Dose related, comparative evaluation of a novel bone-subtraction algorithm in 64-row cervico-cranial CT angiography

PurposeComparative evaluation of a low-dose scan protocol for a novel bone-subtraction (BS) algorithm, applicable to 64-row cervico-cranial (cc) CT angiography (MSCTA).Methods and patientsBS algorithm assessment was performed in cadaveric phantom studies by stepwise variation of tube current and head malrotation using a 64-row CT scanner. In order to define minimum dose requirements and the rotation correction capacity, a low dose BS MSCTA protocol was defined and evaluated in 12 patients in comparison to a common manual bone removal algorithm. Standard MIPs of both modalities were evaluated in a blinded manner by two neuroradiologists for image quality composed, of vessel contour sharpness and bony vessel superposition, by using a five-point score each. Effective Dose (E) and data post-processing times were defined.ResultsIn experimental studies prescan tube current could be cut down to one-sixth of post-contrast scan doses without compromise of bone-subtraction whereas incomplete subtraction appeared from four degrees head malrotation on. Prescan E amounted to additional 1.1 mSv (+25%) in clinical studies. BS MSCTA performed significantly superior in terms of bony superposition for vascular segments C3–C7 (p < 0.001), V1–V2, V3–V4 (p < 0.05, p < 0.001 respectively) and the ophthalmic artery (p < 0.05), whereas vessel contour sharpness in BS MSCTA only proved superior for arterial segments V3–V4 (p < 0.001) and C3–C7 (p < 0.001). MBR MSCTA received higher ratings in vessel contour sharpness for C1–C2 (p < 0.001), callosomarginal artery (p < 0.001), M1, M2, M3 (p < 0.001 each) and the basilar artery (p < 0.001). Reconstruction times amounted to an average of 1.5 (BS MSCTA) and 3 min (MBR MSCTA) respectively.ConclusionThe novel BS algorithm provides superior skull base artery visualisation as compared to common manual bone removal algorithms, increasing the Effective Dose by one-fourth. Yet, inferior vessel contour sharpness was noted intracranially, thus limiting the BS algorithm use to patients with suspected vessel pathology at the skull base level.