Carotid computed tomography angiography with automated bone suppression: a comparative study between dual energy and bone subtraction techniques

Abstract

Objectives: Computed tomography angiography (CTA) is a well-accepted imaging modality to evaluate the vessels of the head and neck. To improve the assessment of vessels at or within the skull base, different techniques to suppress bone in the final CTA image have been developed. We sought to compare the performance of fully automated bone removal based on dual energy (DE) and image registration (bone subtraction CTA [BSCTA]) for the supraaortic vasculature.

Materials and methods: Twenty-seven patients underwent carotid CTA on a dual source CT system. Two scans were performed, one before and one after contrast agent injection in DE mode (simultaneous acquisition of 80 and 140 kV scans). For each patient, data sets were reconstructed for post processing with DE and BSCTA. Two examiners evaluated overall bone suppression and image quality regarding integrity of the vessel lumen of different vessel segments (n = 19) with a 5-point scale (1 = poor, 5 = excellent), CTA source data served as the reference. Wilcoxon signed rank test was used to determine differences between the techniques. Cohen's kappa was used to determine interobserver agreement.

Results: Both bone suppression techniques could be applied successfully in 25 patients with high overall image quality. Two patients were excluded from the evaluation for technical reasons. Interobserver agreement was excellent (kappa = 0.85). Mean score of all vessel segments for overall bone removal was 4.45 +/- 0.76 for DE and 4.33 +/- 1.02 for BSCTA; DE performed better in the vessel segment of the neck (external carotid artery, common carotid artery, V1 to V3 segment of the vertebral artery, subclavian artery), whereas BSCTA was superior in the cranial vessels (C3-C6 segment of the internal carotid artery, ophthalmic artery). Vessel integrity was superior with BSCTA, the mean score of all vessel segments for overall vessel integrity was 3.97 +/- 1.47 for DE and 4.49 +/- 0.98 for BSCTA, the subclavian artery was most frequently truncated using DE post processing (mean score: 1.44 +/- 1.2).

Conclusions: Both techniques provided bone suppression in a fully automated way. DE provided more complete bone suppression in the neck, but at the cost of inferior vessel integrity, especially at the thoracic inlet. BSCTA showed excellent results for vessel integrity and was superior to DE in most of the vessels in or at the skull base.