2240 - Reduction of Motion Artifacts in Four-Dimensional Computed Tomography Using a Novel Reconstruction Algorithm Based on Deformable Image Registration

Purpose/Objective(s): Motion artifacts in four-dimensional computed tomography (4DCT) due to patient breathing irregularities can adversely impact cancer diagnosis, treatment planning, and image-guided radiation therapy. A novel 4DCT reconstruction algorithm, ZeeFree RT, was developed to mitigate these artifacts by reconstructing a separate stack of images for each breathing cycle, followed by deformable image registration utilizing overlapping data from neighboring respiratory cycles. This study evaluates ZeeFree RT against standard reconstruction methods to assess its effectiveness in reducing motion artifacts and improving anatomical accuracy.

Materials/Methods: 4DCT data from 24 lung cancer patients, with half of the scans acquired on a SOMATOM Go.Open Pro scanner and the other half on a NAEOTOM Alpha photon counting CT were analyzed. Sixteen cases were randomly selected, while eight were chosen for their significant breathing irregularities. Each case was reconstructed using four configurations: (1) Standard reconstruction with amplitude-based binning (Standard-AB) (2) Standard reconstruction with phase-based binning (Standard-PB) (3) ZeeFree reconstruction with amplitude-based binning (ZeeFree-AB) (4) ZeeFree reconstruction with phase-based binning (ZeeFree-PB). A thoracic radiation oncologist with greater than 5 years experience and two medical physicists with more than 5 years experience and expertise in 4D scans reviewed and rated the images in a randomized, blinded fashion. Image quality was assessed using: (1) A 5-point Likert scale to quantify stack alignment artifacts, where 1 = severe artifacts – ‘requiring rescan or clinically useable with significant limitations’, 2 = moderate artifacts (>10 mm displacement) – ‘clinically useable with limitations’, 3 = moderate artifacts (>5 mm displacement) – ‘clinically useable without significant limitations’, 4 = minor artifacts – ‘clinically useable without limitations’, and 5 = no detectable artifacts, and (2) A binary (yes/no) rating for anatomical correctness in all respiratory phases.

Results: For both the Likert scale and anatomical correctness assessments, amplitude-based binning (AB) images consistently outperformed phase-based binning (PB) images. ZeeFree RT significantly improved image quality compared to standard reconstruction (p <0.05), reducing anatomically incorrect cases by over 50%. Additionally, the Likert scale ratings indicated that ZeeFree RT often achieved at least one-point higher ratings than standard reconstruction, with no cases where ZeeFree RT received a rating more than one point lower than standard.

Conclusion: ZeeFree RT is a dedicated 4DCT reconstruction technique that mitigates motion artifacts without requiring hardware upgrades. It significantly reduces 4DCT artifacts and maintains or enhances anatomical accuracy compared to standard reconstruction methods, making it a valuable tool for motion assessment in radiation therapy.