Program Information
Voxel-Based Enhancement Pattern Mapping (EPM) Improves the Contrast-To-Noise Ratio (CNR) and Detection of Abnormalities in Multiphase Contrast-Enhanced CT (MPCT)
P Park*, G Choi , E Koay , UT MD Anderson Cancer Center, Houston, TX
Presentations
MO-F-205-9 (Monday, July 31, 2017) 4:30 PM - 6:00 PM Room: 205
Purpose: Current MPCT relies on the tumor to normal tissue contrast seen at each phase limiting the ability to detect subtle changes in enhancement patterns that may differentiate normal from abnormal signals. We developed and characterized a novel voxel-based enhancement pattern mapping (EPM) signal analysis. We investigated the ability of EPM to improve CNR in a controlled phantom study and showed its applicability to clinical cases of hepatobiliary cancers
Methods: EPM was applied to both carefully designed MPCT of virtual phantoms and clinically acquired images from patients with hepatobiliary cancers. The phantoms consisted of liver tissue and tumor with distinct contrast enhancement signals (5 time points) that followed arterial enhancement of tumor and dual uptake of liver. Patient scans involved iodine-based contrast liver protocol CT scans (up to 5 scans acquired in total) under active breath-hold with a real-time motion management system. For analysis, two different images (EPM-Coefficient [EPM-C] and EPM-Distance [EPM-D]) were created. EPM-C was created by mapping the coefficient computed from each voxel’s early enhancement and late washout pattern to piece-wise polynomial functions and EPM-D was created by computing voxel’s mean square error over all time points and compared against the known normal tissue model.
Results: CNRs were calculated on varying tumor size (2mm to 30mm) under varying noise level. When compared to conventional CT images at all given time points, EPM-C and EPM-D showed statistically significant (p<0.05) CNR improvement for tumor sizes down to 4mm (1.24±0.7), as compared to standard gray scale imaging (0.7±0.42). On patient data sets, both EPM images significantly improved the visibility of tumor boundary and heterogeneity within the gross disease area.
Conclusion: Quantification and visualization of voxel-based enhancement pattern increases CNR and detectability of abnormal region as it allows to characterize overall time signal of uptake and washout of contrast through different tissues.
Funding Support, Disclosures, and Conflict of Interest: A part of this research was funded by Project Purple Research Program
Contact Email: