## Half-field Geometry In Mammography

"reference axis" (Figure 3.5). Therefore, in order to compare measured and "nominal" focal-spot sizes, the measurement made at the chest wall needs to be "corrected" to estimate its size at the reference axis.

The reference axis usually bisects the angle formed by the x-ray tube target and the ray perpendicular to the image receptor (at the chest-wall edge of the image receptor). Thus, for an x-ray tube with an effective target angle of 22 degrees (target angle of 16 degrees plus a tube tilt of six degrees), the reference axis will be 11 degrees away from the perpendicular ray at the chest wall. Thus, the nominal focal-spot size is defined, not at the chest wall (where it should be measured), but at a distance out in the x-ray field away from the chest wall. The effective target angle and the SID will determine the reference location in the x-ray field at which the manufacturer will specify the nominal focal-spot size. The focal-spot size at the reference location can be calculated as follows:

• eff-ref-axis ./eff-chest ] 1
• tan ( 0 - 0 )' tan 0

where & is the effective target angle and €> is the angle between the tube target and the reference axis. Thus, in the case of the x-ray tube mentioned above (with the 22 degree target and the 11 degree

Fig. 3.5. Geometry of the mammography x-ray tube (Barnes, 1999).

Coverages-

Fig. 3.5. Geometry of the mammography x-ray tube (Barnes, 1999).

reference axis), feff_ref_axis = feff-chest(0.519). Therefore, if the unit's SID were 65 cm, the required effective focal-spot size (at the chest wall) would be 1.04 mm and the required effective focal-spot size at the reference location would be 0.54 mm. In order to insure that the effective focal-spot dimensions (length and width) do not exceed the recommended values, a nominal focal-spot size of 0.25 mm would be required.

If the manufacturer specifies the focal-spot size not only at a reference axis as just described, but also in the plane perpendicular to the reference axis (Figure 3.6), then a small additional correction will be required. This correction:

is generally quite small, on the order of two percent or less (Barnes, 1991).

Given all these complexities, it is not surprising that a simpler approach has been proposed (Yaffe et al., 1995) in which the limiting resolution, rather than the focal-spot size would be specified.

Fig. 3.6. Geometry associated with reference axis slit length correction between image plane and a plane perpendicular to the reference axis (Barnes, 1994).

The advantage of this approach is that a measure of the imaging performance of the unit (limiting resolution) would be specified, rather than the physical characteristics of a system component (the effective or nominal focal-spot size). ACR recommends measuring the limiting resolution by imaging a high-contrast radiographic bar pattern. The pattern should be placed parallel to the plane of the image receptor, 4.5 cm above the patient support, at a position in the x-ray field closest to the chest-wall edge of the image receptor, centered transversely. To duplicate the requirements specified above, it would be necessary to resolve 12.5 cycles mm1 under these conditions. The most recent recommendations (ACR, 1993; AHCPR, 1994) are that 11 cycles mm1 should be resolved under these conditions for a pattern oriented with the bars perpendicular to the anode-cathode axis and that 13 cycles mm1 should be resolved with the bars oriented parallel to the anode-cathode axis. To be considered resolved, the images of the bars should just appear separated (i.e., be seen as barely separated over 50 percent of their length when viewed at 5 to 10 times magnification). The measurement should be made from a film exposed to an optical density providing maximum resolution (ACR, 1993).

The heat loading capabilities of the focal spot are also an important consideration (Villafana, 1990). The smaller the heat loading capability, the smaller the allowable tube current and the longer the exposure time, the greater the potential for motion unsharp-ness. The heat loading capabilities of the anode can be increased in a number of ways. Increasing the diameter of the rotating anode will lengthen the focal track and spread the heat over a larger area without increasing the focal-spot size. A rotating anode can also be constructed with special heat dissipation features such as a heat absorbing carbon backing. Smaller target angles also spread the heat out over a larger area. As mentioned below, the small voltage ripple provided by high-frequency generators also improves the heat loading capabilities of the x-ray tube. This is the result of more uniform tube current which results in more uniform heating of the focal track.

Many mammographic units have two focal spots, one used for contact mammography and the other used for magnification. Magnifying problematic areas provides considerable improvement in overall image detail (Section 3.1.10). When measured at the chest wall, the large focal spot should meet the criteria discussed above. The requirements applicable to the small focal spot required to permit effective magnification mammography are more stringent. In fact, the limiting resolution of the small focal spot when measured using typical magnification conditions should be no less than that measured for the large focal spot, using typical contact mammogra-phy conditions.

