Large breast compressions: Observations and evaluation of simulations
Compression device (a) with and (b) without padding.
Illustration of employed coordinate system and breast regions (core, outer).
Axial example slices of included cases showing (from left to right) undeformed breast (target), maximally compressed breast (source), difference (target-source), difference after nonrigid registration (target-transformed source), and transformed source. S7 had extra medial padding.
Axial example slices of excluded cases showing (from left to right) undeformed breast (target), maximally compressed breast (source), difference (target-source), difference after nonrigid registration (target-transformed source), and transformed source.
SI projection of the landmarks and the segmentation of the maximally compressed breasts. The landmark positions are marked by o (maximum compression C1) and x (less or not compressed) and connected by lines. The contours illustrate the thickness of the segmented breast in the SI direction.
(a) Comparison of observed mean stretches in the anterior-posterior and superior-inferior directions to the theoretic stretches for incompressible, isotropic material (black line). (b) Comparison of observed stretch ratios for the core and the outer breast regions.
Boxplots and mean value of (a) predicted mean displacement errors for a combination of either anisotropic (aniso) or isotropic (iso) material and heterogeneous (het) or homogeneous (hom) tissue properties. (b) Difference to the anisotropic heterogeneous results.
Illustration of breast shape change during compression for simulations using (a) isotropic material and (b) transverse isotropic material in comparison with (bold black line) the breast shape of the corresponding real mammogram.
Relationship between the breast shape and the progression of the simulated compression for (a) isotropic and (b) transverse isotropic materials. A quadratic function (line) was fitted to the thickness-versus-shape values (crosses) measured during a FE simulation at 4% compression intervals. Circles depict the estimated breast thickness for the breast shape of the corresponding real mammogram predicted from the quadratic function. Very unrealistic and even impossible negative breast thicknesses are estimated for models with isotropic materials.
Overview of subjects (S1–S11), listing compressed side, state of lesion [none, enhanced no pathology, pathology result (IDC: Invasive ductal carcinoma; G: Grade)], subject age, breast volume, percentage of fibroglandular tissue, and applied compressions (C1–C4). Successfully registered subjects are marked with an asterisk.
Overview of nonlinear material models from Krouskop et al. (Ref. 20), Wellman (Ref. 21), and Van Loocke et al. (Ref. 22). The same Young’s modulus was used for compression as for tension, i.e., .
Mean displacement error and material properties after optimizing with respect to the registration result over the unrestricted parameter space for eight subjects (S) and one to four compressions (C). Material parameters are expressed as ratios of the Young’s modulus , namely, the anisotropic stiffness ratio to the coronal plane and the stiffness ratio to fat for fibroglandular tissue , tumor , muscle , and skin . Standard deviation (SD) was employed to measure overall, intrasubject (intraS) and intracompression group (intraC) variability.
Article metrics loading...
Full text loading...