Patient set-up using portal images: 2D/2D image registration using mutual information

Area of Science:

• Medical Physics
• Radiotherapy
• Image Analysis

Background:

• Accurate patient positioning is critical for effective conformal radiation therapy.
• Electronic portal imaging devices (EPIDs) capture images for setup verification, but quantitative interpretation for automated correction is challenging.
• Current methods require robustness and automation for clinical integration across various radiotherapy techniques.

Purpose of the Study:

• To establish the feasibility of 2D/2D image registration for portal/portal and portal/simulator images in radiotherapy.
• To develop and evaluate an automated method for precise patient positioning.
• To ensure the method's applicability to diverse radiotherapy techniques.

Main Methods:

• An automated 2D/2D image registration method was developed, combining calibration and pixel-based algorithms.
• Experiments were conducted using phantoms (with and without a gold standard) and patient data.
• The method was tested with different imaging techniques, including blurred images and small treatment fields.

Main Results:

• The automated 2D/2D registration method proved to be fast, accurate, and robust in phantom studies.
• Performance was maintained even with blurred images and small treatment fields.
• Preliminary patient data indicated the method's potential for clinical application.

Conclusions:

• Mutual information-based 2D/2D image registration is a feasible and effective technique for radiotherapy.
• This automated approach enhances the accuracy and efficiency of patient positioning.
• The method shows promise for routine clinical use in radiation therapy image guidance.