(Sunday, 7/29/2018) 5:05 PM - 6:00 PM
Room: Room 207
Purpose: CT material decomposition has found application in angiography, anthrography, and perfusion scans as well as emerging multi-contrast agent studies. We have previously derived a model-based material decomposition (MBMD) algorithm for direct estimation of material densities from projection data. This MBMD approach has relaxed sampling requirements (not requiring coincident projection rays for each spectral channel as with projection-domain algorithms) permitting a wide variety of acquisition strategies. In this work we explore these capabilities, applying MBMD to novel beam filtration and kV-switching schemes in physical cone-beam CT data for three-material decomposition of water, iodine, and gadolinium.
Methods: A phantom containing four vials of different concentrations of iodine (0-75 mg/mL) and gadolinium (0-19.66 mg/mL) was scanned using four different acquisition strategies. X-ray techniques of 80 and 130 kVp with silver and erbium filtration were acquired. These data were combined to emulate different spatial filtering and kV-switching strategies including split-filter and tiled filter designs. The data were reconstructed using the proposed MBMD algorithm, which simultaneously performed the reconstruction and decomposition.
Results: The MBMD algorithm was able to provide good reconstructions for the novel sampling patterns considered, providing density estimates for water, iodine, and gadolinium. For both iodine and gadolinium, the relative concentrations in each of the four vials were accurately determined. The RMSE of the concentration values were 5.29 mg/mL (iodine) and 1.85 mg/mL (gadolinium) with the split filter, and 5.91 mg/mL and 2.35 mg/mL with the tiled filter.
Conclusion: Preliminary results in physical data investigations show that the MBMD algorithm permits relaxed spatial and spectral sampling conditions through simultaneous reconstruction and material decomposition. We expect to reduce biases in these results through improved physical modeling (e.g., scatter correction and improved spectral calibration). The flexibility of the MBMD technique has the potential to enable new spectral CT designs by relaxing traditional sampling requirements.
Funding Support, Disclosures, and Conflict of Interest: Funded in part by NIH grant F31 EB023783
Keywords
Cone-beam CT, Spatial Filtering, Energy Spectrum
Taxonomy
IM- Cone Beam CT: Dual-energy and spectral
Contact Email
