An equivalent relative utility metric for evaluating screening mammography

Area of Science:

• Medical Imaging
• Signal Detection Theory
• Biostatistics

Background:

• Comparative screening mammography studies often yield ambiguous results.
• Higher sensitivity (detection rate) may be offset by increased false positives (recalls).

Purpose of the Study:

• To introduce an Equivalent Relative Utility (ERU) metric for quantifying comparative screening performance.
• To address limitations in applying utility theory to receiver operating characteristic analysis.

Main Methods:

• Developed the ERU metric based on signal detection theory.
• Defined ERU as the relative utility required to equate two screening systems.
• Estimated ERU using commonly reported recall and detection rates.

Main Results:

• ERU provides a quantifiable measure for comparing screening performance.
• The metric can be estimated without disease prevalence data in prevalence-matched populations.
• ERU analysis was demonstrated on two published screening mammography studies.

Conclusions:

• The ERU metric offers a robust and practical method for evaluating comparative screening mammography performance.
• This approach simplifies performance comparisons by utilizing readily available clinical study data.
• ERU enhances the interpretability of receiver operating characteristic analysis in screening studies.