RSNA 2012 

Abstract Archives of the RSNA, 2012


SSE02-03

Dose Reduction in Digital Breast Tomosynthesis with Prior Image Constrained Compressed Sensing (PICCS)

Scientific Formal (Paper) Presentations

Presented on November 26, 2012
Presented as part of SSE02: Breast Imaging (Digital Breast Tomosynthesis)

Participants

John W. Garrett BA, Presenter: Research Grant, Hologic Inc
Jie Tang PhD, Abstract Co-Author: Nothing to Disclose
Elizabeth S. Burnside MD, MPH, Abstract Co-Author: Research Grant, Hologic, Inc
Janet Beer Garrett MD, Abstract Co-Author: Nothing to Disclose
Guang-Hong Chen PhD, Abstract Co-Author: Research funded, General Electric Company Research funded, Siemens AG Research funded, Varian Medical Systems, Inc Research funded, Hologic, Inc

PURPOSE

Although digital breast tomosynthesis (DBT) is a promising new modality, a typical DBT acquisition does require a slightly higher radiation exposure to the patient than the current standard, digital mammography. The purpose of this study is to evaluate a dose reduction scheme using the Prior Image Constrained Compressed Sensing (PICCS) reconstruction in DBT.

METHOD AND MATERIALS

DBT data sets of an American College of Radiology (ACR) breast accreditation phantom and two cadaver breasts were obtained at varying exposure levels using a Hologic Selenia Dimensions DBT system. An implementation of the PICCS algorithm was tailored for use in DBT, and was used to reconstruct the images with reduced noise. The contrast to noise ratio (CNR) was measured for three regions of interest (ROIs) in the ACR phantom. Five radiologists performed a blinded study on Hewlett-Packard and Hologic SecurView workstations in which they were forced to choose between the commercial reconstruction at full dose and the PICCS reconstruction with 60% dose.

RESULTS

We found the CNR in each ROI to be higher at all dose levels using the PICCS reconstruction method than the currently used reconstruction in product. The CNR was improved by an average of 22.7±3.4% for a large lesion, 22.9±6.7% for a small lesion, and 35.2±7.4% for microcalcifications. In addition, the CNR of reduced dose PICCS reconstructions was higher than the full dose (organ dose: 1.41 mGy chosen with AEC) commercial reconstructions down to 72% (organ dose: 1.02 mGy) for large lesions and 61% (organ dose: 0.86 mGy) for small lesions and microcalcifications. In the low dose cadaver breast study, radiologists chose the reduced dose PICCS reconstruction as superior to the full dose commercial reconstruction for 53% of the cases while choosing the full dose commercial reconstruction 47% of the time.

CONCLUSION

At 60% radiation dose level, the PICCS reconstructions are diagnostically equivalent to the commercial reconstruction at full dose. In addition, PICCS was able to enhance the CNR of low contrast lesions without compromising important features.

CLINICAL RELEVANCE/APPLICATION

Using PICCS, dose can be reduced by 40% in digital breast tomosynthesis without sacrificing diagnostic quality.

Cite This Abstract

Garrett, J, Tang, J, Burnside, E, Garrett, J, Chen, G, Dose Reduction in Digital Breast Tomosynthesis with Prior Image Constrained Compressed Sensing (PICCS).  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12020832.html