Moving Metal Artifact Reduction for Cone-Beam CT (CBCT) Scans of the Thorax Region

Transcription

1 Moving Metal Artifact Reduction for Cone-Beam CT (CBCT) Scans of the Thorax Region Andreas Hahn 1,2, Sebastian Sauppe 1,2, Michael Knaup 1, and Marc Kachelrieß 1,2 1 German Cancer Research Center (DKFZ), Heidelberg, Germany 2 University of Heidelberg, Heidelberg, Germany

2 Metal Artifacts beam hardening 2

3 Metal Artifacts beam hardening + scatter 3

4 Metal Artifacts beam hardening + scatter + directed noise 4

5 Metal Artifacts sharp edges + motion 5

6 Illustration of Conventional Metal Artifact Reduction: Normalized MAR (NMAR) Interpolation Normalized sinogram Interpol. & norm. Normalization Denormalization Input Original sinogram Metal projections Sinogram of tern. im. Corrected sinogram Uncorrected image Metal image Ternary image Corrected image Thresholding Output Meyer, Raupach, Lell, Schmidt, and Kachelrieß, Normalized metal artifact reduction (NMAR) in computed tomography, Med. Phys. 37(10): ,

7 Moving Metal Image-guided radiation therapy (IGRT) CBCT imaging unit mounted on gantry of a LINAC treatment system Slow gantry rotation speed of 6 per second (60 s/360 ) Much slower than clinical CT devices (0.25 s /360 ) Breathing about 10 to 30 rpm (respirations per minute) and thus per scan Linear Accelerator kv Source Detector Projections Motion due to breathing Gantry Rotation Metal subject to respiratory motion 7

8 Problem of Conventional MAR Conventional inpainting-based methods like NMAR do not account for motion. Threshold-based segmentation yields a static metal insert. Metal segmentation in volume Metal mask in projections Reconstructed volume after interpolation 8

9 Problem of Conventional MAR Conventional inpainting-based methods like NMAR do not account for motion. Threshold-based segmentation yields a static metal insert. Metal segmentation in volume Metal mask in projections Reconstructed volume after interpolation 9

10 Accounting for Metal Motion Most MAR methods do not account for motion. Several publications on raw data-based segmentation: Zhang et al.: Reducing metal artifacts in cone-beam CT images by preprocessing projection data, Int J Radiat Oncol Biol Phys 67(3): , March 2007 requires user input Veldkamp et al.: Development and validation of segmentation and interpolation techniques in sinograms for metal artifact suppression in CT, Med Phys 37(2): , February 2010 lacks robustness New approach aims for a combination of image and raw data-based segmentation: Toftegaard et al.: Moving metal artifact reduction in cone-beam CT scans with implanted cylindrical gold markers, Med Phys 41, December 2014 prior knowledge needed (shape of inserts) 10

11 Accounting for Metal Motion Combination of volume and raw data-based metal segmentation: Moving metal artifact reduction (MMAR) Brehm et al.: Moving metal artifact reduction (MMAR): A metal artifact reduction algorithm for flat 280 detector cone-beam CT scans with metal subject to respiratory motion, RSNA Conference Proc., November 2011 Metal segmentation in volume Metal mask in projections Reconstructed volume after interpolation MMAR refined segmentation 11

12 MoCoMAR Aim: Removal of metal artifacts in 3D CBCT volumes. Idea: For NMAR 1 static metal is segmented in a 3D volume. For our new approach (MoCoMAR), NMAR is improved by segmenting the metal in a 4D volume. The 4D volume is reconstructed using the acmoco 2 algorithm. Projections 4D Volume 4D Met. Mask in Volume 3D Met. Mask in Projections acmoco Thresholding Forward projection according to motion bin Corrected Volume NMAR 1: Meyer, Raupach, Lell, Schmidt, and Kachelrieß, Normalized metal artifact reduction (NMAR) in computed tomography, Med. Phys. 37(10): , : Brehm, Paysan, Oelhafen, and Kachelrieß, Artifact-resistant motion estimation with a patient-specific artifact model for motion-compensated cone-beam CT, Med. Phys. 40(10),

13 sagittal axial Results Patient 1 Standard NMAR MMAR MoCoMAR 13

14 sagittal axial Results Patient 1 Standard NMAR MMAR MoCoMAR 14

15 Conclusions MMAR and MoCoMAR outperform NMAR when it comes to moving metal inserts. The proposed method (MoCoMAR) was able to achieve similar results to MMAR. It is, however, computationally more expensive. More patients have to be evaluated to see if there are cases where one of the methods that account for motion outperforms the other. 15

16 Thank You! This presentation will soon be available at Job opportunities through DKFZ s international PhD or Postdoctoral Fellowship programs ( or directly through Marc Kachelriess (marc.kachelriess@dkfz.de). This study was supported by DFG KA 1678/13. Parts of the reconstruction software were provided by RayConStruct GmbH, Nürnberg, Germany. 16