Needle-Induced HU Bias Correction for Quantitative CT Thermometry in CT-Guided Microwave Ablation: An Ex Vivo Feasibility Study.

RATIONALE AND OBJECTIVES: Metallic microwave ablation antennas introduce structured Hounsfield unit (HU) bias that can degrade quantitative CT thermometry. We developed a needle-centered method to suppress its repeatable low-frequency component. MATERIALS AND METHODS: This ex vivo feasibility study used six porcine liver specimens. Three livers provided independent room-temperature HU-temperature calibration, yielding κ=-0.31°C/HU. HU was measured as the mean within a 2-mm-diameter circular ROI centered at the temperature-probe tip. E1 used two no-antenna repeat scans; E2 used three scans at each of two antenna orientations; E5 acquired three scans at each of three orientations, with one representative matched comparison per orientation used for the orientation-specific AAI and low-frequency suppression analysis. E4 comprised two spatially separated, independent 50-W ablation runs in the same liver, evaluated at 3 and 5 min, with nine predefined locations per run. Each run used its own pre-heating baseline and correction profile. RESULTS: Antenna placement produced a repeatable structured HU-bias component beyond baseline variability. Correction reduced large- and mid-scale low-frequency energy by 77.36% and 71.69%, respectively. Across all nine E4 locations, median absolute temperature error decreased from 6.8°C to 6.1°C at 3 min and from 7.1°C to 5.9°C at 5 min. At the three corrected near-antenna locations, median absolute error decreased from 6.8°C to 2.2°C and from 7.1°C to 2.0°C, respectively. CONCLUSION: Needle-centered HU-bias correction improved CT thermometry in controlled ex vivo ablation, particularly at near-antenna locations with measurable bias.