Value of automatic bone subtraction in cranial CT angiography: comparison of bone-subtracted vs. standard CT angiography in 100 patients

Non-contrast-enhanced cranial computed tomography (NECT) and CT angiography (CTA) are the most frequently used modalities in the triage of patients with acute ischemic and hemorrhagic stroke. CTA bone removal can improve the delineation of vasculature closely adjacent to bony structures, which is sometimes limited in standard CTA. The aim of this study was the evaluation of the clinical benefit of bone subtraction (BS) regarding delineation of cerebral vasculature, reading time and depiction of vascular pathologies compared to standard CTA without BS. A total of 100 patients who underwent NECT and supraaortic CTA on a 64-slice CT system were retrospectively included in the study. Bone removal was performed by subtraction of the NECT data from the CTA data using a dedicated workstation. Standard and BS CTA of each patient was reviewed for delineation of cerebral vasculature (grading scale from 1 = “excellent delineation” to 10 = “hardly any delineation”), reading time and depiction of vascular pathologies (standardized catalog) by two blinded readers. For BS data sets, the quality of BS was rated by a combination of the criteria complete bone removal, depiction of vascular structures and sufficient quality for diagnostic evaluation. The use of BS significantly reduced reading time from 4.60 min to 3.49 min (p < 0.001). Performing BS, the quality of vascular delineation of the cerebral arteries, cerebral veins and cavernous segment of the ICA increased significantly as compared to standard CTA (1.70 vs. 2.70; 2.60 vs. 4.12; 2.35 vs. 4.40, all p < 0.001). Consensus reading showed 41 pathologies in 35 patients. Diagnosis was missed or wrong overall in 15 cases, with 3 missed aneurysms (CTA: 2 vs. BS: 1), 8 wrong stenotic findings (CTA: 3 vs. BS: 5) and 4 missed partial thromboses (CTA: 2 vs. BS: 2). Performing BS in supraaortic CTA for the evaluation of cerebral vasculature reduces reading time and improves delineation of vessels. Diagnostic accuracy in general is not improved by BS, as the diagnostic accuracy of stenotic vessel alterations is reduced by potential truncation artifacts, but the detection rate of cerebral aneurysms slightly increases.