Obviously, in principle, the smaller the focal spot the better. However, a tube with a smaller focal-spot size is likely to have a lower output and may have a shorter life expectancy. This reflects only one of the many complex trade-offs that must be resolved in designing or selecting a mammographic system. See Table 3.6 for desirable characteristics of the x-ray source assembly.

### 3.1.4 X-Ray Generator

In order to produce an appropriate beam of adequate intensity, the x-ray generator of a dedicated mammographic unit should meet certain criteria. Given the modest power requirements of screen-film mammography, the x-ray generator power rating only needs to be in the 3 to 10 kW range (AAPM, 1990). However, the usable power will generally be limited by the load capacity (instantaneous heat capacity) of the x-ray tube focal spot.

It is important that the voltage waveform have a reasonably small ripple and, therefore, high-frequency generators are recommended (ACR, 1993; AHCPR, 1994). Not only do they have a small

### TABLE 3.6—Desirable characteristics of the x-ray source assembly.

• Molybdenum target3
• Thin beryllium window

### The filter in use should be displayed on the unit

• X-ray field coverage for 24 x 30 cm cassette
• Focal spot

Resolve 11b and 13c cycles mm-1 at chest wall (object 4.5 cm above support) located directly above chest-wall edge of image receptor aAlternative target and filter combinations may be employed if they provide equivalent contrast-detail perceptibility at equal or reduced patient dose.

bResolution pattern oriented with bars perpendicular to the anode-cathode axis.

cResolution pattern oriented with bars parallel to the anode-cathode axis.

ripple, but they also provide excellent exposure reproducibility (Villafana, 1990; Yaffe, 1991) and higher output exposure rates. A small voltage ripple results in a higher effective operating potential and since the efficiency of x-ray production varies approximately as the second power of the operating potential, a small ripple results in significantly higher output exposure rates. In fact, the output of a high-frequency unit is about a factor of two higher than an unsmoothed single-phase unit (AAPM, 1990; Yaffe, 1991). This results in shorter exposure times or lower input power which has the effect of extending filament life. Alternatively, high-frequency units allow longer SIDs to be employed providing better geometric unsharpness without adversely affecting exposure time. The design and operation of high-frequency generators for mammogra-phy has been described in the literature (Gauntt, 1991).

When the voltage waveform has a small ripple, the tube current will be approximately constant and this results in more temporally uniform heating of the anode and, consequently, a greater single exposure-load capacity. These are important considerations given the small focal spots employed in dedicated mammographic units. The x-ray beam quality of high-frequency units is also more uniform over the exposure time, but somewhat higher than would be the case with a larger ripple and, this in turn, will lead to somewhat lower patient doses for high-frequency systems given the same source assembly and operating potential. High-frequency generators should have an operating potential ripple of less than five percent and an exposure ripple of 10 percent or less (ACR, 1993). However, three-phase units in which the exposure ripple is 20 percent or less, are acceptable. In addition, the exposure waveform should have both a rise time (time until the operating potential is accurate and regulated) and a fall time (time to terminate the exposure) of <16 ms (ACR, 1993).

The generator should provide a means for adjusting the operating potential from 24 kVp to at least 32 kVp in 1 kV steps, as well as a means for compensating appropriately for fluctuations in line voltage (ACR, 1993; Yaffe, 1991). An operating potential somewhat lower (i.e., down to 22 kVp) may be useful, particularly, for specimen radiography, while somewhat higher values (i.e., up to 35 kVp) may be needed for magnification. The operating potential requirements may vary somewhat for target/filter combinations other than Mo/Mo. The operating potential should be displayed and the displayed value should be within ±1 kV of the actual kilovolt peak (kVp) applied to the x-ray tube.

Using the lowest possible operating potential produces an image with the greatest subject contrast which aids in the detection of small calcifications and masses. However, while some units provide an operating potential as low as 22 kVp, radiologists rarely use below 25 kVp for routine grid mammography because at a lower operating potential, the dose increase to the patient is significant while the improvement in image quality is quite small. An operating potential lower than 25 kVp should be reserved for specimen radiography, for coned-down views of questionable areas compressible to a thickness of 2 cm (with or without magnification) and for mammograms of elderly patients whose breasts can be compressed to <2 cm.

Although a low operating potential is always preferable, it may not always be possible in certain situations. These situations include imaging of patients with dense breasts or breasts that are difficult to compress. In these patients, use of a low operating potential can result in underpenetration of the dense regions of the breast, as well as excessive exposure times, particularly on low output units, leading to problems of patient motion and high doses due to film reciprocity law failure. Additionally, patients unable to remain still because of their age, nervousness, or certain medical conditions such as Parkinson's disease, may be impossible to image at the lowest operating potential due to patient motion. Use of a higher operating potential may be necessary to achieve appropriately short exposure times with these patients. See Table 3.7 for desirable characteristics of the x-ray generator.

