Mammography machine compression paddle movement and
observer performance analysis

Ma, WK

Abstract

Full field digital mammography (FFDM) was introduced into the United Kingdom (UK) as a replacement for screen-film mammography (SFM) in 2005. Since then, individual breast screening centres have begun to report blurred images through local audits. Blurring was probably present in SFM as well, however the improvement in contrast resolution in FFDM may have made it more apparent. The sources of blurring include improper imaging techniques, patient movement caused by breathing and heart motion,the viscoelastic motion of the breast, and paddle motion. This thesis aims to test the hypothesis that paddle motion might cause image blur. It investigates whether blurring can be detected visually on technical review monitors and reporting grade monitors.
The thesis presents a method to minimise paddle motion during X-ray exposure. Six papers have been published. Two of these (papers 1 and 2) investigated paddle displacement using linear potentiometers. Three investigated the influence of paddle motion on image quality. Paper 3 investigated whether paddle motion can cause image blur; paper 4 determined the minimum amount of simulated motion required for the visual detection of blurring; and paper 5 evaluated the practitioner’s ability to identify blurring on monitors with different resolutions (2.3 MP and 5 MP). The final research paper (paper 6) investigated a way to reduce paddle displacement settling time; this involved the use of a closed-loop control system.

Results: In papers 1 and 2 paddle displacement followed a bi-exponential function with a settling time of approximately 40 s. The use of average paddle displacement to estimate the amount of paddle motion would underestimate the worst case of the three
different runs of the experiment. The estimated paddle motion would be greatly reduced if the time of exposure is delayed from 5 to 10 s.

In paper 3 all metal ball bearings shown increased in diameters and the range of magnification varied from 1.04 to 1.21. T-test results shown that there was a significant difference (p < 0.05) in the ball bearing diameters between the intensity thresholding and the edge detection methods for all paddle/ compression force combinations. The ball bearing diameters calculated by the intensity thresholding method had higher variability than the edge detection method.

In paper 4 the soft-edged mask method best represented the physical process that caused the blurring effect and was chosen as the standard simulation approach for motion blurring. The ratio between the vertical paddle motion and the horizontal breast motion estimated by the mathematical model is approximately 1:0.3.

In paper 5 the angular size calculation shown that for a viewing distance of 75 cm the screen resolution for 5 MP and 12 MP monitors was better than the observer eyes' resolution. For a viewing distance of 30 cm the observer eyes' resolution was better
than the screen resolution for 2.3 MP, 5 MP and 12 MP monitors. Among all three monitors, image displayed on the 12 MP monitor has the lowest loss in image quality after interpolation.

In paper 6 the simulation results shown that force overshoot is possible for position control system. Force overshoot occurred almost instantaneously for step input and its magnitude is about 10 times larger than the ramp input. Force overshoot and steadystate
error can be eliminated by the use of force control system.

Conclusion: The magnitude of calculated paddle motion is much lower than the minimum amount of simulated motion required for the visual detection of blurring. Mathematical models have shown that vertical paddle motion caused a smaller horizontal breast displacement when compressed. Therefore, there is no sufficient
evidence to support the hypothesis that paddle motion is a cause of image blurring in FFDM.

Citation

observer performance analysis. (Thesis). University of Salford

• Thesis Type: Thesis
• Deposit Date: Feb 5, 2021
• Publicly Available Date: Feb 5, 2021
• Award Date: Jan 29, 2021