Mitchell M. Goodsitt PhD, Presenter: Research collaboration, General Electric Company

Heang-Ping Chan PhD, Abstract Co-Author: Nothing to Disclose

Lubomir M. Hadjiiski PhD, Abstract Co-Author: Nothing to Disclose

Emmanuel G. Christodoulou PhD, Abstract Co-Author: Nothing to Disclose

Sandra Larson PhD, Abstract Co-Author: Nothing to Disclose

Paul L. Carson PhD, Abstract Co-Author: Research collaboration, General Electric Company Research collaboration, Sonetics Ultrasound, Inc Research collaboration, ZONARE Medical Systems, Inc Research collaboration, Light Age, Inc

Scott Zelakiewicz, Abstract Co-Author: Employee, General Electric Company

Andrea Schmitz, Abstract Co-Author: Employee, General Electric Company

Mark Alan Helvie MD, Abstract Co-Author: Institutional Grant, General Electric Company

Chintana P. Paramagul MD, Abstract Co-Author: Nothing to Disclose

Colleen H. Neal MD, Abstract Co-Author: Consultant, Medken LLC

A reader study was performed to evaluate the impact of acquisition geometry (total angle and number of projection views (pv)) on the perception of contrast-detail (CD) objects in digital breast tomosynthesis (DBT).

Modular breast phantoms consisting of slabs that mimic the composition and parenchymal pattern of breast tissue were imaged using a GE prototype DBT system. Two slabs, one with a homogeneous and one with a heterogeneous background were machined to include CD arrays of 25 holes, 1-5 mm in diameter and 0.2-1 mm in depth. Slabs were arranged to create 4 different 5cm thick phantoms with the heterogeneous CD slab and 1 with the homogeneous CD slab. Each phantom was imaged with 12 different acquisition geometries (total angles: 16-64o, pv: 9-21). Two acquisitions were repeated to study reproducibility. Identical x-ray technique factors were used with a mean glandular dose of a digital mammogram (~1.2mGy). Focal slices of the SART-reconstructed CD arrays were selected. 91 image pairs (IPs) were formed from the 14 geometries for each phantom, resulting in a total of 455 IP comparisons in a reader preference study. The IPs were randomized and evaluated by 4 trained readers on a 3 point scale (1= preferred image, 0.5=similar, 0=not preferred). The total % scores for all images and readings for each geometry were compared.

For the heterogeneous CD slab, the 3 highest scoring geometries and percent preferred were: 60o21pv(98%), 64o17pv(80%), and 48o17pv(71%). The lowest scoring were: 16o17pv(10%), 24o9pv(18%), and 24o13pv(35%). For the homogeneous CD slab, the highest scoring were: 60o21pv(87%), 32o17pv(82%), and 48o17pv(77%). The lowest scoring were: 16o17pv(1%), 40o11pv(18%), and 24o9pv(20%). Average % scores for repeat scans were within 4% in the homogeneous CD images and 7% in the heterogeneous CD images.

In general, CD objects are better perceived with wide-angle than narrow-angle DBT. For this study, the 60o21pv acquisition yielded the best perception of CD objects. These results may be applicable to mass perception. The optimal acquisition geometry should be a compromise that accounts for perception of calcifications and soft tissue lesions and scan time.

DBT systems can be designed with a range of acquisition angles and angle increments. This study demonstrates the perception of contrast-detail objects and possibly masses is best with wide-angle DBT.

Goodsitt, M, Chan, H, Hadjiiski, L, Christodoulou, E, Larson, S, Carson, P, Zelakiewicz, S, Schmitz, A, Helvie, M, Paramagul, C, Neal, C, Digital Breast Tomosynthesis: Reader Study of the Effects of Acquisition Geometry on the Perception of Contrast-detail Test Objects.  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13023077.html