### TABLE 3.7—X-ray generator.

• High-frequency generator of 3 to 10 kW
• Operating potential ripple <5%
• Exposure ripple <10%
• Exposure waveform rise time and fall time <16 ms
• Operating potential selection
• 24 to 32 kVp in 1 kV steps
• Displayed value within ±1 kV of actual kVp

3.1.5 X-Ray Beam Geometry

As mentioned above, unlike general radiographic units in which the central ray from the focal spot falls on the center of the image receptor, mammographic x-ray tubes and beam limitation devices are arranged in a half-field geometry [Figure 3.7 (left)]. In this arrangement, the ray which is perpendicular to the image receptor falls on the chest-wall edge of the image receptor (AAPM, 1990; ACR, 1993; Villafana, 1990; Yaffe, 1991). Therefore, the plane defined by the focal spot and the chest-wall edge of the image receptor will be tangent to the chest wall of the patient. If this were not the case, some breast tissue would be projected off the image receptor and would not be imaged [Figure 3.7 (right)].

Fig. 3.7. (Left) correct alignment, excluded from image) (Barnes, 1999).

(Right) incorrect alignment (tissue

COIlHrrtitirtg Diaphragm

Focal Spoi

Colligating Diaphragm

Fig. 3.7. (Left) correct alignment, excluded from image) (Barnes, 1999).

(Right) incorrect alignment (tissue

The SID should be appropriate for the target angle and the focal-spot size of the x-ray tube, and the combination should meet the limiting resolution and coverage criteria given in Section 3.1.4. SIDs for dedicated mammographic units should be >55 cm for contact imaging (and >60 cm for magnification) (ACR, 1993).

Shorter SIDs would require unusually small feff. This would result in longer exposure times if the milliamperes are limited due to reduced focal-spot loadability. Short SIDs also compromise localization procedures, since limited space is available between the x-ray tube head and the patient. For magnification imaging, short SIDs result in a smaller air gap for a given magnification, yielding a higher scatter-to-primary ratio (S/P) and a higher patient dose. In addition, the shorter the SID the greater the beam divergence and, therefore, the greater the difference in magnification from the top to the bottom of the breast, particularly for thicker breasts. Finally, short SIDs result in higher patient doses. This results from the greater proportionate reduction of beam intensity between the entrance surface of the breast and the image receptor simply due to the inverse square law. Assuming a 5 cm separation between the breast entrance surface and the image receptor, a 40 cm SID unit will require about an eight percent higher exposure at the breast entrance than one with a 55 cm SID for the same exposure to the image receptor (Villafana, 1990).

The x-ray unit should also provide means to restrict or collimate the x-ray beam to accommodate the range of image-receptor sizes in use, typically 18 x 24 cm and 24 x 30 cm. This may be accomplished with the use of interchangeable rectangular apertures, moving blade collimators, or both. If interchangeable apertures are provided, each should be clearly labeled to indicate the intended image-receptor size or function. In any event, the x-ray field should not extend beyond the image receptor (the film in the cassette), except at the chest wall where it may not extend beyond by more than two percent of the SID. It is preferable for the x-ray field to extend just to the edges of the image receptor on all sides so that the processed film is black outside the breast image and extraneous light (viewbox glare) will not interfere with image interpretation.5

5Equipment standards for mammography are required by MQSA (1992) and MQSRA (1998). The Act can be found at http://www.fda.gov/ cdrh/mammography, click on "The Act" under "Regulations." Implementation of equipment standards and other criteria for mammography required by MQSA can be found at http://www.fda.gov/ cdrh/mammogra-phy, click on "The Code of Federal Regulations" under "Regulations" and scroll down to Section 900.12(b) Equipment.

The area of the primary x-ray field should be indicated by an illuminator with an intensity of >160 lux at the level of the image-receptor support (ACR, 1993). If the illuminator is intended to be a "light localizer" as defined in the federal x-ray performance standard, then it must comply with the requirements of that standard and the x-ray and light fields should also align properly so that the sum of any misalignments on opposite sides is within two percent of the SID in order that proper positioning of the breast in the x-ray field can be insured.

Ideally, the x-ray beam restriction system should change automatically when the size of the cassette holder or the grid is changed. If this feature is not available, the system should be designed with interlocks to prevent exposure if the wrong size col-limation is selected for the image receptor in use (Yaffe et al., 1995). This will prevent cone cutting and failure to image part of the breast due to interference from the collimator on a large cassette when a small diaphragm is selected. It will also prevent unnecessary exposure to the operator and patient when a large diaphragm is selected for use with a small size cassette, particularly in the MLO view.

