Quantitative breast tomosynthesis: from detectability to estimability.
This work aimed to extend Fourier-based imaging metrics for modeling and predicting quantitative imaging performance. The new methodology was applied to the platform of breast tomosynthesis for investigating the influence of acquisition parameters (e.g., acquisition angle and dose) on quantitative imaging performance.Two quantitative imaging tasks were considered: Area estimation and volume estimation of a 4 mm diameter spherical target. The maximum likelihood estimator yielded training data to generate a size estimation task function, which was combined with the MTF and NPS to predict estimation performance by computing an "estimability index" analogous to the detectability index. Estimation performance for the two tasks was computed as a function of acquisition angle and dose. The results were used for system optimization in terms of quantitation performance and further compared to the detectability index for the detection of the same spherical target.The estimability index computed with the size estimation tasks correlated well with precision measurements for area and volume estimation over a broad range of imaging conditions and provided a meaningful figure of merit for quantitative imaging performance and optimization. The results highlighted that optimal breast tomosynthesis acquisition parameters depend significantly on imaging task and dose. At nominal dose (1.5 mGy), mass detection was optimal at an acquisition angle of 85 degrees, while area and volume estimation for the same mass were optimal at approximately 125 degrees and 164 degrees acquisition angles, respectively.These findings provide an initial validation that the Fourier-based metrics extended to estimation tasks can represent a meaningful metric and predictor of quantitative imaging performance. The optimization framework also revealed trade-off between anatomical noise and system noise in volumetric imaging systems potentially identifying different optimal acquisition parameters than currently used in breast tomosynthesis and CT.
Volume / Issue
Start / End Page
Pubmed Central ID
Electronic International Standard Serial Number (EISSN)
International Standard Serial Number (ISSN)
Digital Object Identifier (DOI)