Real-time, upstream, radiotherapy verification using a CMOS Monolithic Active Pixel Sensor System

Abstract

There is a need for independent quality assurance in radiotherapy, such as IMRT and VMAT. This thesis presents a real-time upstream intra-fraction verification system using an array of CMOS Monolithic Active Pixel Sensors (MAPS). The system is designed for two key functions: real-time, high-precision reconstruction of Multi-Leaf Collimator (MLC) positions to verify beam shape, and accurate dosimetry. This approach provides the direct, sub-second beam monitoring required for the next generation dose guided radiotherapy.

A 2 × 2 sensor array demonstrator was constructed from individual 12 × 14 cm2 sensors to monitor 40 × 40 cm2 fields with ≤ 1% beam attenuation. Using a new leaf position reconstruction algorithm that’s suitable for the inter-sensor gap region yielded position resolutions that remain below 500 μm. For dosimetry, a copper grating was used to separate therapeutic photon and contaminant electron signals. A grating-friendly reconstruction algorithm was developed, achieving worst resolutions of 68 . 59 ± 10. 75 μm for a 10 × 10 cm2 field and 109. 5 ± 24. 7 μm for a 40 × 40 cm2 field.

The system’s performance was validated across all clinical beam parameters (6 MV and 10 MV flattened, and 6 MV FFF) and dose rates (15 to 1323 MU/min). In High Full Well (HFW) sensor mode, it maintained a linear response and sub-300 μm resolution, with positional reconstruction independent of beam energy and dose rate. This algorithm, when applied to complex configurations with adjacent leaves, achieved a resolution of 130 . 5 ± 10 . 8 μm without requiring non-independent corrections.

Dosimetric characterisation confirmed the removal of electron contamination and a field size–independent signal-to-dose calibration. The system provides 2-dimensional dose verification, achieving a measurement uncertainty of 0.095% with a time resolution of 0.15 s and a spatial resolution of 100 μm.

Date of Award	30 Sept 2025
Original language	English
Awarding Institution	University of Bristol
Supervisor	Jaap J Velthuis (Supervisor), Lana Beck (Supervisor) & Richard Hugtenburg (Supervisor)