State-Aware Non-Contact Respiratory Volume Monitoring for Dibh-Guided Radiotherapy
Abstract
Purpose
Respiratory motion increases setup uncertainty in thoracic and abdominal tumor radiotherapy and elevates radiation exposure to organs at risk (OARs), such as the heart and lungs. Deep inspiration breath-hold (DIBH) can effectively reduce OAR dose; however, its clinical application relies on accurate, real-time, and low-burden monitoring of breath-hold stability and tidal volume. This study proposes a contactless tidal volume estimation system for DIBH-guided radiotherapy.
Methods
A depth camera was used to capture three-dimensional surface motion of the thoracoabdominal region, from which a surface-volume surrogate was constructed. A state-aware, surrogate-guided time-series regression model was developed for tidal volume estimation. The model employs dynamic state gating to distinguish between breath-hold and free-breathing phases, and integrates multi-anchor fusion and calibration strategies to enhance prediction stability. Validation was conducted on 22 volunteers with synchronous acquisition of depth data and signals from an Active Breathing Coordinator (ABC) system as the reference standard. Real-time visualization of respiratory state and tidal volume was implemented using a head-mounted display.
Results
The estimated tidal volume demonstrated a strong correlation with the ABC reference (r = 0.968), with a mean absolute error of 0.0987 mL. The average processing time per batch (360 frames) was 7.24 ms, meeting real-time application requirements.
Conclusion
The proposed method enables accurate and stable tidal volume estimation during DIBH without the need for additional wearable devices, while providing real-time visual feedback. It offers a practical solution for respiratory-gated radiotherapy that balances estimation accuracy with patient comfort.