(a) Mammography system with camera and LED side illumination setup attached to compression paddle; (b) example of a camera image for a compressed breast in CC view: the shadow area (segmented by dashed line) represents the area of contact between the breast and the compression paddle; and (c) status display showing compressed breast thickness, C-arm rotation angle, and compression force.
Example of time-registrations of the recorded parameters in CC compression (solid lines) and MLO compression (dashed lines) of the same breast. Circles and squares denote ends of deformation and clamp phases. (Bottom panel) Pressure (kPa) is derived as force (daN)/contact area (dm2); horizontal dotted lines indicate normal diastolic and systolic blood pressures as reference values.
Distributions of compression parameters. Breast thickness is normally distributed; the other parameters are not (based on Shapiro–Wilk test).
Estimated effects for nine pressure-controlled compression protocols in the CC and MLO views: (a) thickness increase relative to the 18 daN observations; (b) required compression force; and (c) corresponding pain score. Distributions of estimated values are shown as box plots (lines are medians; boxes are interquartile ranges; maximum whisker length is 1.5 times interquartile range; and circles are outliers). For comparison, distributions of the 18 daN force-controlled compressions (see Table I ) are included in (b) and (c) as interquartile ranges (light gray horizontal strips).
Models for pressure-controlled protocol estimations: (a) power fit-model used to estimate the thickness at 18 daN compression (arrows 1 and 2); the thickness at 10 kPa compression (arrows 3 and 4); and the force corresponding to 10 kPa compression (arrows 5 and 6). (b) Linear inter-/extrapolation model used to estimate the pain score corresponding to 10 kPa compression (arrows 7 and 8). Pain scores over 10 NRS are “saturated” to 10 (scale maximum).
Compression parameters; sample mean ± relative standard deviation in compression parameters and differences between CC and MLO views. Significance levels of paired t-test or Wilcoxon signed ranks: ***p < 0.001 and *p < 0.05.
Correlation between compression parameters and pain scores (Spearman correlation coefficients). Lower triangle: CC compressions and upper triangle: MLO compressions. Clinically relevant correlations are underlined. Dependent parameters (P = F/A) have “implicit” correlations. Superscript numbers (1) to (5) refer to the main results listed in the text. Significance levels: ***p < 0.001; **p < 0.01; and *p < 0.05.
Multivariate logistic regression models; coefficients of generalized linear mixed effects models for estimating the odds of experiencing severe pain (NRS ≥ 7). Individual models are isolated effects per predictor. Multivariate model 1 includes the two most significant predictors, contact area and baseline pain score, and model 2 is expanded with the next-most significant uncorrelated predictors, deformation phase (CC) and clamp phase (MLO). Significance levels: **p < 0.01 and *p < 0.05.
Compression parameters and pain scores stratified for breast contact area in three groups (MLO view). Comparison of the observations for 18 daN force (mean ± standard deviation) with the estimations for 10 kPa pressure (within squared brackets are the differences with respect to the 18 daN data) and the estimations for 10 daN force (within squared brackets are the differences with respect to the 10 kPa data).
Article metrics loading...
Full text loading...