13.07.2015 Views

Interventional 4-D C-Arm CT Perfusion Imaging Using Interleaved ...

Interventional 4-D C-Arm CT Perfusion Imaging Using Interleaved ...

Interventional 4-D C-Arm CT Perfusion Imaging Using Interleaved ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

PUBLISHED IN: IEEE TRANSA<strong>CT</strong>IONS ON MEDICAL IMAGING, VOL. 31, NO. 4, PAGES 892–906, APRIL 2012 6arterial enhancement [µ w]0.500.250.02∆µ (t) art∆µ tis,h(t)∆µ (t) tis,p00 5 10 15 20 25 30t [s]0 5 10 15 20 25 30 0 0.01tissue enhancement [µ w]η = 1 but, as will be explained in more detail later, we variedthese values to obtain several different curves.The maximum dynamic enhancement A above the staticbaseline value was chosen differently for each set in orderto model different injection methods. In the first set an intravenousbolus injection into the antecubital vein with A =0.3µ w (µ w = 0.18cm −1 is the X-ray attenuation of water)was assumed which is typical for diagnostic <strong>CT</strong> perfusionimaging [21]. In the second set A = 0.5µ w was chosen inorder to model an injection at the aortic arch as it was carriedout in the in vivo studies, see Section IV for details.In each set the time-attenuation curves ∆µ tis (t) in tissuewere computed using the indicator-dilution theory [13], [20],∫ t∆µ tis (t) = ρ CBF ∆µ art (τ)r(t−τ)dτ (9)0{1, for t < T0r(t) = ( )exp −(t−T0)(10)MTT−T 0, for t ≥ T 0 ,with ρ = 1.04g/ml. We set T 0 = 0.632×MTT as suggestedin [22]. According to the central volume theorem [13] MTTcan be computed asMTT = CBV/CBF . (11)Fig. 4. Top: Synthetic time-attenuation curves, here without noise, correspondingto a large artery (∆µ art(t), left scale), to healthy, normallyperfusedtissue (∆µ tis,h (t), right scale) and to pathological, hypoperfusedtissue (∆µ tis,p (t), right scale) assuming a contrast injection at the aortic arch.The plots show the dynamic enhancement (above the static baseline value)relative to the attenuation µ w of water. Note, the arterial peak enhancementis more than 25 times higher than the peak enhancement in tissue. Bottom:Locations of the artery and the tissue regions in the dynamic head phantom.A. Phantom DescriptionWe generated two sets of time-attenuation curves representingtwo different contrast injection methods (intra-venous/ aortic arch injection). An intra-venous injection is commonlyused in diagnostic perfusion <strong>CT</strong> imaging. An injectionat the aortic arch is feasible in interventional C-arm <strong>CT</strong>perfusion imaging since patients will be undergoing arterialcatheterization for their endovascular therapy [9]. Advantagesof the aortic arch injection method are further discussed inSection III-E.Each set contained three different time-attenuation curvesto model the flow of contrast agent through a large artery,a region of normally-perfused tissue and a region of hypoperfusedtissue. In each set the arterial enhancement ∆µ art (t)was modeled using the following gamma-variate function [14],[20]:∆µ art (t) =A(αβ exp(−1)) α τα exp(−τ/β)H(τ) . (8)Here, H(τ) is the unit step function, and α = 3.0 and β = 1.5are shape parameters that were also suggested in [20]. Thedimensionless quantity τ = (t−t 0 )/η depends on the bolusarrival time t 0 and the time scaling factor η which are bothmeasured in s. The factor η controls the full width at halfmaximum of the curve. The initial values were t 0 = 0 andWe chose CBF = 60ml/100g/min and CBV = 4ml/100gto generate a time-attenuation curve ∆µ tis,h (t) for healthy,normally-perfused tissue and chose CBF = 20ml/100g/minand CBV = 4ml/100g to generate a time-attenuation curve∆µ tis,p (t) for pathological, hypoperfused tissue. A plot of allthree time-attenuation curves for the case of the aortic archinjection is shown in Figure 4. Note, the peak enhancement intissue depends linearly on the arterial peak enhancement A —see (8) and (9) — and for the simulated intra-venous injectionit is about 0.011µ w (healthy tissue) and 0.006µ w (pathologicaltissue) which are typical values for the peak enhancementin gray matter in diagnostic perfusion <strong>CT</strong> imaging [21].For each set of time-attenuation curves a dynamic Shepp-Logan-type head phantom µ pha (x,t) was created that contained3 circular regions of interest (ROI) with varying attenuationvalues, see Figure 4. The first ROI (radius 1 mm) modeledan artery with attenuation values µ w +∆µ art (t). The secondand third ROIs (radius 2 mm each) modeled tissue regionswith attenuation values µ w +∆µ tis,h (t) and µ w +∆µ tis,p (t),respectively.The constant attenuation values of the elliptical skull (outerradii 62 mm and 92 mm) and the brain tissue were 2µ wand µ w , respectively. We have chosen the constant attenuationvalues of the two inner ellipses to be 0.95µ w . In order toimprove the reproducibility of our results, a more detaileddescription of this phantom along with relevant source codeis available online [23].B. InvestigationsThe following procedures were applied to each phantomconfiguration (intra-venous / aortic arch injection) individually.We simulated C-arm <strong>CT</strong> scanning of the 2-D phantom with alinear detector array using the scan parameters from set 1 in

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!