Diffusion-Weighted MR Imaging in Acute Ischemia: Value of Apparent Diffusion Coefficient and Signal Intensity Thresholds in Predicting Tissue at Risk and Final Infarct Size
Dong Gyu Naa,d,
Vincent N. Thijsb,
Gregory W. Albersb,
Michael E. Moseleya and
Michael P. Marksa,c
a Department of Radiology, Stanford University Medical Center, Stanford, CA
b Department of Neurology, Stanford University Medical Center, Stanford, CA
c Department of Neurosurgery, Stanford University Medical Center, Stanford, CA
d Department of Radiology, Seoul National University Hospital, Seoul, Korea
View larger version (49K):
[in a new window]
|
FIG 1. Initial DW image and ADC maps at 4.5 hours and follow-up T2-weighted image at 7 days after symptom onset in a patient with acute left MCA infarction. Initial DW image lesion, or ROIDW image CORE (red ROI), was smaller than the final infarct, or ROIFINAL (blue ROI). Initial hyperintense lesion in the left insula is not obvious on the follow-up T2-weighted image at this level. Infarct growth volume was 6.1 cm3 (initial DW image volume, 27.9 cm3; final volume, 34.1 cm3). Quantitative DW images and ADC maps show color-coded regions of infarct growth not identified on initial maps and were generated by using pixel-based cutoff values of ADC ratio of 0.79 and signal intensity ratio of 1.19. Color-coded signals are also seen in the unaffected areas on quantitative DW image or ADC map by using these thresholds and were ignored in the analysis.
| |
View larger version (17K):
[in a new window]
|
FIG 2. Histograms of pixel values for ADC ratio (left) and signal intensity ratio (right). Histogram for ADC ratio shows a leftward shift of the curve in ROIIG (B) compared with the curve for normal tissue, or ROICONTIG (C). There is further a leftward shift of ADC ratios in ROIDW image CORE (A) compared with that in infarct growth (B). Histogram of signal intensity ratio shows similar patterns (with a rightward shift). Pixel values for infarct growth and normal tissue significantly overlap. The overlap was greater with the ADC ratio than with the signal intensity ratio.
| |
View larger version (16K):
[in a new window]
|
FIG 3. Pixel-based ROC curves for ADC, ADC ratio, and signal intensity ratio differentiate infarct growth and final infarct from normal tissue. Areas under the curves for signal intensity ratio were largest in predicting infarct growth (0.705) and final infarct (0.841), and the area under the curve for ADC ratio was slightly larger than that for the absolute ADC value in predicting infarct growth (0.609 vs 0.567) and final infarct (0.760 vs 0.743). Curves did not show that these three values were highly accurate in differentiating infarct growth from normal tissue; each parameter is worse than in the prediction of final infarct. Curves did show that the signal intensity ratio was better than the ADC or ADC ratio for predicting infarct growth or final infarct.
| |
