IWORID 2019 The 21 st International Workshop on

IWORID 2019 The 21 st International Workshop on Radiation Imaging Detectors 07 -12 July 2019 in Kolympari, Greece Image accuracy improvement of interior region of interest (ROI) reconstruction using field modulation CT acquisition Dohyeon Kim*, Hee-Joung Kim*, † Department of Radiation Convergence Engineering, College of Health Science, Yonsei University, Wonju, Korea*, Department of Radiological Science, College of Health Science, Yonsei University, Wonju, Korea†, (Yonsei University) (Medical Physics & Imaging Lab. )

Introduction IWORID 2019 • High risk of radiation dose : < Australian Radiation Protection and Nuclear Safety Agency (ARPANSA), Typical values of effective dose for various medical x-rays> • CT requires high x-ray dose as multiple projection images are to be obtained and processed. • Most of high effective dose is related with CT applications in various medical x-rays.

Introduction IWORID 2019 • Motivation of ROI imaging : • In case of the chronic sinusitis or dental implant, we often focus on sinus and teeth area. • CT imaging is needed for chronic sinusitis surgery or septum diagnosis. • CT imaging is needed for planning dental implant surgery. < Sinus CT image >

Introduction IWORID 2019 • Previous work for ROI imaging : < ROI imaging study > < Study of external information restoration > • Do not perform sampling to maintain the quality of ROI internal information. • The x-ray dose to the patient can be reduced by ROI imaging technique. • Existing studies can not restore anatomical information outside the ROI. • To solve this problem, we present field modulation CT acquisition.

Materials & Methods IWORID 2019 • Schematic diagram of acquisition model : ü ROI imaging acquisition : ü Field modulation acquisition (D. Kim et al, RPC, 2019): Total number of projections : S = n + m Configuration of ROI imaging projections : S×Truncated projections Configuration of ROI imaging projections : (S-n)×Truncated projections + n×Nontruncated projections § D. Kim et al, Radiation Physics and Chemistry, 2019 § D. Kim et al, Journal of Applied Clinical Medical Physics, 2018

Materials & Methods IWORID 2019 • Information of actual acquisition parameter (GERI TVX-IL 1500 H) : Image acquisition set-up table <Nontruncated projection acquisition (full scan mode)> <Truncated projection acquisition (ROI scan mode)> ROI setting < Ray. Safe dosimeter> < RSD Anthropomorphic Head Phantom > SDD / SOD 1178 mm / 273 mm Eff. Detector size 1016× 576 Pixel size 0. 2 mm # of proj (angle interval) 720 (0. 5 °) Recon. image size 1016× 1016 Recon. pixel size 0. 1 mm Reconstruction algorithm FBP (Filtered Backprojection) X-ray condition 90 k. V, 0. 5 m. A

Materials & Methods IWORID 2019 • Information of actual acquisition parameter (GERI TVX-IL 1500 H) : Image acquisition set-up table <Nontruncated projection acquisition (full scan mode)> (a) (b) <Truncated projection acquisition (ROI scan mode)> (c) Obtained sinogram in actual CT system: (a) Full scan sinogram (720 non-truncated) (b) ROI scan sinogram (720 truncated) (c) Field modulation scan sinogram (72 non-truncated + 648 truncated) SDD / SOD 1178 mm / 273 mm Eff. Detector size 1016× 576 Pixel size 0. 2 mm # of proj (angle interval) 720 (0. 5 °) Recon. image size 1016× 1016 Recon. pixel size 0. 1 mm Reconstruction algorithm FBP (Filtered Backprojection) X-ray condition 90 k. V, 0. 5 m. A

Materials & Methods • Flowchart of proposed image restoration method : < Flowchart of the image restoration method based on POCS algorithm > § D. Kim et al, Radiation Physics and Chemistry, 2019 IWORID 2019

Materials & Methods IWORID 2019 • POCS reconstruction algorithm : TV minimization Lower-total variation images ü Flowchart of POCS reconstruction algorithm: Real object Total-variation minimization processing Sinogram (real acquisition data) Simultaneous Algebraic Reconstruction Technique (SART) Difference Back projection Projection data Ratio Sinogram Tomographic image Iteration Virtual object Sinogram Projection data § Sidky and Pan et al, Physics in Medicine & Biology, 2008 Projection Update

Materials & Methods IWORID 2019 • Scout view reconstruction method : <Flowchart of scout view reconstruction method> • Scout view does not have a large number of projection data (less than 5 -6 views) • Scout view may be an image taken at another scan time § Kriti Sen Sharma et al. , Phys. Med. Biol, 2013 6 views

Materials & Methods IWORID 2019 • Quantitative evaluation : Image quality and dose evaluation Bright-band reduction evaluation The structural similarity (SSIM) index SSIM (x, y) = [ l(x, y) ]⋅[ c(x, y) ]⋅[ s(x, y) ] Contrast to noise ratio (CNR) and Figure of merit (FOM) ROI (obj) ROI (bkg) • • where μx, μy, σx, σy, and σxy are the local means, standard deviations, and crosscovariance for images x, y. If α = β = γ = 1 (the default), and C 3 = C 2/2 (default selection of C 3). ROI (obj ) ROI (bkg)

Materials & Methods IWORID 2019 • Dose area product (DAP) value calculation : 15. 24 cm 10. 16 cm < Full view scan dosimetry> 720 FULL : 720 ROI : 20. 32 cm 4. 23 cm 648 ROI + 72 FULL : 714 ROI + 6 FULL :

Results & Discussions IWORID 2019 • Reconstructed images : Interior view (with SSIM values) (a) (c) (b) Ref (f) 0. 23 Ref 0. 34 0. 84 0. 79 (h) (g) (e) (d) (j) (i) 0. 73 0. 91 0. 82 0. 94

Results & Discussions IWORID 2019 • Reconstructed images : Global view (Upper row : sinus, Lower row : dental) Reference ROI scan w/ extrapolation Scout view reconstruction (6 views) Proposed method (72 views)

Results & Discussions IWORID 2019 • Results of FOM and CNR values (interior view) • Field modulation images (using 72 views) showed the best performance in terms of CNR value. • Field modulation images (using 6 views) showed the best performance in terms of FOM compared to other conventional ROI techniques.

Conclusion IWORID 2019 ü We proposed a new ROI-based acquisition method to improve the image quality while reducing the dose. üProposed field modulation acquisition method was applied to actual CT prototype imaging system. ü We also proposed image restoration and image processing method for field modulation tomographic imaging. ü In terms of FOM value evaluation, the proposed method shows superior performance compared to conventional ROI methods.

Medical Physics & Imaging Laboratory Thanks for your attention Presenter’s e-mail address : kimdoo 202@yonsei. ac. kr