Intrafraction Variability and Deformation Quantification in the Breast

Abstract

To evaluate intrafraction variability and deformation of the lumpectomy cavity (LC), breast, and nearby organs. Sixteen left-sided postlumpectomy and 1 bilateral breast cancer cases underwent free-breathing CT (FBCT) and 10-phase 4-dimensional CT (4DCT). Deformable image registration was used for deformation analysis and contour propagation of breast, heart, lungs, and LC between end-exhale and end-inhale 4DCT phases. Respiration-induced motion was calculated via centroid analysis. Two planning target volumes (PTVs) were compared: PTV(FBCT) from the FBCT volume with an isotropic 10 mm expansion (5 mm excursion and 5 mm setup error) and PTV(4DCT) generated from the union of 4DCT contours with isotropic 5 mm margin for setup error. Volume and geometry were evaluated via percent difference and bounding box analysis, respectively. Deformation correlations between breast/cavity, breast/lung, and breast/heart were evaluated. Associations were tested between cavity deformation and proximity to chest wall and breast surface. Population-based 3-dimensional vector excursions were 2.5 ± 1.0 mm (range, 0.8-3.8 mm) for the cavity and 2.0 ± 0.8 mm (range, 0.7-3.0 mm) for the ipsilateral breast. Cavity excursion was predominantly in the anterior and superior directions (1.0 ± 0.8 mm and -1.8 ± 1.2 mm, respectively). Similarly, for all cases, LCs and ipsilateral breasts yielded median deformation values in the superior direction. For 14 of 17 patients, the LCs and breast interquartile ranges tended toward the anterior direction. The PTV(FBCT) was 51.5% ± 10.8% larger (P<.01) than PTV(4DCT). Bounding box analysis revealed that PTV(FBCT) was 9.8 ± 1.2 (lateral), 9.0 ± 2.2 (anterior-posterior), and 3.9 ± 1.8 (superior-inferior) mm larger than PTV(4DCT). Significant associations between breast and cavity deformation were found for 6 of 9 axes. No dependency was found between cavity deformation and proximity to chest wall or breast surface. Lumpectomy cavity and breast deformation and motion demonstrated large variability. A PTV(4DCT) approach showed value in patient-specific margins, particularly if robust interfraction setup analysis can be performed.