Escalating public debate about the value of standard screen-film mammography in detecting breast cancer and decreasing cancer mortality has put radiology researchers on the spot. Those involved in clinical trials of new digital mammography devices are developing ways to ensure that the images are of the highest quality.
Diagnostic Imaging spoke with Dr. Etta Pisano, chief of breast imaging at the University of North Carolina School of Medicine in Chapel Hill and chief investigator of the Digital Mammographic Imaging Screening Trial (DMIST).
DI: Based on your experience, do you have an opinion regarding screen-film versus digital mammography?
Pisano: Studies have shown no statistically significant difference between digital and screen-film mammography, although specificity-the ability to tell cancer from noncancer-does appear to be increased. That increased specificity is attributed at least partially to soft-copy display systems and the ability to manipulate the data.
DI: So the jury is still out?
Pisano: I think any time a new device comes along, the tendency is to assume it is better than what is being done currently. That's a reasonable assumption. But when women's lives are at stake, we don't want even a small change in sensitivity. I'm optimistic that digital will prove to be at least as good, but I would like to see it perform better since the cost is so high.
DI: It sounds like the stakes are high as well.
Pisano: We have to judge if we care more about false positives and reducing those or about false negatives and reducing those. My sense is that the most pressing need is to find cancers earlier. To do that is to increase sensitivity and not specificity.
DI: Will DMIST address the role of observer variability in soft-copy reading of digital mammograms?
Pisano: It may be true, as some researchers have found, that observer variability accounts for much of the difference between screen-film and digital mammography. But over the course of DMIST, which will involve many thousands of scans with readers interpreting in both modalities, any variability should wash out. I expect observer variability will be kind of a nuisance variable that goes across both modalities.
DI: What guidelines exist now for ensuring the quality of digital mammograms?
Pisano: The companies with FDA-approved devices have published manuals on how to improve the quality of the image. But we don't have anything like the ACR recommendations for film mammography in place. And there is clearly a need for standardized rules across companies, so users will be able to assess the quality of their digital mammograms.
DI: When might that happen?
Pisano: I'm on the national panel that we hope will someday be deciding what the requirements for digital mammography will be part of the Mammography Quality Standards Act. The panel needs to work with the ACR and come up with a uniform standard. I don't think we will have to do as much as the manufacturers are currently mandating that we do.
DI: Will we have to wait until the DMIST trial is complete before a standards document is in place?
Pisano: Not at all. We've been collecting quality control data religiously since October 2001 at all of the participating sites. I would like to see the FDA panel that I'm involved with use the data we've acquired.
