Is AI the way forward for reducing metal artifacts in CT? Development of a generic deep learning-based method and initial evaluation in patients with sacroiliac joint implants. Issue 163 (June 2023)

• Record Type: Journal Article
• Authors: Selles, Mark; Slotman, Derk J.; van Osch, Jochen A.C.; Nijholt, Ingrid M.; Wellenberg, Ruud.H.H.; Maas, Mario; Boomsma, Martijn. F.
• Is Part Of: European journal of radiology. Issue 163(2023)
• Publication Date: 2023-06
• Subjects: CT -- Metal artifacts -- Deep learning -- Orthopedic implants -- Sacroiliac joint fusion; CT computed tomography -- dl-MAR deep learning-based metal artifact reduction -- O-MAR orthopedic metal artifact reduction -- SI sacroiliac -- ROI region of interest -- monoE mono-energetic -- DECT dual energy computed tomograpy -- ResNet residual neural network -- FBP filtered back projection -- ResUNet deep residual U-NET -- HU Hounsfield units -- NMAR normalized metal artifact reduction; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2023.110844
• Languages: English
• ISSNs: 0720-048X

Abstract

Highlights: A deep learning based metal artifact reduction algorithm (dl -MAR) was developed by simulation of metal implants on CT scans. O-MAR and dl -MAR corrected images showed significant reduction of metal artifacts in comparison to uncorrected CT-images after SI joint fusion. dl -MAR showed a significantly stronger reduction of metal artifacts than O-MAR. Abstract: Purpose: To develop a deep learning-based metal artifact reduction technique (dl -MAR) and quantitatively compare metal artifacts on dl -MAR-corrected CT-images, orthopedic metal artifact reduction (O-MAR)-corrected CT-images and uncorrected CT-images after sacroiliac (SI) joint fusion. Methods: dl -MAR was trained on CT-images with simulated metal artifacts. Pre-surgery CT-images and uncorrected, O-MAR-corrected and dl -MAR-corrected post-surgery CT-images of twenty-five patients undergoing SI joint fusion were retrospectively obtained. Image registration was applied to align pre-surgery with post-surgery CT-images within each patient, allowing placement of regions of interest (ROIs) on the same anatomical locations. Six ROIs were placed on the metal implant and the contralateral side in bone lateral of the SI joint, the gluteus medius muscle and the iliacus muscle. Metal artifacts were quantified as the difference in Hounsfield units (HU) between pre- and post-surgery CT-values within the ROIs on the uncorrected, O-MAR-corrected and dl -MAR-corrected images. Noise was quantified as standard deviation in HU within the ROIs. Metal artifacts and noise in the post-surgery CT-images were compared using linear multilevel regression models. Results: Metal artifacts were significantly reduced by O-MAR and dl -MAR in bone (p < 0.001), contralateral bone (O-MAR: p = 0.009; dl -MAR: p < 0.001), gluteus medius (p < 0.001), contralateral gluteus medius (p < 0.001), iliacus (p < 0.001) and contralateral iliacus (O-MAR: p = 0.024; dl -MAR: p < 0.001) compared to uncorrected images. Images corrected with dl -MAR resulted in stronger artifact reduction than images corrected with O-MAR in contralateral bone (p < 0.001), gluteus medius (p = 0.006), contralateral gluteus medius (p < 0.001), iliacus (p = 0.017), and contralateral iliacus (p < 0.001). Noise was reduced by O-MAR in bone (p = 0.009) and gluteus medius (p < 0.001) while noise was reduced by dl -MAR in all ROIs (p < 0.001) in comparison to uncorrected images. Conclusion: dl -MAR showed superior metal artifact reduction compared to O-MAR in CT-images with SI joint fusion implants.