Answers are grounded in this paper and cite their sources.
Film digitizers used to convert analogue mammograms into digital images vary in their grayscale response, and this variation propagates into the pixel values on which artificial intelligence (AI) screening classifiers are trained and deployed. The paper treats the resulting calibration problem as a formal measurement problem rather than a data-augmentation problem. A phantom-reanchored calibration operator is defined that maps each digitizer's grayscale response onto a common reference response using a calibration phantom, and the operating characteristics of AI recall decisions under that operator are characterised analytically. The analysis is a design and power study, not an observational one: no mammographic images or patient records were used. Under a declared variance structure with between-cluster standard deviation 6.0, residual standard deviation 4.5 and intraclass correlation 0.64, the before-after-control-impact estimator was evaluated across 2000 replications per design cell for cluster counts 8, 16, 24 and 32 and effect sizes 3, 5 and 8 percentage points. Fifty-four clusters were required to detect a 5 percentage point recall effect at 80 per cent power. The contribution is a calibration-aware evaluation framework that separates digitizer-induced grayscale distortion from genuine differences in diagnostic yield, together with the design parameters needed to test it.
No takes yet. Share an insight, caveat, or question.
Mavhunga et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: