Alejandro Sisniega PhD, Presenter: Research Grant, Carestream Health, Inc

Wojciech Zbijewski PhD, Abstract Co-Author: Research Grant, Carestream Health, Inc

Hao Dang, Abstract Co-Author: Research Grant, Carestream Health, Inc

Jennifer Xu, Abstract Co-Author: Research Grant, Carestream Health, Inc

Joseph Webster Stayman PhD, Abstract Co-Author: Research Grant, Varian Medical Systems, Inc

John Yorkston PhD, Abstract Co-Author: Employee, Carestream Health, Inc

Nafi Aygun MD, Abstract Co-Author: Nothing to Disclose

Vassiliss Koliatsos MD, Abstract Co-Author: Nothing to Disclose

Jeffrey H. Siewerdsen PhD, Abstract Co-Author: Research Grant, Siemens AG Consultant, Siemens AG Research Grant, Carestream Health, Inc Royalties, Elekta AB

CT is sensitive to the detection of fresh blood in the brain (30-50 HU contrast) and is a front line modality for diagnosis of traumatic brain injury (TBI). Benefit to early detection of TBI would be gained from imaging at the point-of-care immediately following suspected injury. We report dedicated cone-beam CT (CBCT) system with image quality sufficient for detection of mild-moderate TBI (e.g., 1-4 mm fresh blood) suitable to point-of-care deployment.

CBCT image quality requires novel system design, high-quality reconstruction, and high-fidelity artifact correction, including x-ray scatter, image lag, veiling glare, and beam hardening. Scatter correction uses a fast Monte Carlo (MC) simulator combining GPU parallelization, variance reduction, and denoising to provide corrections in less than 5 min. Detector lag is corrected by deconvolution with a measured temporal response function. Off-focal radiation and veiling glare are mitigated by deconvolution with the long tails of the detector point spread function. Beam hardening is compensated using the Joseph-Spital approach. The framework was tested on CBCT data of a head phantom including simulated brain and hemorrhages (~40 HU) ranging 2-10 mm diameter. The system design included a flat-panel detector with source-axis and source-detector distance of 58 and 80 cm, respectively, and acquisition protocol 100 kVp, 285 mAs (13.4 mGy). 

Uncorrected CBCT data exhibited non-uniformity (NU) of 165 HU and contrast-to-noise ratio (CNR) of 1.38 (blood-to-brain) with numerous major artifacts. Scatter correction improved the uniformity to NU = 48 HU and increased CNR by 84%. Lag and off-focal glare correction improved uniformity (NU = 46 HU) without increase in noise (CNR = 2.42). Beam hardening correction further improved non-uniformity to 10 HU, particularly at the skull base and peridural periphery. PL reconstruction reduced noise compared to FBP without loss in resolution, yielding CNR = 2.81.

Multi-component artifact correction was essential to achieving CBCT image quality suitable to low-contrast soft-tissue imaging of the brain. Initial results support the development of a novel system for point-of-care TBI detection.

Advanced CBCT artifacts correction techniques allow detection of subtle microhemorrhages in the brain, enabling point-of-care imaging of mild-moderate traumatic brain injury.

Sisniega, A, Zbijewski, W, Dang, H, Xu, J, Stayman, J, Yorkston, J, Aygun, N, Koliatsos, V, Siewerdsen, J, High Performance Cone Beam CT Imaging of Acute Traumatic Brain Injury.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14019381.html