Daniel Hou MD, Presenter: Nothing to Disclose

Chris Davison MD, Abstract Co-Author: Nothing to Disclose

Anja Judith Reimann MD, Abstract Co-Author: Nothing to Disclose

Steven John Co MD, BSc, Abstract Co-Author: Nothing to Disclose

Luck Jan-Luck Louis MD, Abstract Co-Author: Nothing to Disclose

Savvas Nicolaou MD, Abstract Co-Author: Nothing to Disclose

To compare the image noise, grey-white matter differentiation, and subjective image quality of head CT images reconstructed with filtered back projection (FBP) and 100% iterative reconstruction in space (IRIS). As well, to compare the attenuation of intracranial hemorrhage between FBP and IRIS.

20 consecutive emergency room patients undergoing head CT on a dual source 128 slice scanner scanned utilizing our routine reduced dose protocol (120kV, 425mAs) were included. Images were reconstructed using the standard FBP and IRIS. Regions of interest (ROI) were acquired over the globes and lateral ventricles to obtain signal noise. ROIs were also obtained over normal gray matter (GM) and normal white matter (WM) with the Hounsfield units (HU) and standard deviations recorded. Signal to noise (SNR) and contrast to noise (CNR) ratios were calculated in the GM and WM. Subjective image quality was evaluated by two radiologists in a blinded and randomized manner via a four point grading scale. The CTDI volume, dose-length product and patient demographics were also recorded. 10 consecutive emergency room patients were then selected based on the presence of acute subdural hematoma. ROIs were acquired over the areas of hemorrhage with the HU and standard deviations recorded. Statistical analysis comparing both protocols was performed utilizing the student t-test.

The image noise for the examinations reconstructed using IRIS had a lower mean noise than FBP (p<0.001 for noise in the globe and in the ventricles). No statistical significant difference was noted in HU of the GM and WM (p=0.55 in the GM and p=0.59 in the WM). SNR was lower via IRIS than FBP, but no difference was noted in CNR (p<0.001 for SNR and p=0.46 for CNR). Mean DLP was measured to be 1274.8 +/- 63.7 mGycm (min=1188, max=1393) and mean CTDIvol was measured to be 79.86 +/- 0.75 mGy. No statistical significance was seen between HU for hemorrhage between IRIS and FBP (mean for IRIS 72.61HU, mean for FBP 72.16, p=0.85).

Iterative reconstruction in space preserves diagnostic image quality while significantly reducing image noise, enabling the potential for further significant radiation dose reduction.

Utilizing IRIS allows for improved image noise as compared to FBP, allowing for the future potential of radiation dose reductions in head imaging.

Hou, D, Davison, C, Reimann, A, Co, S, Louis, L, Nicolaou, S, Utilizing Iterative Reconstruction in Space for Trauma Head Imaging: Evaluation of Image Noise Reduction and Diagnostic Image Quality.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11034628.html