Room: AAPM ePoster Library
Purpose: is a periarticular response that arises from the presence of monosodium urate (MSU) crystal deposition and joints. Gout can be separated from pseudo gout and other materials using dual energy computed tomography (DECT). The purpose of this study is to use a gout phantom model to quantitatively compare hybrid iterative reconstruction (ASiR-V) to deep learning image reconstruction (TrueFidelity (TF) for GSI).
Methods: gelatin-based, gout phantom was created with uric acid (UA) gout inserts of 3 sizes (10,7,2 mm³) and 5 concentrations (10-50% UA by mass). Pseudogout (calcium hydro phosphate), bone and tendon were also included as these materials are sources of false positives (tendon and pseudogout) and false negatives (bone) in DECT gout assessment.
The phantom was scanned on a Revolution CT (GE Healthcare, Waukesha, WI) with gemstone spectral imaging (GSI) with: 80/140 kVp, 80mm collimation, 0.984 pitch, 335mA tube current, 0.8s rotation, and 0.625mm slice thickness. Images were reconstructed with Filtered Backprojection reconstruction, 50% ASiR-V and three levels of TrueFidelity (low, medium, and high). Gout was segmented on an AW workstation (GE Healthcare, Waukesha, WI) using thresholds applied on Uric Acid(HAP) and HAP(Uric Acid) material images (1200 – 1378 mg/ml on Uric Acid(HAP) and -5-130 mg/ml on HAP(Uric Acid)). The threshold values were determined empirically from clinical gout cases to maximize gout segmented and minimize false positives. Noise was measured as standard deviation in uniform regions of the phantom on 70 keV virtual monochromatic images
Results: gout volume in the phantom was consistent when comparing the FBP, ASIR-V and TrueFidelity reconstructions (FBP: 20.0cm³, ASIR-V 50%: 19.9cm³, TF-L,M,H: 20.5cm³). Noise was reduced in the reconstructions with TrueFidelity compared with FBP (FBP: 5.6, 50% ASIR-V: 4.9, TF-L,M,H: 5.1, 4.6, 4.2).
Conclusion: allows for additional noise reduction in gout image without compromising quantitative accuracy.
Funding Support, Disclosures, and Conflict of Interest: All authors are employees of GE Healthcare.
Not Applicable / None Entered.