Registration of synthetic tomographic projection data sets using cross-correlation

Area of Science:

• Medical Physics
• Radiotherapy Technology
• Image Guided Therapy

Background:

• Accurate patient positioning is critical for effective radiotherapy.
• Current methods like laser alignment require meticulous setup.
• Tomographic registration offers a novel approach using projection data.

Purpose of the Study:

• To develop and evaluate a tomographic registration method for accurate patient positioning.
• To assess the feasibility of computer-mediated refined positioning in radiotherapy.
• To determine the accuracy achievable with this novel registration technique.

Main Methods:

• The tomographic registration process involves manual coarse positioning, tomographic projection set acquisition, and computer-mediated refined positioning.
• Patient alignment is refined by cross-correlating projection sets with planning CTs to determine translation and rotation offsets.
• The algorithm was tested using synthetic data with added Gaussian noise to evaluate stability.

Main Results:

• The developed tomographic registration method achieves patient alignment comparable to standard radiotherapy setups.
• The accuracy requirements for initial patient alignment are relaxed.
• The computer-mediated refined positioning demonstrated stability and achieved alignment accuracy within one projection pixel, even with noise.

Conclusions:

• Tomographic registration provides an accurate and potentially more efficient method for patient positioning in radiotherapy.
• This technique reduces the need for meticulous manual alignment, streamlining the setup process.
• The sub-pixel accuracy achieved demonstrates the clinical viability of tomographic registration for image-guided radiotherapy.