Calculations of dose distributions using a neural network model

Area of Science:

• Medical Physics
• Radiotherapy
• Computational Biology

Background:

• External beam radiotherapy aims to treat tumors while sparing healthy tissues.
• Accurate dosimetric planning is crucial for optimizing dose distribution in radiotherapy.
• Current dosimetric planning methods involve a trade-off between calculation speed and precision.

Purpose of the Study:

• To evaluate the efficacy of neural networks in radiotherapy dosimetric calculations.
• To compare the speed and accuracy of neural networks against traditional methods like Monte Carlo simulations.
• To determine if neural networks can provide a balance between precision and speed for clinical treatment planning.

Main Methods:

• Utilized neural networks for dosimetric calculations in radiotherapy.
• Compared the performance of neural networks with Monte Carlo-based codes (e.g., BEAM).
• Assessed the time efficiency and accuracy of the neural network approach for dose distribution mapping.

Main Results:

• Neural network calculations provide rapid results for dosimetric planning.
• The neural network approach achieves high accuracy, with errors typically below 2% for 2D dosimetric maps.
• Monte Carlo methods, while precise, are computationally intensive and time-consuming, requiring hours for single calculations.

Conclusions:

• Neural networks present a promising alternative for clinical radiotherapy treatment planning.
• This approach overcomes the speed limitations of Monte Carlo methods while maintaining high accuracy.
• The adoption of neural networks can significantly enhance the efficiency and effectiveness of radiotherapy planning.