The aim of this study was to investigate the diagnostic accuracy of computed tomography (CT) for the prediction of ablation zones from microwave ablation (MWA) and cryoablation (CA) in an ex vivo porcine liver model.
Sequential (30 seconds) CT scans were acquired during and after MWA and CA in an ex vivo porcine liver model. We generated 120-kVp equivalent reconstructions of generic dual-energy CT data sets, and comprehensive region-of-interest measurements were statistically correlated with invasive temperature monitoring using Pearson correlation coefficient. Binary logistic regression was performed for prediction of successful ablation.
With the use of pooled data from 6 lesions in 2 separate experiments, correlation analysis of attenuation in Hounsfield units (HU) and temperature yielded r = −0.79 [confidence interval (CI), −0.85 to −0.71] for MWA and r = 0.62 (CI, 0.55 to 0.67) for CA.
For MWA, there was a linear association between attenuation and temperature up to 75°C; thus, linear regression yielded a slope of −2.00 HU/°C (95% CI, −1.58 to −2.41). For CA, a linear association between attenuation and temperature was observed in the cooling phase with a slope of 2.11 HU/°C (95% CI, 1.79 to 2.58). In MWA treatment, binary logistic regression separated less than 70°C and greater than 70°C with 89.2% accuracy. Within the ice ball, temperatures above and below −20°C were distinguished with 65.3% accuracy.
Our experiments reveal several difficulties in predicting ablation zone temperature from CT attenuation. Microwave ablation leads to gas production in the tissue, which degrades the accuracy of noninvasive temperature measurement, especially at higher temperatures. In CA, CT thermometry is limited by ice ball formation, which leads to homogeneous attenuation, nearly independent of temperature. Further research is needed to define the role of CT thermography in ablation zone monitoring in liver malignancies.