The goal of this university-industry collaborative grant is to improve the imaging performance of clinical digital breast tomosynthesis (DBT). Our objectives are twofold: 1. Increase the inherent signal to noise ratio (SNR) of DBT by improving the detective quantum efficiency (DQE) of amorphous selenium (a-Se) full-field digital mammography (FFDM) detectors, and optimizing the x-ray delivery scheme. 2. Establish image quality evaluation methods that quantify the improvement in lesion conspicuity with different DBT implementations, with the results verified on a prototype DBT system. Our objectives will be accomplished through the following specific aims: (1) Develop a new detector with improved low dose and temporal performance for DBT.
Aim 1 will lead to the development and clinical translation of a new a-Se flat-panel imager that can double the DQE of the detector for each projection view of DBT, and increase the DQE by 30% for the higher energy beams used in contrast enhanced DBT imaging applications. (2) Develop a task-specific SNR framework to guide the optimization of DBT image acquisition for the detection of breast lesions.
Aim 2 will lead to the development of a frequency domain task-based image evaluation framework that quantifies the impact of detector performance, radiation delivery scheme and reconstruction filters on the detection tasks of breast lesions using reconstructed images. (3) Develop a DBT simulation platform with realistic 3D digital breast phantom and a computerized observer to quantify breast lesion conspicuity and verify optimization results.
Aim 3 complements Aim 2 to form a balanced evaluation methodology. It will investigate to what extent linear system assumptions can be used, and provide an image evaluation method where linear system theory does not apply, i.e. non-stationary background and system response, and non-linear reconstruction algorithms. (4) Verify optimization strategies for different lesion detection tasks on a prototype DBT system.
Aim 4 will implement the imaging system hardware improvements and use the framework developed in Aims 2 and 3 to guide the development of new, practical strategies to improve breast lesion conspicuity in a clinical prototype DBT system. Digital breast tomosynthesis has been receiving increasing attention since it was proposed more than a decade ago, yet more knowledge is needed in how to best utilize its 3D imaging capability while overcoming inherent system limitations.
We aim to take advantage of the capabilities of university-industry collaboration to contribute to this knowledge base. This will support rapid clinical translation of improvements in both system hardware and optimization methodologies.
The goal of this university-industry collaborative grant is to develop engineering methods to improve the clinical imaging performance of digital breast tomosynthesis (DBT) for the early detection of breast cancer. The improvements will be quantified by imaging theory and verified by clinical feasibility testing. The knowledge gained will be used to guide the design of future multi-institutional clinical trials.
