Related Experiment Video
Updated: Apr 16, 2026
08:25
Radiation Planning Assistant - A Streamlined, Fully Automated Radiotherapy Treatment Planning System
Published on: April 11, 2018
16.2K
Evaluation of mathematical algorithms for automatic patient alignment in radiosurgery
Kenneth M Williams1, Reinhard W Schulte2, Keith E Schubert3
1Loma Linda University, Loma Linda, CA, USA School of Computer Science and Engineering, California State University, San Bernardino, CA, USA kwilliams@llu.edu.
Technology in Cancer Research & Treatment
|March 18, 2015
Summary
Anatomy-based image registration automates patient alignment for intracranial radiosurgery, improving accuracy and efficiency. Three algorithms (phase correlation, mutual information, ECC) showed clinical adequacy, potentially eliminating fiducial markers.
Area of Science:
- Medical Physics
- Radiosurgery Technology
- Image Analysis
Background:
- Patient alignment is critical for accurate intracranial radiosurgery.
- Current methods often rely on implanted fiducial markers and manual alignment, which can be invasive and time-consuming.
Purpose of the Study:
- To evaluate the efficacy of 2D image registration algorithms for automated patient alignment in intracranial radiosurgery.
- To compare the accuracy and speed of these algorithms against the standard manual alignment procedure.
Main Methods:
- Four 2D image registration algorithms were analyzed: phase correlation, mutual information (MI) maximization, enhanced correlation coefficient (ECC) maximization, and iterative closest point (ICP).
- Digitally reconstructed radiographs (DRRs) from CT scans served as reference images.
- Orthogonal digital X-ray images from the treatment room were used as captured images for alignment.
Main Results:
- Phase correlation, MI maximization, and ECC maximization algorithms demonstrated clinically adequate translational accuracy and improved speed compared to manual alignment.
- The ICP algorithm did not yield clinically acceptable results.
- Three of the four tested algorithms showed comparable or superior performance to the standard alignment technique.
Conclusions:
- Automated, anatomy-based 2D image registration algorithms show promise for improving intracranial radiosurgery alignment.
- These techniques may reduce or eliminate the need for invasive fiducial markers.
- Further research into combining algorithms could optimize registration for clinical application.
