Mammogram missile

News

The US Army Space and Missile Defense Command's (SMDC) Technical Center is investigating whether missile defence research can be used to improve mammogram imagery.

At the seventh annual Space and Missile Defense Conference, held on August 18 in Huntsville, Alabama, Craig Shriver of the Clinical Breast Care Project at Walter Reed Army Medical Center in Washington, DC, told attendees that mathematical algorithms used for missile defence might be able to be used to improve mammogram imagery.

Using the technique, Shriver believes surgeons will be able to see how big and what shape a cancer is, and use that information to improve planning for breast cancer surgery.

"We have developed the detection and discrimination algorithms, but we still have to test them on many mammograms," said Pete Kirkland, a senior research scientist with the Technical Center. "We can detect anomalies; however, discriminating cancerous from a noncancerous (tissue) is the big issue."

Kirkland said the team is working three specific tasks: using imaging devices to distinguish between cancerous and noncancerous anomalies; using 2D gel protein expression to determine early if a person might have cancer; and taking clinical and lifestyle data from Walter Reed and Vital Solutions in Huntsville to develop a ‘Response Surface Model’ to predict the probability of patients getting cancer.

"We are working with Jeffrey Zelickson, president of Vital Solutions, to develop a database that will be used to train the Response Surface Model to predict the probability that a patient will get cancer," Kirkland said. The model being developed would likely be used for prediction in a clinical setting.

"We think our approach is solid, but we do not want to make any claims until we test it against a big data set. We have more than 1,000 mammograms that we are processing. I think once we process most of those, we will have an answer,” he added.

"We are cautious because we have to be right," Kirkland said. "Whatever we do must have greater than a 90% probability of success. A radiologist would use this tool to help them make a diagnosis; it will not take the place of the radiologist.”

Dr. Robert Smith, a Huntsville radiologist, is donating his time to the effort as well. He is providing information about what radiologists look for when screening mammograms. The information is helpful when developing the algorithms.