An Automated Patient-Specific Collimator Positioning Tool for Minibeam Radiotherapy Treatments
Abstract
Purpose
X-ray Minibeam Radiotherapy (MBRT) is an emerging technology recently translated to first-in-human treatments. MBRT uses narrow beamlets generated by specialized collimators to produce spatially fractionated dose distributions with high peak-to-valley dose ratios. These ratios improve as the treatment collimator is positioned closer to the patient; however, placement must also maintain target coverage and avoid collisions with the patient surface. This creates a challenging six-degree-of-freedom positioning problem, particularly for patients with irregular surface anatomy or nearby obstructions such as ears or shoulders. This work presents an interactive, automated optimization application for patient-specific collimator placement in MBRT.
Methods
A graphical user interface was developed to extract patient surface geometry from simulation CT DICOM images and initialize collimator placement using the gross tumor volume (GTV) and its spatial relationship to the external contour. Collimator position was optimized in six degrees of freedom using objectives that minimized collimator-to-target distance and maximized the fraction of the target within the collimator projection, subject to a constraint preventing collimator–patient collision. The application provides real-time visual feedback, including three-dimensional renderings, surface distance heatmaps, and beam’s-eye-view target coverage. Upon user acceptance, the optimized collimator geometry is exported as an RT-STRUCT to enable fabrication of a patient-specific 3D-printed collimator holder.
Results
The application rapidly produced clinically acceptable collimator positions while maintaining target coverage. Compared to manual placement, optimized solutions achieved closer collimator proximity, with a mean reduction in collimator-to-surface distance of 1.35 ± 0.83 mm (n = 12), improved consistency, improved target coverage, and substantially reduced planning time (~5 minutes versus ~65 minutes per case).
Conclusion
An interactive, automated collimator optimization tool was developed to streamline MBRT planning, enabling improved positioning accuracy, efficiency, and reproducibility.