Duke Breast Models Prepared for Allergan & Ansys Paul Segars Greg Sturgeon
Types of Data Breast MRI (prone) Patient CT (supine) Breast CT (prone) Resolution: 0.85 mm XY; 1.6 mm Z Breast CT (prone) Resolution: 0.97 mm XY; 3.27 mm Z Resolution: 0.155 to 0.25 mm
Two Sources of Breast Models Full Body XCAT Models Isolated High Resolution Breast Models Based on patient CT data with ‘morphed’ structures Segmented breast CT data
Full Body XCAT Models Based on patient CT data ~3000 structures modeled as individual NURBS surfaces. Some objects built from patient specific segmentations Some ‘morphed’ from detailed reference phantoms
Extrapolating the FOV Part of the torso or breast may be outside of the reconstruction in some CT scans. NURBS surfaces created for the overall body habitus and breasts. Geometry is extrapolated from imaging and patient BMI. Full-body XCAT models do not currently have patient specific fibroglandular tissue or ducts modeled. Possible to segment inner tissue from CT if not truncated
Variability of Full Body XCAT Models
High Resolution Breast Models Segmented Breast CT .155 mm voxel resolution Segmented into 6 classes (skin + 4 fibroglandular levels + adipose) Imaged in pendant geometry Limited pectoral muscle / no chestwall
Size / Density Variability of High Resolution (isolated) Breast Models
Addition of Breast (based upon Breast CT) to XCAT model +
FE compression using template hex mesh Modeled gravity and compression to allow for simulated mammography and breast tomosynthesis imaging. Chest wall modeled simplistically Patient specific chest wall information wasn’t available for these cases. Use of template mesh allowed for easy mapping between breast geometries. Allow for breasts to be mapped to consistent shell.
Subpopulation of breast models – mapped to consistent geometry Subpopulation of larger less dense breasts. Mapped to a consistent compressed breast ‘shell’. Allows of modeling of breast texture independent of geometry.