Modeling dose distributions from portal dose images using the convolution/superposition method

Area of Science:

• Medical Physics
• Radiation Oncology
• Image Reconstruction

Background:

• Post-treatment dose verification is crucial for ensuring accurate radiotherapy delivery.
• Exit dose measurements using portal imaging offer a way to assess the dose delivered to a patient.
• Current methods may require knowledge of the incident radiation fluence.

Purpose of the Study:

• To develop and validate an iterative convolution/superposition algorithm for reconstructing patient dose distributions from exit dose measurements.
• To assess the accuracy and convergence of the proposed method.
• To determine if dose reconstruction is possible without prior knowledge of the incident fluence.

Main Methods:

• An iterative convolution/superposition algorithm was developed using an extended phantom incorporating patient CT data and an electronic portal imaging device (EPID).
• The algorithm reconstructs dose distributions by iteratively predicting and refining the primary energy fluence (PEF) from measured portal dose images.
• Dose reconstruction was performed by back-projecting the converged PEF and convolving with a dose deposition kernel.

Main Results:

• The algorithm successfully reconstructed dose distributions in phantom studies using 6 MV photon beams.
• Reconstructed dose volumes agreed within 3% of forward computation dose volumes, demonstrating accuracy.
• The method demonstrated accurate dose reconstruction without requiring prior knowledge of the incident fluence.

Conclusions:

• The iterative convolution/superposition method provides an accurate and robust approach for post-treatment dose verification using exit dose measurements.
• This technique enhances radiotherapy quality assurance by inferring delivered dose distributions directly from portal images.
• The ability to reconstruct dose without incident fluence information offers a significant advantage in clinical practice.