Impact of heel effect and ROI size on the determination of contrast-to-noise ratio for digital mammography systems

The European Guidelines for quality control in digital mammography specify a procedure for measuring contrast-to-noise ratio (CNR) using a 0.2mm thickness of aluminium with different thicknesses of Plexiglas. The relationship between ROI size and heel effect and how this affects CNR measurement is investigated in this work for DR and CR systems. The measured relative noise for the CR images was found to be strongly dependant on the ROI size due to the heel effect. After applying heel effect correction there was very little dependence on ROI size. The relative noise in the images from the DR system showed very little dependence on ROI size. The heel effect also distorted the CNR measurement on CR images when larger ROI is used. However the use of multiple small ROIs led to a result that was essentially the same as if a heel effect correction had been applied. The appropriate ROI size which should be used for CNR measurement was found to be 0.25 × 0.25 cm. Using this size the heel effect had an insignificant impact on the measurement of relative noise and CNR. This approach has the advantage that only a single image is required for each measurement. The application of heel effect correction with CR systems requires two images and complex image processing. The current suggestion in the European guidelines to use a 2 × 2cm ROI is inappropriate for CR systems and leads to an error of 8% to 18% in CNR determination due to the heel effect.