A healthy 4-month-old female domestic pig under general anesthesia was used for this experiment. The pig was housed in the central animal facility of the Charité following the 2010/63/EU guidelines as well as the recommendation of the GM-Solas (Gesellschaft für Versuchstierkunde, Freiburg, Germany). At the end of the experiment, the pig was euthanized under deep general anesthesia.

A total of three MWAs were performed in the healthy pig liver and monitored with CT. During each MWA, the ablation probe (Emprint System, Medtronic, Meerbusch, Germany) was set to a power of 100 Watt, and ablation was performed for 5 min, as commonly done in clinical practice [15, 16]. The probe was placed in the liver, and two custom-made fiberoptic thermometers (Optocon and Weidmann Technologies, Dresden, Germany) were inserted parallel to the probe for continuous temperature monitoring during the procedure. Individual ablation areas were placed sufficient distance from each other. At the end of the experiment, the liver was removed for sampling and further examination. After liver resection, the ablation lesions were bisected and examined macroscopically (Supplementary Fig. 1, A1–A3). The samples were photographed on a sheet of scaled millimeter paper, which allows precise sizing of the ablation area with dedicated software (MWANecrosisMeasurement, Fraunhofer Institute for Digital Medicine MEVIS, Bremen, Germany) [17].

Each MWA was monitored by CT, for which 20 spectral scans were acquired without table movement using 16-cm detector coverage (Canon Aquilion ONE Prism; Canon Medical Systems, Otawara, Japan). The following scan parameters were used: rapid kVp switching between 80 and 135 kVp, 1 s rotation time, and 500 mA tube current. The first ten scans were acquired every 30 s beginning before ablation started (T 0) and ending with the time of peak temperature (T max) to cover the upslope phase. The next 10 scans were acquired at 60-s intervals and covered the downslope, postablation phase. After withdrawal of the MWA probe, a CECT scan in the portal venous phase following administration of an intravenous (IV) contrast agent (100 ml Imeron 400 MCT, Bracco, Konstanz) was acquired for each ablation (Supplementary Fig. 1, B1–B3). CECT was performed according to a previously established protocol with a fixed amount of contrast medium [7, 18].

The CT system generated virtual mono-energetic reconstructions in a medium soft-tissue kernel using Advanced Intelligent Clear-IQ Engine (spectral AICE) with 75 keV. Primary reconstructions resulted in 0.5-mm slice thickness volume stacks. Motion artifacts were reduced by applying a double-registration approach consisting of elastic registration followed by rigid registration, selecting the probe tip at the time of maximum temperature in the center of the ablation zone as the reference point [19].

Based on macroscopic measurement, circular ROIs were defined specifically for each ablation to determine average attenuation (HU) in unenhanced CT scans during the upslope phase. These ROIs exceeded the largest macroscopically measured diameter by a few millimeters and were determined as follows: ablation A—28 + 2 mm, ablation B—22 + 2 mm, ablation C—44 + 2 mm (data presented as mean value of two measured diameters on each side of the lesion after bisection + small extension zone of 2 mm). These data and invasively measured temperatures served to calculate the slope using pooled data of all three ablations. A dedicated research CTT software (Canon Medical Systems, Japan) was used to visualize the ablation area. Here, (at least) two CT scans must be uploaded: one scan from before ablation (T 0) and another from the time of peak temperature (T max). To create a thermography map after uploading the registered CTT dataset, a slope is required, which was calculated before. The result is a colored map superimposed on the uploaded CT scan at T max. Red was chosen for better visualization of the ablated area and represents the temperature changes (ΔT) in the tissue. For the present study, the sensitivity of the colored map was set to indicate temperature changes greater than 33 K (ΔT ≥ 33 K). Thus, it can be assumed that at a core body temperature of 37 °C, tissue necrosis has been achieved at 70 °C within the colored zone (T core + ΔT > 70 °C). The largest extent of the ablation area in CTT was measured in the para-axial plane, perpendicular to the inserted probe. The ablation areas determined using CTT were compared with macroscopic size measurements and the ablation areas identified by CECT (Fig. 1). The ablation zone in CECT can be detected orthogonal to the probe insertion region. It is characterized by a lower tissue density with an enhancing margin, whereas a differentiated distinction of histological zones is not possible [7, 10].

Ablation areas determined by macroscopy and two CT imaging modalities. macro, macroscopy; CTT, computed tomography thermography; area measured using the slope of − 1.96 HU/°C. CECT, contrast-enhanced computed tomography

A patient with unresectable renal cell carcinoma (RCC) at the left renal pole was selected for retrospective analysis of the potential of CTT in a clinical setting. The patient underwent RFA of the RCC over 10 min using the RITA Starburst Semi-Flex applicator (Rita Medical Systems, Milwaukee, WI, USA). Pre- and postablation CT scans were acquired using the same acquisition parameters. The two scans were then selected to retrospectively determine the ablation zone based on CTT.

Average HU values determined in the ROIs placed in unenhanced CT scans were matched with the corresponding temperatures measured invasively during the upslope phase. Spearman’s correlation coefficient was calculated after performing data normality tests, and linear regression was performed for pooled data of all three ablations in the upslope phase. All statistics were planned in consultation with a biometric expert. Calculations were carried out using GraphPad Prism (Version 9.5.1, San Diego, California) software. A p-value < 0.05 was considered statistically significant.