Computing Diffusion Rates in T2-dark Hematomas and Areas of Low T2 Signal
Joseph A. Maldjian
,a,
John Listeruda,
Gul Moonisa and
Faez Siddiqia
a From the Departments of Radiology (J.A.M., G.M., F.S.) and Psychiatry (J.L.), Hospital of the University of Pennsylvania, Philadelphia, PA.
View larger version (33K):
[in a new window]
|
FIG 1. T2-bright left temporal hyperacute hematoma (intracellular oxyhemoglobin).
A, Axial T1-weighted image (600/20/1) (left) shows isointense blood products. Axial fast spin-echo T2-weighted image (4000/85/1) (middle) shows high signal intensity. Diffusion-weighted isotropic image (10000/125/1) (right) shows increased signal relative to the brain.
B, Trace ADC map displayed without background masking (left) shows hematoma to have diffusion rates comparable with the brain. Trace ADC map displayed with 2% background masking (right) shows similar appearance for this T2-bright hematoma.
| |
View larger version (43K):
[in a new window]
|
FIG 2. T2-dark left frontal acute hematoma (intracellular deoxyhemoglobin with some intracellular methemoglobin).
A, T1-weighted image (600/20/1) (left) shows predominantly isointense blood products (deoxyhemoglobin) with some areas of hyperintensity centrally (intracellular methemoglobin). Fast spin-echo T2-weighted image (4000/85/1) (middle) shows predominantly low signal intensity corresponding to intracellular blood products. Diffusion-weighted isotropic image (10000/125/1) (right) shows marked hypointensity relative to the brain. This would intuitively be expected to correspond to fast diffusion (compare with CSF in ventricles).
B, Trace ADC map displayed without background masking (left) shows hematoma to have diffusion rates comparable with the brain, with some dropped points at the periphery. The dropped points are caused by background noise variation resulting in negative ADC values. Note that the diffusion within the hematoma is neither restricted nor fast relative to the brain. Trace ADC map displayed with 2% background masking (right) shows dramatic loss of pixels within hematoma.
| |
View larger version (36K):
[in a new window]
|
FIG 3. T2-dark right frontal subacute hematoma (intracellular methemoglobin).
A, T1-weighted image (600/20/1) (left) shows predominantly hyperintense blood products. Fast spin-echo T2-weighted image (4000/85/1) (middle) shows predominantly low signal intensity corresponding to intracellular blood products. Diffusion-weighted isotropic image (10000/125/1) (right) shows T2 blackout effect with marked hypointensity relative to the brain.
B, Trace ADC map displayed without background masking (left) shows hematoma to have diffusion rates comparable with the brain, with some dropped points within the lesion. Trace ADC map displayed with 2% background masking (right) shows dramatic loss of pixels within hematoma.
| |
View larger version (17K):
[in a new window]
|
FIG 4. Plot of average ADC of hematomas for method 1 (expected values) versus methods 2, 3, and 4 (at 2% masking) with line fits. A nearly perfect correspondence is shown between ADC values computed using method 1 and method 2 (mean ROI) (intercept = -0.13, slope = 1.2, R2 = 0.98, standard error = 0.038). Although method 3 (negative masked voxels) shows a slope close to identity, the fit is not as good as for method 2 (intercept = 0.056, slope = 0.98, R2 = 0.71, standard error = 0.144). Method 4 shows a poor correspondence with method 1 (intercept = -0.076, slope = 0.759, R2 = 0.36, standard error = 0.233), most notably at the lower ADC values
| |
View larger version (13K):
[in a new window]
|
FIG 5. Average ADC plotted against percent background masking. For the T2-dark hematomas, a dramatic decrease in ADC values is evident as more of the masked voxels (with values of 0) are included in the computation. For the T2-bright hematomas, there is only a small decrease in ADC values across the range of masking values. There is no variation in the ADC of white matter values, which shows that background masking becomes an issue only with areas of very low T2 signal
| |
AJNR - American Journal of Neuroradiology