Click here to


Are you sure ?

Yes, do it No, cancel

Effect of Multi-Slit Collimator Motion On SparseCT Image Quality for Low-Dose CT Examinations

B Chen1*, E Kobler2 , T Allmendinger3 , A Sodickson4 , D Sodickson5 , R Otazo6 , (1) NYU School of Medicine, New York, NY, (2) Graz University of Technology, Graz, ,(3) Siemens Healthineers, Forchheim, ,(4) Brigham and Women's Hospital, Boston, MA, (5) NYU School of Medicine, New York, NY, (6) Memorial Sloan Kettering Cancer Center, New York, NY


(Thursday, 7/18/2019) 7:30 AM - 9:30 AM

Room: 221AB

Purpose: SparseCT is a practical compressed sensing approach for CT dose reduction, which undersamples each view along the row dimension with a multi-slit collimator (MSC). The MSC is mounted between tube and patient and moves along the row direction to change the undersampling pattern along the row dimension for each view. This study aims to investigate the impact of MSC motion on SparseCT image quality.

Methods: A SparseCT prototype was built with the MSC installed on a state-of-art clinical CT scanner. The MSC is a tungsten plate with periodic slits parallel to detector row direction. The slit separation is 3 times wider than the slit width, such that the dose reduction factor is 3. A liver phantom was scanned repeatedly at various MSC locations, each sampling different rows. The MSC was static during each scan, but “dynamic MSC� scans were retrospectively simulated by stitching together projections from different scans. Six MSC motions were tested, including 3 patterns (linear, back-and-forth, and random) and 2 speeds (1 and 5 row(s)/projection). The dynamic MSC scans were reconstructed iteratively using a compressed sensing reconstruction algorithm that enforces 3D sparsity using total variation regularization. Image quality for different motions were compared in terms of PSNR and SSIM.

Results: Increasing MSC motion speed significantly improved PSNR and SSIM while the effect of motion pattern was negligible. Higher motion speeds also markedly reduced undersampling artifacts observed around high attenuation, high frequency objects such as the spine. The best PSNR and SSIM were achieved using a combination of linear motion and a speed of 5 rows/projection.

Conclusion: The motion of the MSC has a significant impact on the performance of SparseCT. Higher motion speed yields more incoherent undersampling artifacts and thus improves reconstruction quality.

Funding Support, Disclosures, and Conflict of Interest: This work was funded by NIBIB (U01 EB018760)


Not Applicable / None Entered.


IM- CT: Development (New technology and techniques)

Contact Email