SU-E-J-87: Building Deformation Error Histogram and Quality Assurance of Deformable Image Registration

Area of Science:

• Medical Imaging
• Image Registration
• Quality Assurance

Background:

• Deformable Image Registration (DIR) systems are crucial for medical imaging but require robust quality assurance (QA) to quantify systematic errors.
• Traditional landmark-based approaches for DIR QA have limitations, only identifying errors at specific visible points.

Purpose of the Study:

• To develop a quantitative method for assessing the systematic error of Deformable Image Registration (DIR) systems.
• To establish a standardized Quality Assurance (QA) procedure for DIR systems.

Main Methods:

• Adapted a Deformation Vector Map (DVM) comparison approach to overcome landmark limitations.
• Generated a ground truth DVM and corresponding image set (R') to quantify DIR systematic error.
• Developed the Deformation Error Histogram (DEH) for quantitative analysis of registration errors, tested on CT to CBCT registration.

Main Results:

• Identified higher errors in skin regions interfacing with air and mobile joints like shoulders.
• Quantified average errors for regions of interest: CTV (0.4mm), Brain stem (1.4mm), Shoulders (1.6mm), and Normal tissues (0.7mm).

Conclusions:

• Successfully developed the Deformation Error Histogram (DEH) approach to quantify DVM uncertainty.
• The DEH and available datasets can serve as a tool for establishing DIR system QA guidelines.
• This method allows users to define acceptable systematic error thresholds for DIR systems.