There are two standard mammographic projections: a mediolateral oblique (MLO) view and a craniocaudal (CC) view (Fig. 69-2). Correct positioning is crucial to avoid missing lesions situated at the margins of the breast. The MLO view is taken with the X-ray beam directed from superomedial to inferolateral, usually at an angle of 30–60°, with compression applied obliquely across the chest wall, perpendicular to the long axis of the pectoralis major muscle (Fig. 69-3A). The MLO projection is the only projection in which all the breast tissue can be demonstrated on a single image. A well-positioned MLO view should demonstrate the inframammary angle, the nipple in profile, and the nipple positioned at the level of the lower border of the pectoralis major, with the muscle across the posterior border of the film at an angle of 25°–30° to the vertical (Fig. 69-2A).

For the CC view, the X-ray beam travels from superior to inferior. Positioning is achieved by pulling the breast up and forward away from the chest wall, with compression applied from above (Fig. 69-3B). A well-positioned CC view should demonstrate the nipple in profile. It should demonstrate virtually all of the medial tissue and most of the lateral tissue except the axillary tail of the breast. The pectoralis major is demonstrated at the centre of a CC film in approximately 30% of individuals and the depth of breast tissue demonstrated should be within 1 cm of the distance from the nipple to the pectoralis major on the MLO projection (Fig. 69-2B).

Additional Projections

Supplementary views may be taken to solve specific diagnostic problems.^1,2 For example, the CC view can be rotated to visualise either more of the lateral or medial aspect of the breast, compared to the standard CC projection. Localised compression or ‘paddle views’ can be performed. This involves the application of more vigorous compression to a localised area using a compression paddle (Fig. 69-4). These views are used to distinguish real lesions from superimposition of normal tissues and to define the margins of a mass. A true lateral view may be used to provide a third imaging plane in order to distinguish superimposition of normal structures from real lesions or to increase the accuracy of wire localisations of non-palpable lesions. The true lateral view is performed with the mammography unit turned through 90° and a mediolateral or lateromedial X-ray beam.

Magnification views are most frequently performed to examine areas of microcalcifications within the breast, to characterise them and to establish their extent. Magnification views are typically performed in the craniocaudal and lateral projections. The magnified lateral view will demonstrate ‘teacups’ typical of benign microcalcifications, described later in the chapter. Mammographic technique may need to be modified in women with breast implants. Silicon and saline implants are radio-opaque and may obscure much of the breast tissue. Consequently, mammography is of limited diagnostic value in some women. The Eklund technique can be employed to displace the implant posteriorly, behind the compression plate, maximising the volume of breast tissue that is compressed and imaged.³ Mammography-induced implant rupture has not been reported to date.

Breast Compression

Compression of the breast is essential for good mammography, for the following reasons:

Mammography uses ionising radiation to image the breast. The risks of ionising radiation are well known and any exposure needs to be justified, with doses kept as low as possible. The radiation dose for a standard two-view examination of both breasts is approximately 4.5 mGy.⁴ The average effective dose of radiation from mammography is equivalent to 61 days of average natural background radiation.⁵

Dose is more of an issue in a population screening programme, where women who may never develop breast cancer are being exposed to radiation. It has been estimated that the risk of inducing a breast cancer in women screened in the United Kingdom National Health Service Breast Screening Programme (NHSBSP) is 1 in 100,000 per mGy. A risk–benefit calculation has established that the benefits of screening far outweigh the risk of inducing a cancer, with the ratio of lives saved to lives lost calculated as approximately 100 : 1.⁴

Digital mammography systems have the potential to reduce patient dose without loss of image quality.

The Detector

Traditionally, the mammographic image has been recorded on film, but this has been superseded by digital technology. Manufacturers have developed a number of different approaches to producing a digital mammogram. The first type of digital system developed for mammography used photostimulable phosphor computed radiography (CR). This uses an imaging plate coated with a phosphor to replace the traditional screen/film mammography cassette. The imaging plate, stored in a conventional-looking cassette, is exposed in the usual fashion in a conventional (analog) mammography machine. A latent image is stored in the phosphor after exposure. The imaging plate is scanned by a laser beam in a plate reader and light is emitted in proportion to the absorbed X-rays. The emitted light is then detected by a photomultiplier system and the resulting electrical signal is digitised to produce the image.

