Digital breast tomosynthesis: Systems, characterization and simulation

Abstract

Digital breast tomosynthesis (DBT) is an advanced imaging application for breast cancer detection, which makes use of a number of 2D X-ray projection images over a limited angular range to reconstruct quasi 3D reconstruction images. As such, first an introduction to digital X-ray tomosynthesis is given, after which existing limited angle tomosynthesis methods are presented. In the next section DBT geometry and its development to date is discussed. Details are given on geometries currently available from different manufacturers, and recent advances. Part of this section presents the available relevant reconstruction methods. Next, we look into the DBT detectors and their performance evaluation. Current DBT detectors are based on amorphous silicon (a-Si) and amorphous selenium (a-Se) thin film transistor (TFT) technology. However, complementary metal-oxide-semiconductor (CMOS) active pixel sensor (APS) digital X-ray detectors have the potential to replace a-Si:H TFT detectors in DBT in the near future, due to a smaller pixel pitch, low electronic noise, faster frame rate and negligible image lag. The performance of current DBT detectors is evaluated mainly by: automatic exposure control performance, response function, noise analysis, detector element failure, and system projection modulation transfer function (MTF). Following this we discuss image quality measurements, because they are essential for the evaluation and optimization of DBT systems. They should represent relevant clinical tasks, such as the detection of microcalcifications and masses in mammographic backgrounds. Currently the CDMAM phantom is used for contrast-detail analysis (i.e., the required threshold contrast to detect discs of various diameters) of the reconstructed images. The TOR MAM phantom can also be used to score the visualization of discs, filaments and specks for various contrast levels. The parameter Z-resolution is used to assess the inter-plane spread of reconstruction artifacts associated with 1 mm diameter aluminum spheres (contained in a specific three-dimensional phantom). Furthermore, the system MTF in the x-y plane is used to take into account all sources of blurring in the DBT system: detector MTF, additional sources of unsharpness and the reconstruction algorithm. The final part of the chapter describes image simulation methods for DBT optimization. Briefly, DBT is currently under consideration for its use in breast cancer screening in Europe, in combination with 2D mammography or alone. Several parameters (such as image acquisition parameters, detector response, system geometry, radiation dose, and image processing and reconstruction algorithms) are studied for their effect on image quality and the investigation of the optimum use of DBT. Traditionally, large clinical trials are required to evaluate these parameters over a large number of women. Such trials are time consuming, expensive and require irradiating asymptomatic women. Alternatively, several research groups use virtual clinical trials (based on image simulation methods) to optimize DBT parameters in fast, radiation-free, and costeffective ways. This part of the chapter reviews several simulation methods in DBT and the applications in evaluating its effectiveness.