Encrypted login | home

Program Information

Evaluation of a Novel Algorithm for Improved 4DCT Resolution

C Glide-Hurst

C Glide-Hurst1*, P Klahr2 , J Briceno1 , I. J. Chetty1 , (1) Henry Ford Health System, Detroit, MI, (2) Philips Healthcare, Highland Heights, Ohio


MO-FG-204-5 (Monday, July 13, 2015) 4:30 PM - 6:00 PM Room: 204

Purpose: Accurate tumor motion characterization is critical for increasing the therapeutic ratio of radiation therapy. To accommodate the divergent fan-beam geometry of the scanner, the current 4D-CT algorithm utilizes a larger temporal window to ensure that pixel values are valid throughout the entire FOV. To minimize the impact on temporal resolution, a cos² weighting is employed. We propose a novel exponential weighting (“exponential”) 4DCT reconstruction algorithm that has a sharper slope and provides a more optimal temporal resolution.

Methods: A respiratory motion platform translated a lung-mimicking Styrofoam slab with several high and low-contrast inserts 2 cm in the superior-inferior direction. Breathing rates (10-15 bpm) and couch pitch (0.06-0.1 A.U.) were varied to assess interplay between parameters. Multi-slice helical 4DCTs were acquired with 0.5 sec gantry rotation and data were reconstructed with cos² and exponential weighting. Mean and standard deviation were calculated via region of interest analysis. Intensity profiles evaluated object boundaries. Retrospective raw data reconstructions were performed for both 4DCT algorithms for 3 liver and lung cancer patients. Image quality (temporal blurring/sharpness) and subtraction images were compared between reconstructions.

Results: In the phantom, profile analysis revealed that sharper boundaries were obtained with exponential reconstructions at transitioning breathing phases (i.e. mid-inhale or mid-exhale). Reductions in full-width half maximum were ~1 mm in the superior-inferior direction and appreciable sharpness could be observed in difference maps. This reduction also yielded a slight reduction in target volume between reconstruction algorithms. For patient cases, coronal views showed less blurring at object boundaries and local intensity differences near the tumor and diaphragm with exponential weighted reconstruction.

Conclusion: Exponential weighted 4DCT offers potential for improving image sharpness in 4DCT. Comparisons with true object extent using static CTs of varied phantom positions are necessary to confirm these findings. Future work with additional patient cases is warranted.

Funding Support, Disclosures, and Conflict of Interest: Research partially sponsored by Philips Healthcare

Contact Email: