DBT has been shown to be a promising modality for detecting breast cancer, offering higher cancer detection rates and fewer false positives than traditional 2D mammography. However, these benefits come at the cost of higher radiation exposure; the two DBT clinical scenarios approved by the U.S. Food and Drug Administration (FDA) could deliver a 1.4 to 2.3 times higher radiation dose than mammography, according to senior author Kenji Suzuki, PhD, of the of the Illinois Institute of Technology in Chicago.
DBT for annual screening could increase cumulative radiation exposure and the lifetime risk for radiation-induced breast cancer; therefore, it is important to reduce DBT radiation dose as much as possible, Suzuki said. To address this issue, the group developed a supervised image-processing technique based on a deep-learning method called neural network convolution (NNC).
The researchers' NNC converts low-dose DBT images to "virtual" high-dose images; noise and artifacts are significantly reduced in the low-dose images, while breast tissue and subtle structures such as tiny microcalcifications are maintained. This results in approximately 80% dose reduction, according to Suzuki.
What's more, a blinded observer study involving 35 breast radiologists and a database of 51 clinical cases found that virtual high-dose images acquired in half-dose acquisitions were equivalent in image quality to real, full-dose images, according to the team, which included presenter Junchi Liu of the Illinois Institute of Technology and Dr. Laurie Fajardo of the University of Utah.
"Unlike other deep-learning models, our original NNC deep learning requires a very small number of training cases, which is an extremely important property in our medical imaging domain," Suzuki told AuntMinnie.com. "Substantial radiation dose reduction by our technique would benefit patients by reducing the risk of radiation-induced cancer from DBT screening."
