Program Information

Optimizing Four Dimensional Cone Beam Computed Tomography Projection Allocation to Respiratory Bins

R OBrien*, C Shieh , J Kipritidis , P Keall , Radiation Physics Laboratory, Sydney Medical School, The University of Sydney, Sydney, SW, Australia

Presentations

TH-E-17A-5 Thursday 1:00PM - 2:50PM Room: 17A

Purpose: Four dimensional cone beam computed tomography (4DCBCT) is an emerging image guidance strategy but it can suffer from poor image quality. To avoid repeating scans it is beneficial to make the best use of the imaging data obtained. For conventional 4DCBCT the location and size of respiratory bins is fixed and projections are allocated to the respiratory bin within which it falls. Strictly adhering to this rule is unnecessary and can compromise image quality. In this study we optimize the size and location of respiratory bins and allow projections to be sourced from adjacent phases of the respiratory cycle.

Methods: A mathematical optimization framework using mixed integer quadratic programming has been developed that determines when to source projections from adjacent respiratory bins and optimizes the size and location of the bins. The method, which we will call projection sharing, runs in under 2 seconds of CPU time. Five 4DCBCT datasets of stage III-IV lung cancer patients were used to test the algorithm. The standard deviation of the angular separation between projections (SD-A) and the standard deviation in the volume of the reconstructed fiducial gold coil (SD-V) were used as proxies to measure streaking artefacts and motion blur respectively.

Results: The SD-A using displacement binning and projection sharing was 30%-50% smaller than conventional phase based binning and 59%-76% smaller than conventional displacement binning indicating more uniformly spaced projections and fewer streaking artefacts. The SD-V was 20-90% smaller when using projection sharing than using conventional phase based binning suggesting more uniform marker segmentation and less motion blur.

Conclusion: Image quality was visibly and significantly improved with projection sharing. Projection sharing does not require any modifications to existing hardware and offers a more robust replacement to phase based binning, or, an option if phase based reconstruction is not of a suitable quality.

Funding Support, Disclosures, and Conflict of Interest: This project is supported by an NHMRC Australia Fellowship and NHMRC project grant number 1034060

Contact Email: