Feasibility of a low-dose orbital CT protocol with a knowledge-based iterative model reconstruction algorithm for evaluating Graves' orbitopathy

To evaluate the clinical feasibility of low-dose orbital CT with a knowledge-based iterative model reconstruction (IMR) algorithm for evaluating Graves' orbitopathy.

Low-dose orbital CT was performed with a CTDI_vol of 4.4 mGy. In 12 patients for whom prior or subsequent non-low-dose orbital CT data obtained within 12 months were available, background noise, SNR, and CNR were compared for images generated using filtered back projection (FBP), hybrid iterative reconstruction (iDose⁴), and IMR and non-low-dose CT images. Comparison of clinically relevant measurements for Graves' orbitopathy, such as rectus muscle thickness and retrobulbar fat area, was performed in a subset of 6 patients who underwent CT for causes other than Graves' orbitopathy, by using the Wilcoxon signed-rank test.

The lens dose estimated from skin dosimetry on a phantom was 4.13 mGy, which was on average 59.34% lower than that of the non-low-dose protocols. Image quality in terms of background noise, SNR, and CNR was the best for IMR, followed by non-low-dose CT, iDose⁴, and FBP, in descending order. A comparison of clinically relevant measurements revealed no significant difference in the retrobulbar fat area and the inferior and medial rectus muscle thicknesses between the low-dose and non-low-dose CT images.

Low-dose CT with IMR may be performed without significantly affecting the measurement of prognostic parameters for Graves' orbitopathy while lowering the lens dose and image noise.