More recently, full-field flat-panel detectors have been developed. One type of detector consists of a phosphor layer coated onto a light-sensitive thin-film transistor (TFT) array composed of amorphous silicon. Charge from the TFT array produced in response to the light emission from the phosphor is measured and digitised. The above digital systems require multiple conversion steps in the acquisition of the image: X-ray energy is converted into light energy, which is then converted into electrical energy. Multiple conversion steps are inefficient and have the potential to degrade image quality. Systems that avoid these conversion losses are described as being more ‘direct’. Some manufacturers use amorphous selenium or silicon dioxide in the detector, allowing the energy of the X-ray photon to be directly converted into electrical energy.

Digital Mammography in Clinical Practice

There are clear logistical advantages to digital mammography, including the potential to improve patient throughput, as traditional screen/film mammography is labour intensive, with time taken in handling cassettes, loading/unloading film and processing. Digital mammography equipment interfaces directly with picture archiving and communication systems (PACS), leading to further increased efficiencies associated with image storage and display, with soft-copy reporting from high-resolution (5 megapixel) monitors.

It is important to establish whether an improvement in image quality can translate into an improvement in cancer detection. A powerful test of the potential of digital mammography to improve cancer detection rates is in a screening setting. Several early studies found that digital mammography was at least equivalent to screen/film mammography in terms of cancer detection rates.^6,7 The larger Oslo II study, which randomised over 25,000 women to either conventional or digital mammography, showed an increase in cancer detection in the women undergoing a digital mammogram, but this did not quite reach statistical significance (p = 0.053).⁸

To detect significant differences between the two techniques, the population size needs to be large, as the cancer detection rate in a screening population is around 6 per 1000 women screened. The North American Digital Mammographic Imaging Screening Trial (DMIST) enrolled 49,500 women.⁹ This study found that overall the diagnostic accuracy of digital and conventional mammography was similar. However, there were some groups of women where digital mammography outperformed screen/film mammography, showing significantly improved diagnostic accuracy. These were women under the age of 50, those with dense breast parenchyma and women who were pre- or perimenopausal. Encouragingly, it is in these groups of women that conventional screen/film mammography had shown reduced sensitivity for detecting breast cancer.

Computer-Aided Detection

Computer-aided detection (CAD) is a software system that is designed to assist the film reader by placing prompts over areas of concern, to reduce observational oversights. CAD systems are highly sensitive for detecting cancers on screening mammograms. CAD will correctly prompt around 90% of all cancers, with 86–88% of all masses and 98% of microcalcifications correctly marked. Specificity is much more of a problem with a high rate of false-positive prompts. The number of false prompts will vary according to the level of sensitivity at which the system is set; typically, there are between two and four false prompts per standard set of mammogram images.^10–12

The routine use of CAD remains controversial and there is no consensus in the literature as to whether CAD improves film reader performance; despite this, CAD is used in the interpretation of screening mammograms in around 75% of cases in the United States.¹³ Some prospective studies of the use of CAD in the screening setting have shown a significant improvement in a single film reader’s performance when CAD software is applied, whereas others have shown no effect on cancer detection rates.^12,14 A large multicentred retrospective review of the use of CAD in the interpretation of screening mammography in the United States found its use associated with a decrease in the specificity of screening, with an increase in recall rates for only a very minimal improvement in sensitivity, largely the result of a non-significant increase in the detection of ductal carcinoma in situ.¹³

It is difficult to extrapolate the findings of these studies to the UK breast screening programme (NHSBSP), where virtually all films are double read by two readers. Double reading is known to increase cancer detection rates by 4–14% compared to single reading, and so the question is whether one reader using CAD could produce results equivalent to double reading and whether CAD is a more cost-effective solution to recruitment problems than training non-medically qualified film readers. The Computer-Aided Detection Evaluation Trial (CADET II) was a large prospective trial of over 30,000 women that reported a comparable cancer detection rate for single reading with CAD to double reading but with a small but significant increase in recall rate.¹⁵ A further sub-analysis has suggested that at the present time CAD is not a cost-effective alternative to double reading in the NHSBSP in view of the increase in recall rates.¹⁶

Digital Breast Tomosynthesis

One of the limitations of mammography is that it produces a two-dimensional (2D) radiographic view of a three-dimensional structure and, as a consequence, a cancer may not be detected due to overlapping normal glandular tissue obscuring the presence of a tumour. Lesions may be simulated by the superimposition of normal tissue, leading to unnecessary recalls following screening mammography. These factors result in a reduction in the sensitivity and specificity of mammography. Breast tomosynthesis is an emerging digital mammographic technique where thin slices through the breast are reconstructed from multiple low-dose projections acquired at different angles of the X-ray tube. The resulting thin sections can be scrolled through by the reporting radiologist, with the potential to alleviate the effects of tissue superimposition.

