2387 - Accuracy of a Physician-Free Workflow for Daily Adaptive Stereotactic Partial Breast Radiation Therapy (A-SPBI)

D. D. M. Parsons¹, S. Domal¹, D. Salazar¹, J. Visak¹, M. Arbab¹, F. Holgado¹, X. D. Li¹, C. Okoro¹, N. Wandrey¹, Z. Iqbal², P. G. Alluri¹, H. Nickens¹, H. Keen¹, E. Chambers¹, J. Cleaton¹, A. R. Godley¹, D. J. Sher¹, S. N. Badiyan², M. H. Lin¹, and A. S. Rahimi¹; ¹Department of Radiation Oncology, University of Texas Southwestern Medical Center, Dallas, TX, ²Department of Radiation Oncology, UT Southwestern Medical Center, Dallas, TX

Purpose/Objective(s): x-ray based adaptive radiotherapy allows markedly smaller PTV margins for patients treated with adaptive stereotactic partial breast irradiation (A-SPBI). However, this treatment approach imposes increased demands on the treatment team, particularly physicians. We hypothesize that with proper training the physician’s involvement at the treatment console can be reduced by delegating contouring and planning tasks to therapists under physicist supervision. Materials/

Methods: 23 prospective patients undergoing A-SPBI were included, with two treatment plans generated for each adaptive fraction. The first plan utilized contours edited by therapists under physicist supervision (without physician supervision), while the second plan utilized contours edited by physicians. Contours included the tumor bed, ipsilateral chest wall, bilateral breasts, lungs, adjacent skin, heart, and left anterior descending artery. The plan without physician supervision were evaluated against physician-edited contours for target coverage (V_100% and V_95%) and relevant organ-at-risk (OAR) planning constraints. Additionally, the Dice coefficient and Hausdorff distance we compared for the target volumes. Performance trends were analyzed based on fraction number and tumor bed volume, tumor bed volume change from simulation and fiducial density.
Results: Analysis of 97 adaptive fractions revealed Dice coefficients of 0.95±0.06, 0.98±0.03 and 0.98±0.02 for the tumor bed, CTV and PTV, respectively. Similarly, Hausdorff distance was 0.67±0.69, 0.75±0.71 and 0.78±0.73 mm. Target coverage differences included tumor bed V_100% of -0.00±0.21%, CTV V_100% and V_95% of -0.33±0.91% and -0.10±0.44%, respectively, and PTV V_100% and V_95% of -1.01±2.22% and -0.61±1.33%, respectively. Importantly, the CTV V_100% and V_95% in the plan without physician supervision was covered by =95% in 99.0% and 100% of plans, respectively. No significant correlations were observed with fraction number and tumor bed volume, tumor bed volume change from simulation and fiducial density. OAR doses showed no measurable differences between contour sets.
Conclusion: Delegating contouring and planning tasks to non-physicians is feasible to reduce physician workload without compromising treatment quality during on-line A-SPBI. Review of the contours in online/offline manner can further ensure on-going quality and consistency. Table: Summary between the MD supervised and unsupervised A-SPBI plans for Dice coefficient, Hausdorff distance and target coverage for tumor bed, CTV and PTV.

Metric	Target	Mean	Standard Deviation
Dice Coefficient	Tumor Bed	0.95	0.06
	CTV	0.98	0.03
	PTV	0.98	0.02
Hausdorff Distance	Tumor Bed	0.67 mm	0.69 mm
	CTV	0.75 mm	0.71 mm
	PTV	0.78 mm	0.73 mm
V100%	Tumor Bed	-0.00%	0.21%
V100%	CTV	-0.33%	0.91%
V95%		-0.10%	0.44%
V100%	PTV	-1.01%	2.22%
V95%		-0.61%	1.34%