2413 - Automated Stereotactic Radiosurgery Planning for Brain Metastases Based on Dose Prediction: A Dosimetric and Efficiency Analysis

Purpose/Objective(s): To develop an automated planning model based on dose prediction for stereotactic radiosurgery (SRS) of brain metastases and to evaluate its dosimetric quality, planning efficiency, and clinical feasibility.

Materials/Methods:

We retrospectively reviewed 128 patients with brain metastases treated with SRS. Patients had 1-6 metastases with prescription doses ranging from 8-22 Gy in 1 fraction, 24-36 Gy in 2 fractions, 18-39 Gy in 3 fractions, 36 Gy in 4 fractions, 25-36 Gy in 5 fractions, to 36 Gy in 6 fractions. Automated and manual plans were generated for each patient. We compared plan quality metrics including the conformity index (CI) and homogeneity index (HI) for the planning target volume (PTV), as well as organs-at-risk (OAR) doses (brain V12, V6, V3, maximum dose to brainstem and optic pathway), and planning time. The Shapiro-Wilk test was used to assess the normality of data distributions. Paired t-tests were used for normally distributed data, and Wilcoxon signed-rank tests for non-normally distributed data. A p-value <0.05 was considered statistically significant.

Results: The PTV conformity index was comparable between manual and automated plans [0.84 (0.83-0.91) vs. 0.82 (0.77-0.90), p=0.170]. Automated plans demonstrated significantly higher low-dose brain volumes: V3 was 269.06 cc (74.02-368.13 cc) vs. 237.55 cc (68.20-305.04 cc) (p<0.001), and V6 was 107.96 cc (25.84-134.91 cc) vs. 97.07 cc (23.41-129.73 cc) (p=0.011). The maximum dose to the brainstem was also significantly higher with automated planning [863.96 cGy (126.45-1292.78 cGy) vs. 793.13 cGy (164.78-1195.91 cGy), p=0.048]. No significant differences were observed in brain V12 (p=0.426) or maximum dose to the optic pathway (p=0.912). Automated planning significantly reduced planning time compared to manual planning [2.45 min (1.58-3.80 min) vs. 90.0 min (48.0-210.0 min), p<0.001].

Conclusion:

Dose prediction-based automated SRS planning for brain metastases substantially reduces planning time from hours to minutes while maintaining comparable target conformity to manual planning. However, significantly increased low-dose brain volumes (V3, V6) and brainstem maximum dose were observed. Clinical implementation requires careful attention to low-dose brain exposure and brainstem dose optimization, with manual refinement when necessary to ensure treatment safety.