The role of tomosynthesis is still being defined, and there is debate surrounding its use as an adjunct or replacement for conventional 2D digital mammography. Its role in diagnosis and screening requires clarification. It has the potential to be an additional tool, for the work-up of screen-detected abnormalities replacing traditional ‘paddle’ views.¹⁷ When used as a screening tool, studies show an improvement in specificity, with a reduction in recall rates of up to 11%; improvements in sensitivity over conventional 2D mammography are not so clear.¹⁸

Ultrasound

The main indications for ultrasound are:

Elastography is an ultrasound technique that can provide additional information based on tissue stiffness or hardness. The concept that malignant lesions feel firmer or stiffer than the surrounding breast tissue is well recognised from clinical palpation. There are two methods of producing an elastography image or elastogram: strain elastography, where the operator gently manually compresses the breast tissue, and shear wave elastography, where pulses are generated by the transducer producing transverse shear wave propagation through the breast tissue. The main advantage of shear wave elastography is that the technique is quantitative and highly reproducible. Information regarding stiffness can be displayed as a black and white or colour overlay onto the grey-scale image (Fig. 69-5). Features on the elastogram that can be measured include quantitative elasticity (stiffness) in kPa and size ratios relative to conventional grey-scale imaging. In general, breast cancers tend to be stiff, with benign lesions or normal tissue appearing softer (elasticity <80 kPa). Invasive breast cancers often produce areas of stiffness that are larger than the grey-scale abnormality, likely due to changes in the tumour-associated stroma. Elastography has the potential to improve the specificity of breast ultrasound for differentiating benign from malignant masses, reducing the number of benign biopsies. In a recent study, the use of shear wave elastography resulted in a significant improvement in the specificity of breast mass assessment from 61.1 to 78.5%.¹⁹

On mammography, fibroadenomas are seen as well-defined, rounded or oval masses (Fig. 69-8A). Coarse calcifications may develop within fibroadenomas, particularly in older women (Fig. 69-9).

Ultrasound features have been described that are characteristic of benign masses.²⁰ These include hyper­echogenicity compared with fat, an oval or well-circumscribed lobulated or gently curving shape and the presence of a thin echogenic pseudocapsule. If these features are present with no features suggestive of malignancy, then a mass can be confidently classified as benign.

These features are demonstrated by fibroadenomas (Fig. 69-8B). Most fibroadenomas are isoechoic or mildly hypoechoic relative to fat, with an oval shape and lobulated contour. A thin echogenic pseudocapsule may be seen. Percutaneous biopsy may be avoided in women under the age of 25, where the risks of any mass being malignant are very small;²⁰ however, in most cases, even though the mass appears benign, percutaneous biopsy is undertaken to confirm the diagnosis.

Fibroadenomas must be distinguished from well-circumscribed carcinomas; this is done by percutaneous biopsy. Phyllodes tumour can also have a similar appearance to fibroadenoma, leading to diagnostic difficulties (Fig. 69-10). The pathological characteristics can also be similar to those of large fibroadenomas. Most phyllodes tumours are benign, but some (less than 25%) are locally aggressive and may even metastasise.²¹ When a diagnosis of phyllodes tumour is made, surgical excision must be complete with clear margins to prevent the possibility of recurrence. Many larger fibroadenomas (over 3 cm) and those that show a rapid increase in size are excised in order to avoid missing a phyllodes tumour.

On mammography, they may be seen as a well-defined mass, commonly in a retroareolar location (Fig. 69-11A). Sometimes the mass is associated with microcalcifications. On ultrasound, they typically appear as a filling defect within a dilated duct or cyst (Fig. 69-11B). On aspiration, any cyst fluid may be bloodstained. As it is impossible to differentiate papillomas from papillary carcinomas on imaging criteria, percutaneous biopsy is required.

Lipomas are benign tumours composed of fat. They present clinically as soft, lobulated masses. Large lipomas may be visible on mammography as a radiolucent mass (Fig. 69-12A). On ultrasound their characteristic appearance is that of a well-defined lesion, hyperechoic compared with the adjacent fat (Fig. 69-12B).

Hamartomas are benign breast masses composed of lobular structures, stroma and adipose tissue—the components that make up normal breast tissue. They occur at any age. On imaging they may be indistinguishable from other benign masses, such as fibroadenomas. Sometimes large hamartomas are detected on screening mammograms and are impalpable (Fig. 69-13). On mammography they classically appear as large, well-circumscribed masses containing a mixture of dense and lucent areas, reflecting the different tissue components present. Diagnostic difficulty may be encountered because percutaneous biopsy specimens may be reported as normal breast tissue.