TH-EF-BRB-09: Real-Time Ultrasound Monitoring with Speckle Tracking in Abdominal Stereotactic Body Radiation Therapy

Abstract

Purpose: Ultrasound is ideal for real-time monitoring with high soft tissue contrast, non-ionization, portability, and cost effectiveness. No studies have investigated clinical feasibility of real-time ultrasound monitoring for abdominal stereotactic body radiation therapy (SBRT) under active breath control (ABC). We are able to monitor target motion using 4D ultrasound and speckle tracking. Methods: An arm-bridge system (ABS) was designed to allow clinician to freely move and lock ultrasound probe. A ceiling mounted infrared camera was calibrated to track probe position using an reflector on the probe. An abdominal ultrasound phantom was secured on a programmable respiratory motion platform to simulate the 20 second expiration and inspiration phases of the ABC with 10 mm inferior translation for 6 cycles. The probe was coupled to the phantom using gel. 4D ultrasound B-mode images were simultaneously acquired at each breath-hold for monitoring. A 3D normalized cross-correlation template matching algorithm was developed to track speckle and feature Point-of-interest (POI) motion at shallow, medium and deep positions in B-mode images. The reproducibility of breath-hold was evaluated by comparing speckle tracking and motion platform position as a function of ROI size. Results: During the repeated respiration movement, the probe position variation was minimal (< 0.5 mm at all time in all directions). For target motion monitoring, ultrasound speckle tracking showed high reproducibility (0.005±0.005 mm AP and 0.01±0.01 mm SI for all ROIs, and 0.03±0.02 mm RL for the shallow and medium ROI). For the deep speckle ROI, the RL error was 0.04±0.04 mm likely due to reduced resolution with depth. Deep ROI showed increased RL tracking accuracy with increased ROI size. Conclusion: Our ABS was able to maintain fixed probe position during the respiration movement with high fidelity. Ultrasound speckle tracking was able to precisely determine the motion over repeated breath-hold for real time monitoring. This work is supported by NIT grant R01CA161613.