M J van Amerongen1, P Mariappan2,3, P Voglreiter4, R Flanagan2, S F M Jenniskens5, M Pollari6, M Kolesnik7, M Moche8, J J Fütterer5,9. 1. Department of Radiology and Nuclear Medicine, Radboudumc, Nijmegen, The Netherlands. Martin.vanAmerongen@radboudumc.nl. 2. NUMA Engineering Services Ltd., Louth, Ireland. 3. Department of Mathematics and Statistics, IIT Tirupati, Tirupati, India. 4. Institute of Computer Graphics and Vision, Graz University of Technology, Graz, Austria. 5. Department of Radiology and Nuclear Medicine, Radboudumc, Nijmegen, The Netherlands. 6. Department of Computer Science, Aalto University School of Science and Technology, Espoo, Finland. 7. Fraunhofer Institute for Applied Information Technology FIT, Sankt Augustin, Germany. 8. Department of Interventional Radiology, Helios Park-Klinikum Leipzig, Leipzig, Germany. 9. Robotics and Mechatronics (RaM), University of Twente, Enschede, The Netherlands.
Abstract
OBJECTIVES: Radiofrequency ablation (RFA) can be associated with local recurrences in the treatment of liver tumors. Data obtained at our center for an earlier multinational multicenter trial regarding an in-house developed simulation software were re-evaluated in order to analyze whether the software was able to predict local recurrences. METHODS: Twenty-seven RFA ablations for either primary or secondary hepatic tumors were included. Colorectal liver metastases were shown in 14 patients and hepatocellular carcinoma in 13 patients. Overlap of the simulated volume and the tumor volume was automatically generated and defined as positive predictive value (PPV) and additionally visually assessed. Local recurrence during follow-up was defined as gold standard. Sensitivity and specificity were calculated using the visual assessment and gold standard. RESULTS: Mean tumor size was 18 mm (95% CI 15-21 mm). Local recurrence occurred in 5 patients. The PPV of the simulation showed a mean of 0.89 (0.84-0.93 95% CI). After visual assessment, 9 incomplete ablations were observed, of which 4 true positives and 5 false positives for the detection of an incomplete ablation. The sensitivity and specificity were, respectively, 80% and 77% with a correct prediction in 78% of cases. No significant correlation was found between size of the tumor and PPV (Pearson Correlation 0.10; p = 0.62) or between PPV and recurrence rates (Pearson Correlation 0.28; p = 0.16). CONCLUSIONS: The simulation software shows promise in estimating the completeness of liver RFA treatment and predicting local recurrence rates, but could not be performed real-time. Future improvements in the field of registration could improve results and provide a possibility for real-time implementation.
OBJECTIVES: Radiofrequency ablation (RFA) can be associated with local recurrences in the treatment of liver tumors. Data obtained at our center for an earlier multinational multicenter trial regarding an in-house developed simulation software were re-evaluated in order to analyze whether the software was able to predict local recurrences. METHODS: Twenty-seven RFA ablations for either primary or secondary hepatic tumors were included. Colorectal liver metastases were shown in 14 patients and hepatocellular carcinoma in 13 patients. Overlap of the simulated volume and the tumor volume was automatically generated and defined as positive predictive value (PPV) and additionally visually assessed. Local recurrence during follow-up was defined as gold standard. Sensitivity and specificity were calculated using the visual assessment and gold standard. RESULTS: Mean tumor size was 18 mm (95% CI 15-21 mm). Local recurrence occurred in 5 patients. The PPV of the simulation showed a mean of 0.89 (0.84-0.93 95% CI). After visual assessment, 9 incomplete ablations were observed, of which 4 true positives and 5 false positives for the detection of an incomplete ablation. The sensitivity and specificity were, respectively, 80% and 77% with a correct prediction in 78% of cases. No significant correlation was found between size of the tumor and PPV (Pearson Correlation 0.10; p = 0.62) or between PPV and recurrence rates (Pearson Correlation 0.28; p = 0.16). CONCLUSIONS: The simulation software shows promise in estimating the completeness of liver RFA treatment and predicting local recurrence rates, but could not be performed real-time. Future improvements in the field of registration could improve results and provide a possibility for real-time implementation.
Authors: Peter Schullian; Edward Johnston; Gregor Laimer; Daniel Putzer; Gernot Eberle; Arno Amann; Maria Effenberger; Manuel Maglione; Martin C Freund; Alexander Loizides; Reto Bale Journal: Eur Radiol Date: 2020-10-30 Impact factor: 5.315