SU-E-J-138: Fast 2-D Fiducial Marker Detection on Sequential MV Projections in Arc Therapy

Area of Science:

• Medical Physics
• Radiation Oncology
• Image Processing

Background:

• Intrafraction motion during arc radiotherapy can significantly impact treatment accuracy for prostate cancer.
• Real-time detection of this motion is crucial for adaptive dose adjustments.
• Current methods may lack the precision or speed required for effective intrafraction motion management.

Purpose of the Study:

• To develop and validate an automated method for detecting intrafraction motion in prostate cancer patients undergoing arc radiotherapy.
• To track the movement of implanted fiducial markers using 2D MV images acquired during treatment.
• To enable precise, real-time adjustments of radiation dose based on detected intrafraction motion.

Main Methods:

• Implantation of four gold fiducial markers in the prostate.
• Acquisition of 2D MV images during 360-degree gantry rotation using a Varian Linac with RapidArc.
• Image preprocessing including edge detection and morphological operations, followed by marker center detection constrained by planning CT data and a 2D correction algorithm.

Main Results:

• The method was validated on 191 projections from four treatment fractions, with manually indicated marker positions serving as ground truth.
• A mean detection error of less than 0.5 mm (standard deviation 0.6 mm) was achieved.
• Execution time was under one second per image, with most errors occurring at moving leaf boundaries.

Conclusions:

• The developed method effectively detects intrafraction motion during arc radiotherapy using only projected MV images.
• This automated approach shows promise for improving the accuracy of radiotherapy for prostate cancer.
• Further research could refine marker visibility assessment, particularly in complex scenarios like moving leaf boundaries.