BLUE RIBBON POSTER IMAGING: Iodine Quantification Error Due to Non-Water Backgrounds: A Phantom Study
Abstract
Purpose
Material quantification in spectral CT assumes a known voxel background, such as water in iodine maps or hydroxyapatite in bone edema maps. These assumptions are never met in vivo, where voxels typically contain mixtures of tissues with differing effective atomic numbers and dual-energy ratios. This study quantifies iodine density quantification errors arising from incorrect background material assumptions in spectral CT material decomposition.
Methods
Two sets of custom phantoms containing sodium iodide (an iodinated contrast surrogate) were constructed with different background materials: 70% isopropyl alcohol and 20% sodium chloride in pure water. Iodine concentrations ranged from 0 to 17.0 mg I/mL, representing clinically relevant values. The background materials spanned a range of dual-energy ratios, with 70% isopropyl alcohol exhibiting a ratio lower than water and 20% sodium chloride exhibiting a ratio higher than water, mimicking in vivo variation associated with tissues such as fat and dense liver. Phantoms were scanned using a fast kV-switching dual-energy CT system (Discovery 750 HD, GE HealthCare, Waukesha, WI) and reconstructed using clinically available iodine (water) material density maps. Additionally, attenuation vectors from NIST for the background materials and sodium iodide were used for material decomposition, generating iodine (true background), sodium iodide (water), and sodium iodide (true background) image sets. Quantification error was defined as the difference between measured and known iodine concentrations.
Results
The 70% isopropyl alcohol background produced systematic underestimation, while the 20% sodium chloride background produced systematic overestimation, consistent with theoretical predictions based on deviations from water the dual-energy ratio. Errors were reduced when the decomposition material pair reflected the true phantom composition.
Conclusion
Non-water backgrounds representative of clinical conditions introduced predictable iodine quantification errors in spectral CT, which can be mitigated by appropriate material decomposition choices.