FIGURE 11 Schematic of the effect of automatic exposure control (AEC) on the imaging of a non-dense breast (a) and a dense (b) breast. Because of AEC it is difficult to relate absolute brightness to tissue composition. For example, fatty regions in a dense breast will usually appear darker than the same regions in a predominantly fatty breast.

calibration object include the overall thickness, the size of the steps, and appropriate compositions of the phantom material. Phantom materials, representing the X-ray attenuation of breast fat and fibroglandular tissue are readily available commercially. These can be combined to mimic intermediate compositions [43]. A radiograph of the breast and the calibration object is shown in Fig. 13.

Using the calibration object, a relationship such as that illustrated in Fig. 14 is obtained for each clinical image. For the exposure conditions used, each brightness could be obtained with many combinations of breast thickness and composition. If the total compressed breast thickness is known for a particular point in the image, then only a single combination of effective thicknesses of adipose and fibroglandular tissue adding to that total thickness is consistent with that image brightness.

The breast thickness can be estimated by means of the breast compression gauge of the mammography unit. A more accurate value of thickness is possible with a direct radiographic measurement using the triangulation device whose image is seen in Fig. 13. With this device, a precision of approximately 1 mm is possible. The thickness variation in the margin and elsewhere in the breast can be estimated using a technique similar to that of the previous section, or estimated

FIGURE 12 Calibration object for determination of volumetric mammographic density.

FIGURE 13 Mammogram with calibration object. Radiographic gauge for breast thickness measurement is shown at upper right.

FIGURE 14 Example of an empirical model that relates brightness in the mammogram (gray level) to the thickness and composition of breast tissue. The model has been constructed from the image of the calibration object shown in Fig. 12.

with assumptions about how the thickness varies in the periphery.

0 0

Post a comment