Impact of common artefacts on mass visibility in contrast-enhanced mammography: A virtual imaging trial.

BACKGROUND

Artefacts in contrast-enhanced mammography (CEM) can compromise diagnostic image quality. These artefacts may obscure clinically relevant findings or mimic pathological features, potentially leading to misinterpretation.

PURPOSE: A virtual imaging trial technique was used to quantify the impact of breast-in-breast and motion artefacts on lesion contrast and signal-difference-to-noise ratio (SDNR) in CEM.

METHODS

Computational breast phantoms representing four breast density categories were generated, and 5 mm diameter non-spiculated masses were embedded centrally and peripherally to study the influence of density and the breast-in-breast artefact. Motion artefacts were simulated using two techniques: horizontal and vertical translation of the entire phantom between the low-energy (LE) and high-energy (HE) acquisitions, and simulation of partial paddle release between the two exposures. A Monte Carlo (MC) method was used to model the LE and HE images generated by a Siemens MAMMOMAT B.brilliant system. An offline algorithm with motion correction generated the recombined images. Lesion contrast and SDNR were evaluated for simulated masses with clinically relevant iodine concentrations derived from patient data and QC phantoms.

RESULTS

The breast-in-breast artefact causes contrast and SDNR reductions of up to 76% for peripheral lesions. There was a slight increase in these parameters as breast density increased, due to the fixed weighting factor used by the recombination algorithm. Simulated rigid and non-rigid motion reduced SDNR by up to 60% before motion correction. Activation of the motion correction algorithm preserved SDNR for small in-plane displacements, although some residual artefacts remained for non-rigid motion.

CONCLUSIONS

This study highlights the challenges posed by non-uniform breast thickness at the borders and emphasizes the importance of motion correction. Virtual imaging trials provide a powerful, controlled method to study these effects, offering insights that are challenging to obtain from clinical or physical phantom studies alone.