Abstract Archives of the RSNA, 2014
Jessica Lee Nute MS, Presenter: Nothing to Disclose
Megan Jacobsen, Abstract Co-Author: Nothing to Disclose
Dawid Schellingerhout MD, Abstract Co-Author: Nothing to Disclose
Jim W. Pennington, Abstract Co-Author: Nothing to Disclose
Adam Grant Chandler PhD, Abstract Co-Author: Employee, General Electric Company
Dianna D. Cody PhD, Abstract Co-Author: In-kind support, General Electric Company
Intracranial hemorrhage and calcification with Single-Energy CT (SECT) attenuation levels below 100 HU cannot be reliably differentiated using currently clinically available means. Calcification is typically benign but hemorrhage can carry a risk of intracranial bleeding and contraindicate use of anticoagulant therapies. A biologically relevant phantom was constructed to examine the effects of lesion size, lesion location, and scan technique on the differentiating power of Dual-Energy CT (DECT).
Spectrally-equivalent brain material was fit into the cranial cavity of an anthropomorphic head phantom. Cylindrical lesion models (diameters: 0.5, 1.0 and 1.5cm) were created by adding calcium carbonate or ferric oxide to the background brain material. Seven sets of lesion models were created at matched SECT HUs from 40 to 100HU. Lesion models of each size were placed in the cerebrum, while just the largest lesion size was placed in the skull base. The phantom was scanned using a SECT routine brain protocol to verify the HU matching of the lesions as well as five DECT protocols representing CTDIvol levels from 27 to 81mGy each using four image thicknesses: 1.25, 2.5, 3.75 and 5mm. Each scan was reconstructed using a water/calcium material density pair. A two-component, 3D Gaussian mixture model was applied using a 50/50 training/validation approach. Accuracy of differentiation was calculated by comparing predicted voxel assignments with actual voxel identities.
Accuracy of differentiation improved with increasing dose, image thickness and lesion size. Accuracy was also notably poor in the skull base region. Using our current analysis method, differentiation was feasible in the cerebrum down to 70HU with close to 90% accuracy when 5mm images and 67mGy CTDIvol were applied to the 1.5cm lesions.
SEHU matched hemorrhage and calcification models less than 100HU could be distinguished using DECT. Future work will include expanded scan acquisition parameter sets and more sophisticated statistical analysis, which may provide stronger results.
The ability to distinguish between intracranial calcifications and hemorrhages using dual energy CT may help guide the use of anti-coagulant medications.
Differentiation of Low-attenuation Intracranial Hemorrhage and Calcification Using Dual-energy Computed Tomography in a Phantom System. Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14008032.html