Current pulse-echo ultrasound scanners provide high quality images of internal organs, such as the liver, kidneys, prostate and breast. They operate by transmitting pulses of sound into the tissue. The pulse can be focused and steered along a fixed direction. Reflectors in the beam give rise to echoes, which are displayed as dots on a screen. Imaging is done by sweeping the pulsed beam through the region to be scanned – like a search light scanning the night sky, and detecting and displaying echoes from objects in the sound path.
Figure 1. Typical human liver image. Click for larger image.
Standard ultrasound images though of good quality, are not quantitative because equipment technical settings on the scanner can be varied freely at the discretion of the sonographer, there is no absolute calibration of the echo data, and each patient differs somewhat in how rapidly the tissues absorb the ultrasound energy. Moreover, the signal and image processing used by conventional scanners “throws away” some of the echo information, probably the most important being the actual frequency content of the echo data.
We are developing methods for quantifying ultrasound attenuation and ultrasound backscatter levels from data acquired by ultrasound scanners. Specific projects include:
Measuring “Absolute” Backscatter and Attenuation
Methods for doing this apply precise calibrations; process each frequency in the pulse individually; and utilize models for the tissues and ultrasound beams. When done properly, we provide data to the clinician on the absolute scattering properties and the sound beam absorbing properties of the tissue. There appears to be diagnostically useful information in these properties.
We developed our methods using laboratory apparatus [1]. The techniques were ported to clinical scanners, where they were applied to the detection of “diffuse liver disease” (fatty liver; cirrhosis; livers that are changed by drugs) [2,3] We currently are updating our equipment to acquire additional backscatter vs. frequency information on more human livers. We are also working with radiologists and surgeons to apply our methods to intraoperative ultrasound, where sensitive scanning methods are needed to guide surgical resection of cancer.
Determining the number of “subresolution” scattering targets
It is easy to compute the “first moment” of a quantity, such as the echo signal amplitude from a region of interest. Just take the average value. Interestingly, we find that when we compute “higher moments” of the echo signals we can obtain additional types of information. For example, even though the ultrasound scanner is not capable of “resolving” individual scatterers in the tissue, it is possible to estimate the number of “scattering targets” from specific relationships of these higher moments. This type of analysis has been applied in the fishing industry for quite some time. We are exploring this concept for ultrasound tissue characterization [4].
“Sizing-up” scattering targets
We are testing methods for estimating scatterer sizes using information mainly on how ultrasound echo signals vary with the frequency. The relevant measure for the scatterer size is the “spatial autocorrelation function,” a statistical measure of spatial variations in acoustical properties of a medium. We can differentiate “tissue mimicking” phantoms that have different sized scatterers even though the individual scatterers cannot be resolved by the scanner [5]. We are working to try out these methods in tissues. We hypothesize scatterer sizes will be different in the nodular tissues of cirrhotic liver than in normal liver, and experiments are being designed to test this hypothesis.
VSA (Video Signal Analysis)
Access to “raw” data from a scanner, and hardware and software to process that data, can be done in research; but thus far it has not been practical to make such data generally available. Another method for producing quantitative data from ultrasound imagers, being developed at Wisconsin, uses the image data itself. It is attractive because many modern scanners are now being linked to work stations for archiving and reviewing images, so this type of quantitation will be readily available clinically.
VSA involves careful calibration of the ultrasound image data so that changes in image brightness can be related to echo amplitude variations at the transducer. It also involves a reference phantom to eliminate the effects of the equipment and sonographer setting changes. Preliminary tests are very promising [6], and ongoing research will tell us exactly what the limits of detection are for this new modality.