Semi-automated hippocampal avoidance whole-brain radiotherapy planning

Abstract

Background: Hippocampal avoidance whole-brain radiotherapy (HA-WBRT) is designed to spare cognitive function by reducing radiation dose to the hippocampus during the treatment of brain metastases. Current manual planning methods can be time-consuming and may vary in quality, necessitating the development of automated approaches to streamline the process and ensure consistency.

Purpose: To automate hippocampal avoidance whole-brain radiotherapy (HA-WBRT) planning.

Methods: Our algorithm automatically contours organs-at-risk (OARs) and the hippocampal-avoidance brain target. The algorithm generates planning structures from given contours, utilizing preset beam parameters and dose constraints for optimization. If the dose constraints are unmet, "hotspot" contours will be created to improve dosimetry. The algorithm was written with RayStation's scripting feature and was compared with clinically approved manual HA-WBRT plans for 20 retrospective patients using target and OAR dose metrics, with statistical analysis performed using the Student's t-test. In the qualitative review, an experienced radiation oncologist blindly scored both the manual plans and autoplans for qualitative review. Lastly, IMRT QA was performed to determine the plans' deliverability.

Results: The autoplans demonstrated a better target coverage with a more uniform dose. With a prescription dose of 3000 cGy, autoplans achieved higher D_95% (3026 cGy vs. 2998 cGy, p = 0.02) and lower D_max (3337 cGy vs. 3533 cGy, p < 0.01) for the target. The maximum OAR doses were substantially lower in the eyes of autoplans (1727 cGy vs. 2176 cGy, p < 0.01), while the other OARs had similar maximum doses to those of the manual plans. The autoplans met all of the in-house dose constraints, and the minimum dose to the hippocampus was reduced by 5% compared to the manual plans; the average MU was 1376 ± 329 MU for the manual plans and 1141 ± 64 MU for the autoplans. Autoplan generation took an average of 100.2 ± 16.3 minutes (range 62.9-127.9 min). In the qualitative review, the average scores were 4.9 ± 0.4 for the autoplans and 3.4 ± 1.0 for the manual plans. The gamma criteria results for IMRT QA were 96.4 ± 2.1% for the autoplans and 91.6 ± 5.3% for the manual plans.

Conclusions: Our rule-based autoplanning algorithm produces high-quality plans that are comparable to those of manual planning, demonstrating autoplanning's potential to reduce HA-WBRT planning time while ensuring consistent plan quality.

Keywords: autocontouring; autoplanning; hippocampal avoidance WBRT; whole brain radiotherapy.

FIGURE 1

Workflow of the HA‐WBRT autoplanning algorithm. The planning structures and beam configurations were determined, and then initial optimization was performed. Dose goals were evaluated on the final plan. If the dose goals were not met, the dose constraints on the hotspots were added and the dose constraints on normal structures were adjusted for further optimization.

FIGURE 2

Boxplots for target and maximum OAR doses. The maximum, D _2%, D _95%, and D _98% doses for the targets, along with the maximum doses for the OARs, were plotted for both manual plans and autoplans.

FIGURE 3

Dose distributions and DVH for autoplans and manual plans. An example of the dose distribution of autoplans and manual plans, alongside DVH curves, for the same patient. The autoplans exhibited a more uniform dose distribution, with fewer hotspots (indicated by light purple) within the target area. In the DVH, the autoplans (represented by solid lines) and manual plans (represented by dashed lines) show similar dose distributions for OARs, while the autoplans demonstrate improved target coverage. Similarly, the bottom figure illustrates the mean DVH for each organ for both the autoplans (solid lines) and manual plans (dashed lines), demonstrating improved target coverage and lower OAR doses.