Summary
A total of 2986 subjects, with breast sizes and radiographic densities representative of a screening population and compressed thicknesses from 3.5 to 6 cm, were simulated and imaged on in-silico versions of DM and DBT systems using fast Monte Carlo x-ray transport. Images were interpreted by a computational reader detecting the presence of lesions. The in-silico trial (VICTRE) was designed to replicate a comparative trial from a previous regulatory submission. The endpoint was the difference in area under the receiver-operating-characteristic curve between modalities (delta-AUC) for lesion detection. Using a fully-crossed design, VICTRE was sized for a standard error (SE) of 0.01 in delta-AUC, half the uncertainty seen in the comparative trial.
Expensive and lengthy clinical trials delay regulatory evaluation of innovative medical technologies affecting patient access to high-quality medical products. Sophisticated simulation tools are increasingly being used in device development, but are rarely used in regulatory applications. We investigate a new paradigm for evaluating digital breast tomosynthesis (DBT) as a replacement for digital mammography (DM), using exclusively in-silico methods.
Researchers studying radiomics will be able to evaluate features for robustness across a variety of scanners. Features can be calculated using the researchers own software or third party software such as IBEX (imaging biomarker explorer). Related publications: http://journals.lww.com/investigativeradiology/Abstract/2015/11000/Measuring_Computed_Tomography_Scanner_Variability.3.aspx The following paper was generated on different imaging modalities but the same phantom, this is a related but independent paper with a different set of authors: http://tinyurl.com/zm7tr7p |
This data set was provided to TCIA by Authors: Mackin, Dennis; Fave, Xenia; Zhang, Lifei; Fried, David; Yang, Jinzhong; Taylor, Brian; Rodriguez-Rivera, Edgardo; Dodge, Cristina; Jones, Aaron Kyle; and Court, Laurence.
|
