Other Imaging Modalities

There are applications when it is not practical to use the mammogram for assessment of breast density and instead, it is preferable to use another imaging method for this purpose. One example is in young women (under age 40) who do not normally receive routine mammograms. Another may be in high-risk women where it maybe advisable to monitor changes in the breast more frequently than is considered appropriate with mammography. In both cases, an imaging method that does not employ ionizing radiation would be desirable. Two modalities can be immediately considered — ultrasound and magnetic resonance imaging. Not only do these avoid radiation concerns, but they also produce three-dimensional image data, thereby facilitating volumetric analysis.

With ultrasound, differences between the acoustic scattering properties within fibroglandular tissue and fat [44-46] give rise to differences in image texture. Work is currently underway to determine features that allow effective segmentation of dense tissue from fat. One problem is that the thin slices normally produced in ultrasound scans used for diagnostic purpose are likely to be too noisy due to speckle, making segmentation unreliable. This maybe overcome by combining slices (Fig. 15) to form thick image slabs with noise reduction due to averaging. Another issue is that the ultrasound gain controls are normally used interactively to optimize the diagnostic value of the image by compensating for tissue attenuation. This process is both patient- and operator-dependent. For quantitative work, it will be necessary to adopt a mode in which

FIGURE 15 4 MHz ultrasound breast images of a 71-year-old subject (a) single transverse slice and (b) 1 cm-thick slice consisting of the average of 10 images spaced 1 mm apart. Darker regions represent fatty tissue while bright regions represent fibroglandular tissue. Note the reduction of speckle noise and "mammographic" appearance of this image. (c) Segmented image showing fibrous (white) and fatty (gray) tissues. Images courtesy of Dr. Stuart Foster.

FIGURE 15 4 MHz ultrasound breast images of a 71-year-old subject (a) single transverse slice and (b) 1 cm-thick slice consisting of the average of 10 images spaced 1 mm apart. Darker regions represent fatty tissue while bright regions represent fibroglandular tissue. Note the reduction of speckle noise and "mammographic" appearance of this image. (c) Segmented image showing fibrous (white) and fatty (gray) tissues. Images courtesy of Dr. Stuart Foster.

machine settings are constant for the entire imaging procedure or else a method to correct for alterations in acquisition parameters.

Some work has also been done to develop magnetic resonance imaging for characterizing the composition of the breast [47]. It has been shown that a pulse sequence that allows separate "fat" and "water" images to be produced provides data that correlate very well with X-ray mammographie density analysis [48]. These methods can use saturation techniques to exclude irrelevant tissue from the measurement. Unlike the long times required for diagnostic MRI, the low-resolution tissue composition images can be acquired in as little as 18 seconds.

0 0

Post a comment