2382 - Evaluation of Automated Skin Flash for Breast Cancer Patients

T. Nykanen^1,2, V. Raatikainen¹, A. Kärnä¹, and T. Koivumäki¹; ¹Deparment of Medical Physics, Hospital Nova of Central Finland, Wellbeing Services County of Central Finland, Jyväskylä, Finland, ²Deparment of Physics, University of Jyväskylä (JYU), Survontie 9 C, 40014 Jyväskylä, Finland, Jyväskylä, Finland

Purpose/Objective(s): A new automated skin flash feature for a treatment planning algorithm prototype was evaluated on breast cancer patients simulating body outline changes. Materials/

Methods: Study was done retrospectively on ten breast cancer patients’ CT-data. The results calculated with the new automated skin flash (SF) feature were compared to plans using a conventional virtual bolus (VB) method [1]. Both plans were calculated using a 13 mm PTV expansion extending outside the breast contour. A 16 mm thick bolus was used in VB. The original plan was normalized to ‘Target mean’ of the PTV-skin structure, where 3 mm from the surface of the body was removed. Four tangential VMAT fields or four tangential VMAT fields with integrated static angle ports were used for all patients. Swelling of breast tissue was simulated with 4, 8 and 12 mm expansions on the breast contour. HU of -100 was used for skin flash, virtual boluses and the expansion. The original plans were recalculated on the expanded CT-data. The recalculated plans were normalized to have the same monitor units as the original plans. PTV-skin 95 % planned dose coverage and mean skin dose were recorded from the original plan and the expanded plans. The PTV-skin structure was recreated on the expanded plans. Skin structure was defined as 5 mm from the surface of the body structure. Wilcoxon singed-rank test was used to evaluate statistically significant difference between VB and SF plans with the level p<0.05.
Results: The original plans with the VB and SF had comparable PTV coverage (Table 1): The VB plan median PTV coverage ( was 93.8 % and with SF 94.9 %. Median skin doses were 91.0 % and 89.9 % respectively. Median PTV coverage dropped by 1.8, 3.3 and 6.1 percentage points (pp) with the VB and 2.2, 4.5 and 8.9 pp with the SF in 4, 8 and 12 mm expanded plans compared to the original plans. Skin dose dropped by 3.1, 4.5 and 5.3 pp and by 3.2, 4.8 and 5.8 pp respectively. With VB and SF all expanded plans had a skin dose of over 84 % and 82 %, respectively. Without PTV expansion skin dose of ca. 60 % with 10 mm of swelling has been reported [1].
Conclusion: The new automated skin flash method provides clinically comparable mean dose in the near skin volume compared to the manual virtual bolus method. Abstract 2382 – Table 1 * Denotes statistically significant difference between VB and SF (p<0.05)

	PTV (%)	Skin (%)
VB Original plan	93.8[91.6, 96.3]	91.0[89.2, 91.7]
SF Original plan	94.9[94.3, 95.4]	89.9[88.3, 91.1]*
VB 4 mm expansion	92.0[89.4, 94.8]	87.9[86.0, 88.8]
SF 4 mm expansion	92.7[91.2, 94.0]	86.7[84.1, 88.0]*
VB 8 mm expansion	90.5[87.3, 92.8]	86.5[85.2, 87.3]
SF 8 mm expansion	90.4[88.9, 92.0]	85.1[83.1, 86.3]*
VB 12 mm expansion	87.7[83.4, 90.5]	85.7[84.2, 86.7]
SF 12 mm expansion	86.0[82.6, 88.6]	84.1[82.3, 84.5]*