Scanned particle-beam tracking with beam correction based on predictive volumetric imaging: A simulation study.

Area of Science:

• Particle therapy
• Radiation oncology
• Medical imaging
• Image-guided therapy

Background:

• Real-time tracking irradiation in spot-scanning particle therapy aims to improve treatment time and accuracy for moving targets.
• System processing delays (image acquisition, synthesis, correction assessment, response) currently limit clinical translation, causing discrepancies due to tumor motion.
• Predictive modeling is proposed to compensate for these latency-induced discrepancies, particularly for volumetric image prediction crucial for energy correction.

Purpose of the Study:

• To evaluate the dosimetric effectiveness of particle-beam tracking irradiation utilizing predictive volumetric imaging under varying system latency conditions.

Main Methods:

• A predictive volumetric imaging technique was developed, combining surrogate-driven volumetric image synthesis with surrogate position prediction using a Long Short-Term Memory (LSTM) network.
• A linear regression model was established to derive internal deformation from surrogate displacement for each voxel using 4D CT data.
• Dosimetric simulations were performed on lung, liver, and pancreatic cancer patient data (4DCT, internal marker trajectories) with simulated latencies (133.3, 266.6, 400.0 ms), comparing no-latency tracking, latency without prediction, and latency with prediction.

Main Results:

• Tracking irradiation with prediction achieved doses comparable to the benchmark (no latency).
• Without prediction, CTV dose differences exceeded 5% at all latencies for lung tumors and at 400.0 ms for liver/pancreas tumors.
• With prediction, dose differences for CTVs were generally below 5% (lung) and below 3% (liver/pancreas) even at 400.0 ms latency, with minimal deviations for organs at risk.

Conclusions:

• The proposed tracking irradiation with predictive volumetric imaging demonstrates dosimetric accuracy comparable to the treatment plan across various latencies for lung, liver, and pancreas tumors.
• This technique shows promise as an effective motion management strategy in spot-scanning particle therapy.
• Further development of treatment devices and planning protocols is recommended for clinical implementation to enhance dosimetric accuracy and treatment efficiency.