Iterative reconstruction maintains or improves image quality while reducing dose in cranial CT

Iterative reconstruction improves the quality of images of stroke patients’ brains while also allowing for radiation dose reduction, according to researchers at Goethe University in Frankfurt, Germany.

Looking at 51 cranial CT studies, the team compared images reconstructed by filtered back projection (FBP) with sinogram-affirmed iterative reconstruction (SAFIRE) at five strength levels.

They found the SAFIRE reconstruction at strength level 4 produced the best images, unsurpassed in both subjective and objective criteria. Meanwhile this reconstruction allowed a dose reduction of 24 percent with no losses to the sharpness of visuals on ischemic lesions.

The research is running in the September edition of the American Journal of Roentgenology.

With its multiple steps of algorithmic computation, iterative reconstruction is considered a costlier alternative to the filtered back projection method of CT imaging reconstruction, as the latter directly calculates the image in one step.

The authors note that the “second generation” of SAFIRE they evaluated combines raw, data–based iterations for artifact reduction with image-based iterations using a pixel-level regularization that finely adjusts for image noise.

They cite previous studies showing this technique can improve image quality while lowering dose in CTs of the chest, coronary arteries, abdomen, cervical spine and other regions.

The new findings show low-dose cranial CT with SAFIRE S4 as superior to standard-dose cranial CT with conventional reconstruction as assessed by subjective criteria and at least not inferior regarding objective criteria, they report.

“Other vendor-specific iterative reconstructions show the same tendency,” the authors point out. “The subjective evaluation showed similar results in standard- and low-dose cranial CT, which implies the possibility of further dose reduction.”