BY-NC-SA 3.0). Right: magnified view of a cluster of microcalcifications, the many small, bright, irregular specks that are the other main sign of cancer on a mammogram (Guilherme Pioli Resende, Radiopaedia.org, rID 82983, CC BY-NC-SA 3.0).
Bone stands out in a radiograph because calcium absorbs X-rays so strongly. A breast tumor has no bone: it is soft tissue inside soft tissue. Mammography is designed for that problem. The physics follows from 2.2 X-ray radiography, but design choices, from the tube to the detector, are aimed at seeing differences between soft tissues and tiny specks of calcium.
The breast contains two main tissues: fat, and fibroglandular tissue, which is composed of the glands and ducts together with the connective tissue that supports them. The attenuation coefficient $\mu$ of the two tissues are very close, and a tumor is closer still to fibroglandular tissue.
Left: μ against photon energy for fat (orange), fibroglandular tissue (blue), tumor (red, dashed), and calcification (green), log scale. At 20 keV, the mammography energy, fibroglandular tissue attenuates about 1.6 times as much as fat, a tumor only about 5% more than fibroglandular tissue, and calcification about 25 times as much; at 60 keV, the chest radiograph energy, the differences shrink. NIST XCOM with the compositions of Hammerstein et al. (1979).
Mammography is primarily interested in two types of findings, both of which are challenging:
Masses and calcifications are the two commonest signs but not the only ones. Cancer can also appear as an architectural distortion or an asymmetry, and the shape, margins, and distribution of a finding matter as much as its presence. In 2.2 X-ray radiography, we learned that the visibility of a structure came down to the extra attenuation it adds along a ray: the difference in $\mu$ between the structure and its surroundings, times its thickness (the $\Delta$ of Eq. 2.2.6). Both targets above are therefore easiest to see where the differences in $\mu$ are largest, and the figure shows where that is: fibroglandular tissue attenuates about 15% more than fat at 60 keV but about 55% more at 20 keV, so the same lump gives almost ten times the contrast at the lower energy, because photoelectric absorption, which is sensitive to composition, dominates there. For this reason, mammography tubes run at about 25 to 35 kV, in comparison to 50 to 120 kV in general radiography.
Fraction of photons transmitted through a breast of half fat and half fibroglandular tissue. Compressing from 8 cm to 5 cm lets seven times as many photons through at a standard mammography energy level.
Low energy improves contrast but increases dose. At 20 keV, about 4% of photons pass through 5 cm of tissue (about 40% at 60 keV); the rest are removed from the primary beam by absorption or scattering. Absorption deposits photon energy in tissue, while scattered photons may carry energy away. Each extra centimeter halves the transmitted photon count. Mammography reports the mean glandular dose, the energy absorbed per kilogram of fibroglandular tissue, about 1 to 2 mGy per view. A four-view examination has an effective dose of about 0.4 mSv, roughly twenty times that of a chest radiograph (Hendrick, 2010). The main way to get more photons to the detector for less dose is to make the breast thinner (next section).
Choices in device design such as focal spot and detector are intended for small targets. The focal spot is about 0.3 mm wide for mammograms, compared with about 1 mm in general radiography: a focal spot of width $a$ blurs every edge by about $a(M-1)$, with $M$ the magnification from 2.2 X-ray radiography. The blur must stay smaller than a microcalcification. The detector elements are 50 to 100 µm wide, about half the size used in general radiography; a 23 cm × 29 cm field at 70 µm gives the 3328 × 4096 pixel image mentioned above, stored with 12 to 14 bits.
A mammography unit. The tube head is at the top of a C-arm and the detector housing at the bottom; the focal spot sits directly above the chest wall, so one edge of the beam runs down the chest wall and tissue against the wall is in the image. The paddle presses the breast (dashed: before compression) down to 2 to 8 cm on the detector housing, with an anti-scatter grid between breast and detector.
The breast is compressed between the detector housing and a plastic paddle, typically to a thickness of 2 to 8 cm. This can be painful for patients, but has the following benefits: