Non-contrast photon counting computed tomography of the head: optimized modeling of off-focal radiation to reduce calvarium-related tissue inhomogeneity

TL;DR

A new CT software reduces brain tissue inhomogeneity near the skull in non-contrast photon counting CT scans, improving image quality.

Contribution

A new software approach optimizes off-focal radiation modeling to reduce artifacts in head CT scans.

Findings

The new software significantly reduced signal differences between gray and white matter near the calvarium (p < 0.001).

Qualitative analysis showed improved gray-white differentiation in virtual monoenergetic images at 65 keV (p < 0.001).

Mathematical correction of off-focal radiation effectively reduced cortical brain tissue inhomogeneity.

Abstract

Photon counting CT (PCCT) is a promising technique for neuroradiological CT examinations. In initial studies on non-contrast PCCT of the head (NCCT), however, artifacts close to the calvarium were noticed, which lead to an inhomogeneous representation of the brain tissue. In this study, a new software for image reconstruction to reduce artifacts is evaluated.

In the new CT software developed by the manufacturer, off-focal radiation was remodeled and is mathematically corrected in the NCCT in data processing during image formation. For the evaluation, 60 patients with an NCCT in the currently used software and 44 patients in the new software were included retrospectively. A detailed quantitative analysis using multiple regions of interest and a qualitative analysis with a reading by experienced radiologists was performed to evaluate image quality and tissue homogeneity below the calvarium.

The new software reduced the inhomogeneity of the cortical brain tissue near the calvarium. As a quantitative measure, there is a clear reduction of the signal difference of the gray and white matter at different distances from the calvarium (p < 0.001). In the qualitative analysis, the inhomogeneity of the brain tissue was reduced, and the gray-white differentiation improved (p < 0.001) in the clinically used virtual monoenergetic image at 65 keV.

Optimized modelling and mathematical correction of the off-focal radiation in the new software led to an effective reduction of the inhomogeneity of the cortical brain tissue and thus improved image quality.

Figures