Room: Room 207
Purpose: This work reports the design, characterization, implementation, and assessment of multiple aperture devices (MAD) for achieving dynamic fluence field modulation (FFM) on an experimental cone-beam CT (CBCT) bench.
Methods: MAD filters consist of a series of thin tungsten bars of varying widths and spacings. Relative motions between two MADs produce MoirÃ© patterns that effect a wide range of fluence patterns. The dual-MADs are mounted on linear actuators permitting controlled motion on an experimental CBCT bench. Dynamic FFM is achieved through a combination of view-dependent MAD actuation for beam shape modulation and mAs for amplitude modulation. To correct for gains associated with the MADs, an algorithm was developed to account for focal spot shifts during/between scans and spectral effects from a hardened x-ray beam at the edge of the tungsten bars. Phantom-specific FFM was designed for an ellipse phantom (25.8x14.1 cm) based on flatness and minimum mean variance criteria. The effect of FFM on image quality was assessed in terms of the variance and local noise power spectrum (NPS) in filtered-backprojection (FBP) reconstructions.
Results: The experimental FFM for the ellipse phantom was achieved over 0.44 mm of actuation motion and closely matches the target profiles for both design objectives. The proposed correction algorithm effectively removes high-frequency ring artifacts otherwise present in conventional data processing. Moreover, phantom details including wires and contrast inserts are preserved. The effects of FFM on noise properties follow theoretical expectations: flat FFM criterion produces nearly isotropic NPS and uniform variance across the phantom; while the minimum mean variance FFM results in the lowest mean variance in all cases investigated in this work.
Conclusion: The dual-MADs system provides effective fluence and image quality control in a compact volume with small relative motions. Thus, this work has the potential to facilitate FFM in a clinical CT system.
Funding Support, Disclosures, and Conflict of Interest: This work is supported by NIH grant U01EB018758.