Needle artifact reduction during interventional CT procedures using a silver filter

Abstract

Background: MAR algorithms have not been productized in interventional imaging because they are too time-consuming. Application of a beam hardening filter can mitigate metal artifacts and doesn't increase computational burden. We evaluate the ability to reduce metal artifacts of a 0.5 mm silver (Ag) additional filter in a Multidetector Computed Tomography (MDCT) scanner during CT-guided biopsy procedures.

Methods: A biopsy needle was positioned inside the lung field of an anthropomorphic phantom (Lungman, Kyoto Kagaku, Kyoto, Japan). CT acquisitions were performed with beam energies of 100 kV, 120 kV, 135 kV, and 120 kV with the Ag filter and reconstructed using a filtered back projection algorithm. For each measurement, the CTDIvol was kept constant at 1 mGy. Quantitative profiles placed in three regions of the artifact (needle, needle tip, and trajectory artifacts) were used to obtain metrics (FWHM, FWTM, width at - 100 HU, and absolute error in HU) to evaluate the blooming artifact, artifact width, change in CT number, and artifact range. An image quality analysis was carried out through image noise measurement. A one-way analysis of variance (ANOVA) test was used to find significant differences between the conventional CT beam energies and the Ag filtered 120 kV beam.

Results: The 120 kV-Ag is shown to have the shortest range of artifacts compared to the other beam energies. For needle tip and trajectory artifacts, a significant reduction of - 53.6% (p < 0.001) and - 48.7% (p < 0.001) in the drop of the CT number was found, respectively, in comparison with the reference beam of 120 kV as well as a significant decrease of up to - 34.7% in the artifact width (width at - 100 HU, p < 0.001). Also, a significant reduction in the blooming artifact of - 14.2% (FWHM, p < 0.001) and - 53.3% (FWTM, p < 0.001) was found in the needle artifact. No significant changes (p > 0.05) in image noise between the conventional energies and the 120 kV-Ag were found.

Conclusions: A 0.5 mm Ag additional MDCT filter demonstrated consistent metal artifact reduction generated by the biopsy needle. This reduction may lead to a better depiction of the target and surrounding structures while maintaining image quality.

Keywords: CT artifact; CT filter; Interventional CT.

Fig. 1

Metal artifacts from biopsy needles can obscure targets. In this example, a planning CT is shown on the left, and an intraprocedural CT fluoroscopy image on the right. The target is a 1.8 cm lymph node which is easily visualized on the left image (red arrow) but completely obscured by an artifact on the right. Interventional device artifacts commonly manifest themselves in this way, hindering both visualization of important anatomy that will be traversed by the needle, and localization of the target

Fig. 2

Shown are the positions in the CT image where the profiles were drawn to obtain the values of FWHM, FWTM, and the absolute error in HU used to quantify the artifacts generated from the biopsy needle. FWHM = full width half maximum, and FWTM = full width at one-tenth of maximum (FWTM). The needle was inserted from the back of the right lung until the needle tip reached the region of tissue at chest level

Fig. 3

CT image showing the ROIs positions (yellow rectangles) used to measure the image noise

Fig. 4

Axial CT images of the Lungman phantom with the needle inserted into the lung. The top row shows CT images reconstructed with a soft tissue window (ww/wl of 400/50) and the bottom row with a window for lung (ww/wl of 1700/− 700). Beam energy is indicated from left to right: 100 kV, 120 kV, 135 kV, and 120 kV with Ag additional filter

Fig. 5

Profiles of the artifacts for the beam energies of 100 kV, 120 kV, 135 kV, and 120 kV-Ag filter. Profiles represent the Hounsfield unit change with the distance passing through the artifact. a profile on the needle (i.e., the yellow region in Fig. 2), b profiles in the needle tip artifact (i.e., the green region in Fig. 2), c profiles in the needle trajectory path artifact (i.e., the blue region in Fig. 2), and d profiles in the directions of the needle path on needle trajectory artifact (i.e., the red region in Fig. 2). The shaded regions depict standard deviations in profiles