Current techniques for reducing metal streak artifacts in CT images have not achieved widespread clinical use, and they introduce new artifacts. We present a a simple new method, called the Metal Deletion Technique (MDT), and compare it to existing methods.

Simulated projection data was calculated for a phantom containing water, soft tissue, bone, and iron. Experimental projection data was obtained for 11 consecutively identified patients with metal streak artifact scanned at Stanford Hospital’s Siemens Sensation 64 CT scanner. Each scan was reconstructed using filtered backprojection (FBP), linear interpolation (LI), selective algebraic reconstruction technique (SART), and MDT. The MDT algorithm works as follows. First, an adaptive filter is applied to the experimental projection data that effectively expands the detectors in regions with low photon counts. Next, the initial image is constructed using a combination of FBP and LI. Finally, metal pixels are erased, and filtered backprojection is iterated 4 times. On each iteration, rays that pass through metal are replaced with forward projected values from the previous iteration.

The simulations showed that MDT reduces artifacts due to photon counting noise, beam hardening, and motion, while avoiding new artifacts. MDT had the lowest root-mean-square error: 76% less than FBP, 42% less than LI, and 17% less than SART. For the experimental scans, a blinded comparison showed that MDT had the best image quality 100% of the time (95% confidence interval: 78 to 100%). No manual steps or adjustment of parameters was necessary. In two of the scans, MDT changed the radiologic diagnosis. In one case, peri-rectal lymphadenopathy obscured by hip replacement artifacts were identified in a patient with a history of rectal cancer. In the second case, a pacemaker lead that appeared to perforate through the right ventricle was seen to be non-perforated.

MDT reduces metal streak artifacts and produces better quality images than FBP, LI, or SART, for a wide range of scans.

MDT is a promising new technique for reducing metal streak artifacts in CT scans. It is the first metal artifact reduction technique that has been shown to change the radiologic diagnosis.