The means of collimation should be readily changeable after the technologist has positioned the patient. This is particularly important while collimating for magnification. In this regard, a diaphragm located in the front of the tube head is inconvenient, since to change the diaphragm, the technologist must move the patient's head. Units should be designed so that the technologist can exchange the diaphragm or adjust the collimator from either side of the unit, rather than from the front.

When collimators are provided, the blades should not jam and technologists should be able to move each blade independently of the others. Movement of the C-arm or vibrations in the tube head should not loosen the blade and make it move as this could lead to cone cutting. See Table 3.8 for desirable characteristics of the x-ray beam geometry.

### 3.1.6 X-Ray Beam Energy and Intensity

Beam quality is a critical parameter in mammography and the HVL should be kept low in order to maximize subject contrast. The lower limit on the HVL is set by federal standards for purposes of patient protection. In the mammographic operating potential range, the limit [100 millimeters of aluminum (mm Al)] is defined as kVp/100. However, given the nature of the compression paddles used in mammography, the HVL should be no less than

### TABLE 3.8—Desirable characteristics of the x-ray beam geometry.

• X-ray field
• Half-field geometry
• Perpendicular ray from focal spot to chest-wall edge of image receptor
• Does not extend beyond image receptor except at chest wall
• At chest wall it may extend beyond by <2% of SID

### • SID

• Appropriate for the focal-spot size (resolve 11 and 13 cycles mm-1)
• gt;55 cm for contact imaging
• gt;60 cm for magnification imaging
• Collimation
• Accommodate range of image-receptor sizes
• Automatically change with size of cassette holder or grid or have interlocks
• Readily changeable after patient is positioned
• Interchangeable apertures clearly labeled kVp/100 + 0.03 (ACR, 1999). To insure appropriate image contrast, the HVL should also be no higher than kVp/100 + C, where C = 0.12 mm Al for Mo/Mo, 0.19 mm Al for Mo/Rh, and 0.22 mm Al for Rh/Rh (ACR, 1993). These standards should be met when the compression device is in the x-ray beam and the HVL measurement is made under the compression device at the location of the normal breast entrance surface. Since excessively soft x-ray beams result in increased patient dose with no improvement in image quality, it is critical that the lower HVL limit be met. Since excessively hard x-ray beams result in a loss of contrast, the upper HVL limit should also be met.

The HVL should not decrease by >20 percent upon removal of all materials [i.e., the compression device between the x-ray filter and the breast (ACR, 1993)]. In this instance, the HVL should be measured at a point in the x-ray field 4 cm from the chest-wall edge of the image receptor and centered transversely. Most of the minimum filtration should be provided by the selective filter and not by other beam hardening materials, such as a glass window on the x-ray tube, a permanently installed glass mirror in the beam limiting device, or an inordinately attenuating compression device.

The x-ray beam output is also critical. Insufficient output can result in excessively long exposure times resulting in problems with patient motion. With some images, motion may be noticeable when exposure times exceed 1 s and may become a significant problem at times of 2 s or more (Feig, 1987). With inadequate compression, considerable motion unsharpness can be seen with times as short as 0.2 s (NCRP, 1986). With long exposure times, patient doses may also be high due to the effects of reciprocity law failure of the film. Alternatively, insufficient output may require the use of higher than optimal kilovolt peak settings and this can result in inadequate subject contrast. When the focal spot intended for contact mammography is used, dedicated mammographic x-ray units with molybdenum targets and filters should be capable of delivering 200 pC kg1 s1 for 3 s at 28 kVp at the location of the breast entrance surface under the compression device (ACR, 1993). This is a significantly higher output exposure rate than has been previously recommended (AAPM, 1990). The output should be measured under the compression device, 5 cm above the top surface of the image-receptor support and 4 cm out from the chest-wall edge of the image receptor (centered transversely). Because of differences in x-ray tubes, x-ray generator design, unit geometry, SID, etc., a specification based on a value of tube current can be misleading in predicting radiation output (AAPM, 1990). For this reason, the output should be specified in terms of pC kg1 s1.

As mentioned in a previous section, a half-field geometry is used in mammography. Consequently, the heel effect is more pronounced than in general radiography and the beam intensity can fall significantly from the chest wall to the nipple edge of the image receptor (Table 3.9). The HVL also changes, increasing with increasing distance from the chest wall. See Table 3.10 for desirable characteristics of x-ray beam energy and exposure rate.

TABLE 3.9—Heel effect for a 30 kVp exposure, and HVL versus distance from chest wall at 60 cm SID (Barnes, 1991).

Distance from

Relative