Purpose: To demonstrate that a fast and low‐dose 4DCBCT can be used to the lung ventilation of small animals for treatment assessment. As 4DCBCT scan goes faster, fewer projections are acquired and under‐sampling‐induced streaking artifacts will be rampant if the conventional reconstruction algorithm is used. The enabling method to be proposed is the PICCS algorithm, which is capable of reconstructing streak free images from vastly under‐sampled projections by using a high SNR prior image. Methods: A 20‐minute 4DCBCT scan of a mouse was performed and 7200 projections were acquired. By under‐sampling these projections, faster 4DCBCT scans with acquisition times equivalent of 4, 2, 1.4, and 1 minute are simulated. Both FBP and PICCS were used for image reconstructing these faster scans and these images were compared with the original slow scan for performance evaluation. A non‐linear deformable registration algorithm was applied to the reconstructed dynamic images to generate lung ventilation maps. The fidelity of the generated lung ventilation maps was used as a metric for quantitative evaluation to determine the optimal scan time. Results: The qualitative evaluation shows unnoticeable differences between PICCS images from faster scans and FBP images from the original scan. The quantitative evaluation is ongoing to demonstrate the equivalence of PICCS images from faster scans and FBP images from the original slow scan, in terms of the accuracy of the extracted lung ventilation maps. Conclusion: The proposed method shows promises to accelerate small animal 4DCBCT scans by an order of magnitude (data acquisition time down to about 2 minutes) without impacting the faithfulness of the lung ventilation map. Significant radiation dose reduction (∼10) is also expected.
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging