Three-dimensional seed reconstruction from an incomplete data set for prostate brachytherapy.

Area of Science:

• Medical Physics
• Radiology
• Surgical Technology

Background:

• Accurate 3D seed localization is crucial for intra-operative dosimetry in prostate brachytherapy.
• Current methods rely on multiple X-ray projections, often failing due to clustered or overlapping seeds.
• Computed tomography (CT) is not feasible for intra-operative use.

Purpose of the Study:

• To develop a novel algorithm for robust seed reconstruction from incomplete X-ray projection data.
• To overcome limitations of existing algorithms that require complete seed identification across all projections.
• To improve the accuracy and reliability of intra-operative seed localization in prostate brachytherapy.

Main Methods:

• Developed a new algorithm utilizing epipolar imaging geometry and pseudo-matching.
• The algorithm requires only one complete projection or can handle incomplete data across all three projections.
• Successfully applied the method to clinical cases with overlapping seeds.

Main Results:

• The algorithm successfully reconstructs seeds even with incomplete datasets.
• Achieved 100% reconstruction of implanted seeds in certain scenarios, depending on undetected seed distribution.
• Demonstrated effectiveness in clinical cases with challenging seed overlap.

Conclusions:

• The developed algorithm significantly enhances seed reconstruction accuracy in prostate brachytherapy.
• It offers a more robust solution for intra-operative seed localization, especially when dealing with overlapping seeds.
• This advancement can lead to improved dosimetry and treatment outcomes.