Iterative 2D deconvolution of portal imaging radiographs

Area of Science:

• Medical Physics
• Radiotherapy Technology
• Image Processing

Background:

• Portal imaging is crucial for verifying patient positioning in modern radiotherapy (IMRT, IGRT).
• Image blurring in portal images degrades quality, hindering clear visualization of anatomical structures.
• Accurate patient positioning is essential for successful radiotherapy treatment outcomes.

Purpose of the Study:

• To develop and evaluate a 2D iterative deconvolution method for reducing image blurring in portal imaging.
• To identify and quantify the main physical and geometrical contributors to portal image blurring.
• To improve the diagnostic quality of portal images for enhanced radiotherapy accuracy.

Main Methods:

• A 2D iterative deconvolution algorithm was developed to address image blurring.
• Components contributing to blurring (secondary electron transport, EPID pixel size, geometrical penumbra) were measured separately.
• Line-spread kernels were modeled as Lorentz functions, characterized by a width parameter (λ).
• Portal images were deconvolved using experimentally determined λ values (0.5mm for 6 MV, 0.65mm for 15 MV).

Main Results:

• Secondary electron transport, EPID pixel size, and geometrical penumbra were identified as major blurring contributors.
• Photon scattering within the patient was found to be a less significant factor.
• The deconvolution method, using Lorentz function-derived point-spread functions, effectively reduced image blurring.
• Quantification via modulation transfer function demonstrated significant improvement in image sharpness and contrast.

Conclusions:

• The developed 2D iterative deconvolution method successfully enhances portal image quality.
• Improved image sharpness and contrast facilitate clearer visualization of anatomical structures.
• This technique has the potential to improve the accuracy of patient positioning in radiotherapy.