Breast tomosynthesis using the multiple projection algorithm adapted for stationary detectors

Area of Science:

• Medical Imaging
• Radiology
• Algorithm Development

Background:

• Breast Tomosynthesis (BT) aims to improve breast cancer detection.
• Evaluating novel reconstruction algorithms is crucial for clinical implementation.
• The Multiple Projection Algorithm (MPA) has shown promise in other imaging applications.

Purpose of the Study:

• To investigate the clinical validity of the Multiple Projection Algorithm (MPA) for Breast Tomosynthesis (BT).
• To compare MPA performance against Back Projection (BP) using real clinical projection data.
• To assess the impact of phantom thickness and pre-filtering on image quality.

Main Methods:

• Utilized a CIRS-BR3D phantom (3-6 cm thickness) with 2D mammography and 25 projections (50° arc).
• Adapted MPA for partial isocentric rotation with a stationary detector.
• Developed a Back Projection (BP) algorithm for comparison; evaluated pre-filtering effects.

Main Results:

• MPA and BP showed comparable performance in feature detection and image appearance.
• MPA demonstrated faster overall processing times.
• Pre-filtering improved Breast Tomosynthesis (BT) image quality, especially for thicker phantoms and superimposed structures.
• Increased phantom thickness limited detection of smaller features, but filtered BT slices enhanced visualization over 2D mammograms.

Conclusions:

• The Multiple Projection Algorithm (MPA) is time-efficient, compliant, and suitable for clinical Breast Tomosynthesis (BT) systems.
• Breast Tomosynthesis (BT) offers superior visualization of small features and dense backgrounds compared to 2D mammography.