Volumetric imaging during head and neck radiation therapy using a Kalman filter tracking approach

Area of Science:

• Medical Physics
• Radiotherapy
• Image-guided therapy

Background:

• Volumetric imaging during treatment delivery is crucial for head and neck (H&N) cancer radiotherapy.
• Motion management strategies are needed to improve geometric accuracy and protect organs at risk (OARs).

Purpose of the Study:

• To develop and evaluate a Kalman filter (KF) tracking approach for real-time 3D H&N anatomy imaging using 2D x-ray projections.
• To assess the accuracy of KF tracking for tumors and organs at risk (OARs) compared to no tracking.

Main Methods:

• A KF approach was developed to generate 3D images from 2D x-ray projections in real-time.
• The KF incorporated 3D-to-2D projection geometry and deformable vector fields (DVFs).
• A simulation study used H&N patient data, comparing KF tracking to standard care (no tracking) using centroid error, Hausdorff distance, and Dice coefficient.

Main Results:

• KF tracking significantly reduced mean tumor centroid error across all axes compared to no tracking.
• The 95th percentile Hausdorff distance for tumor contour was smaller with KF tracking.
• KF tracking achieved higher Dice coefficients for tumors and OARs (e.g., brainstem, spinal cord) than no tracking.

Conclusions:

• The developed KF tracking method enables accurate real-time volumetric imaging of H&N anatomy during treatment delivery.
• The findings suggest potential clinical applicability for improving geometric accuracy and OAR sparing in H&N radiotherapy.