Neural network-based tissue attenuation correction in dual-band cherenkov imaging for dose distribution verification

Area of Science:

• Medical Physics
• Optical Imaging
• Radiation Oncology

Background:

• Cherenkov imaging offers potential for real-time dose verification in radiation therapy.
• Tissue heterogeneity causes Cherenkov signal attenuation, limiting accurate dose quantification.
• Advanced correction methods are needed to overcome these limitations.

Purpose of the Study:

• To develop and validate an advanced attenuation correction method for Cherenkov imaging.
• To improve the accuracy of Cherenkov-based dose quantification in the presence of tissue heterogeneity.
• To enhance the reliability of Cherenkov imaging for radiation dose assessment.

Main Methods:

• A Neural Network-Based Attenuation Correction Model (NN-BACM) was developed using dual-band Cherenkov images (550 nm and 660 nm).
• Optical features were extracted from dual-band images to train the neural network for predicting attenuation correction factors.
• The model was tested under uniform and non-uniform dose conditions, evaluating dose distribution uniformity and correlation with planned doses.

Main Results:

• The NN-BACM significantly improved dose distribution uniformity, reducing the coefficient of variation from 20.8%-33.8% to 1.2%-1.8% under uniform dose fields.
• In non-uniform fields, the correction method effectively reduced tissue-induced Cherenkov light attenuation.
• Corrected Cherenkov-based dose distributions showed high consistency with actual surface doses.

Conclusions:

• The proposed NN-BACM effectively compensates for Cherenkov signal attenuation caused by tissue heterogeneity.
• This method enhances the accuracy of dose quantification in Cherenkov imaging.
• The findings provide a foundation for precise quantitative dose assessment in clinical applications.