RSNA 2019

Abstract Archives of the RSNA, 2019


SSC13-03

Comparison of Digital Mammograms, Breast Tomosynthesis, and Synthetic Mammograms for Detection of Masses: An Observer Performance Study

Monday, Dec. 2 10:50AM - 11:00AM Room: S503AB



FDA Discussions may include off-label uses.

Participants
Liesbeth Vancoillie, Leuven, Belgium (Presenter) Nothing to Disclose
Lesley Cockmartin, Leuven, Belgium (Abstract Co-Author) Nothing to Disclose
Nicholas Marshall, Leuven, Belgium (Abstract Co-Author) Nothing to Disclose
Hilde Bosmans, PhD, Leuven, Belgium (Abstract Co-Author) Stockholder, Qaelum NV Research Grant, Siemens AG Research Grant, General Electric Company Research Grant, Agfa-Gevaert Group

For information about this presentation, contact:

liesbeth.vancoillie@uzleuven.be

PURPOSE

Compare detectability of masses in digital mammography (DM), digital breast tomosynthesis (DBT) and synthetic mammograms (SM) of 5 vendors with a 3D structured (L1) phantom with embedded mass-like lesions.

METHOD AND MATERIALS

L1 is a hemispherical shaped phantom filled with PMMA spheres and water plus nine 3D-printed lesions: 5 non spiculated (diameter 1.6mm to 6.2mm) and 4 spiculated masses (diameter 3.8mm to 9.7mm). DM, DBT and SM images were acquired on these systems: Fujifilm Amulet Innovality ST, GE HC Senographe Pristina, Hologic Selenia Dimensions, IMS Giotto Class and Siemens Mammomat Revelation. Three dose levels were studied (12 acquisitions at each level): automatic exposure control (AEC) level and manually set at AEC/2 and 2xAEC. A 4-alternative forced choice reading paradigm was used. Threshold diameter to reach a 62.5% correct score was evaluated (dtr). One-way analysis of variance was performed to test for significant differences among dose levels/modalities.

RESULTS

For GE, Giotto and Siemens, DBT performed significantly better than SM, while SM showed no difference with DM. For Fuji and Hologic, there was no significant difference between DBT and SM, while DM was inferior to DBT and SM. The dtr values for non-spiculated masses for DM/DBT/SM at AEC dose, were respectively: Fuji: 4.21mm/2.12mm/2.89mm; GE: 4.87mm/2.15mm/4.21mm; Hologic: 7.21mm/1.87mm/3.28mm; Giotto: 4.96mm/2.37mm/4.30mm; Siemens: 4.44mm/2.22mm/4.64mm and for spiculated masses: Fuji: 6.65mm/2.58mm/3.66mm; GE: 4.27mm/2.04mm/3.55mm; Hologic: 5.04mm/2.54mm/3.68mm; Giotto: 4.97mm/2.37mm/4.30mm; Siemens: 5.6mm/2.99mm/4.67mm. Dose did not impact detection of both mass types for GE and Hologic. For Fuji, Giotto and Siemens DBT, decreasing the dose lead to a significantly inferior dtr for spiculated masses and in the case of Fuji and Siemens also for non-spiculated masses.

CONCLUSION

Detectability of mass-like lesions was higher in DBT compared to DM and SM, except for Fuji and Hologic, where SM was not different from DBT. Increasing dose only influenced mass detection in DBT. The L1 phantom demonstrated the superiority of DBT compared to DM for mass detection, for all 5 systems.

CLINICAL RELEVANCE/APPLICATION

For most vendors, SM, in its current stage of development, cannot be recommended as a stand-alone modality if equal mass detectability as in DBT is required.

Printed on: 03/01/22