Adamkiewicz artery visualisation using iterative model-based reconstruction in low-dose computed tomograpyhy (CT)...

The use of iterative model-based reconstruction (IMR) for the artery of Adamkiewicz (AKA) visualisation using low-dose computed tomography (CT) angiography remains understudied. Therefore, we aimed to compare the image quality and AKA visualisation using a novel knowledge-based IMR with those using hybrid iterative reconstruction (iDose⁴) and filtered back projection (FBP) under a low-dose CT protocol.

Forty-five patients (73.3% male) with aortic aneurysm or dissection who underwent 256-slice multidetector row CT with a low-dose CT protocol (100 kVp and 20 mA) were enrolled. The acquired raw data were reconstructed using FBP, iDose⁴, and IMR and blindly analysed by two observers. Quantitative analysis assessed signal-to-noise ratio (SNR) of the aorta and contrast-to-noise ratio (CNR) of the anterior spinal artery measurement relative to the spinal cord on multiplanar reformatted images. Qualitative analysis evaluated AKA visualisation and its continuity with the intercostal or lumbar artery using a four-point scale.

In the quantitative analysis, both SNR and CNR of IMR (SNR: 35.2 ± 10.2; CNR: 3.8 ± 0.5) were significantly higher than those of iDose⁴ (SNR: 21.5 ± 4.9; CNR: 3.1 ± 0.8) and FBP (SNR: 14.1 ± 3.1; CNR: 2.6 ± 0.8) (all P<.05). In the qualitative analysis, the mean AKA visualisation score was also significantly better in IMR (3.6 ± 0.6) than in iDose⁴ (3.1 ± 0.8) and FBP (2.6 ± 0.8) (P<.05). The prevalence of assessable AKA was the highest in IMR (93.3%), followed by iDose⁴ (75.6%) and FBP (51.1%).

The IMR algorithm provided superior image quality and AKA visualisation compared with iDose⁴ and FBP when using low-dose CT angiography.