Estimation of three-dimensional intrinsic dosimetric uncertainties resulting from using deformable image registration...

Area of Science:

• Medical Physics
• Radiotherapy
• Image Analysis

Background:

• Deformable image registration (DIR) is crucial for dose mapping in radiation therapy.
• Intrinsic uncertainties in DIR can significantly impact the precision of calculated dose distributions.
• Accurate assessment of dose uncertainty is essential for patient safety and treatment efficacy.

Purpose of the Study:

• To present a general procedural framework for assessing point-by-point precision in mapped dose.
• To quantify the dose uncertainty arising from the intrinsic uncertainty of deformable image registration (DIR).
• To evaluate this framework for an arbitrary patient, demonstrated using a 4D lung CT example.

Main Methods:

• A three-step process was developed to obtain dose uncertainty.
• Step 1: Iterative DIR procedure to obtain clusters of points for each voxel.
• Step 2: Quantify spatial uncertainty using dispersion of points (Method A: 1D, Method B: 3D).
• Step 3: Combine spatial uncertainty with mapped dose to compute point-by-point dose standard deviation.

Main Results:

• Demonstration on a 4D lung CT showed standard deviation of inconsistency vectors up to 9.2 mm (mean sigma 1.3 mm).
• Maximum estimated dose uncertainties were 29.65 Gy (Method A) and 21.81 Gy (Method B).
• Significant volumes (602 cm³ for A, 1422 cm³ for B) exhibited dose uncertainties exceeding 10.00 Gy.

Conclusions:

• The proposed procedure is a valuable tool for evaluating mapped dose precision due to intrinsic DIR uncertainty.
• The framework is flexible and adaptable to different intrinsic error models.
• This method enhances the reliability of dose calculations in image-guided radiation therapy.