Using synthetic CT for partial brain radiation therapy: Impact on image guidance

Area of Science:

• Medical Physics
• Radiation Oncology
• Medical Imaging

Background:

• Advancements in synthetic computed tomography (synCT) from magnetic resonance (MR) imaging enable MRI-only treatment planning for brain cancers.
• The performance of synCT for image-guided radiation therapy (IGRT) requires thorough evaluation.

Purpose of the Study:

• To compare the geometric equivalence of digitally reconstructed radiographs (DRRs) generated from synCT and conventional CT scans.
• To quantify the performance of synCT for partial brain IGRT.

Main Methods:

• Generated synCTs by combining multiple MR sequences (ultra-short echo time, T1, T2, FLAIR) using voxel-based weighted summation.
• Compared DRRs from synCT and CT using overlap index, Dice similarity coefficient, and Jaccard index.
• Assessed 2D-2D rigid registrations for 1490 fractions to quantify shifts and differences between CT and synCT.

Main Results:

• DRR agreements (overlap index, Dice, Jaccard) exceeded 0.95, indicating robust geometric equivalence.
• Normalized mutual information results were equivalent in 75% of evaluated quadrants.
• Population average absolute shifts were clinically insignificant (<1 mm), though synCT showed sensitivity in postsurgical cases.

Conclusions:

• The geometric accuracy of DRRs derived from synCT is robust for brain cancer IGRT.
• Observed statistical differences between CT and synCT registrations were not clinically significant.
• Further research is needed to optimize synCT generation for postsurgical brain cancer patients.