Estimation of Dual-Input Blood Volumes Using Dynamic Contrast-Enhanced MRI

1
Estimation of Dual-Input Blood Volumes Using
Dynamic Contrast-Enhanced MRI Michael H.
Rosenthal, MD, PhD Hiroto Hatabu, MD,
PhD Francine Jacobson, MD, MPH
2

• Traditional Perfusion Analysis
• Exponential kinetic models
• Homogeneous compartments and voxels
• Limited emphasis on dual-input vascular
  supplies
• Challenges
• Subvoxel characterization
• Measurement and modeling of dual inflows
• Temporal resolution

3

• Composite Voxels
• Consider tissue voxels as a mixture of reference
  vascular signals
• Estimate weighting factors ai using least squares
• Subtracted signal used to isolate enhancement

4

• Monte Carlo Simulation
• Numerical phantom
• Six vascular compartments
• Eleven mixed targets
• Gaussian white noise to test SNR from 0.1 to 5.0
• ROIs from 1 to 900 pixels
• Sampling from 1/s to 0.1/s
• 206,250 iterations

5

• Results
• Standard error in vessel contents 3
• SNR 0.2 and ROI 100 pixels at 0.1 Hz
• SNR 1 and ROI 25 pixels at 0.1 Hz
• Standard error in vessel contents 1
• SNR 1 and ROI 81 pixels at 0.1 Hz

6
Clinical DCE-MRI Example
7
Characterization of Pulmonary Tissues in Clinical
DCE-MRI Cases
Tissue of Interest Pulmonary Arterial Fraction Volume Systemic Arterial Fraction Volume
Pleural Mesothelioma 0.04 0.29
Normal Lung 0.11 0.02
Chronic Atelectasis 0.06 0.17
Normal Lung 0.06 0.02
8
Conclusions Viewing voxels as mixtures of
reference signals allows subvoxel estimation in a
simple numerical phantom Early anecdotal
promise in clinical applications Prospective
clinical evaluation in progress
9
This work was supported in part by RSNA Research
and Education Foundation Resident Research Grant
RR0826. Questions?