Ablation zone prediction: A comparison of two manufacture provided models.

This submission has open access
Submission ID :
GEST2021-40
Submission Type
Submission Topic
Purpose :
Microwave ablation (MWA) has been firmly established as a treatment option for both primary and secondary malignancies of the liver. While modern MWA machines have improved over prior versions, ability to accurately predict the ablation zone remains questionable. One commonly utilized 2450MHz system (Emprint, Metronic, Minneapolis, MN) provides both in vivo and ex vivo models for ablation zone (AZ) size prediction. The purpose of this abstract is to evaluate which is most accurate.
Materials & Methods :
Between 1/1/2015 and 2/1/2021 170 patients who underwent 224 ablations with a single MWA system at a single academic center were retrospectively reviewed. The patient’s electronic medical records were reviewed for to determine patient demographics, treatment variables, and follow up. AZ dimensions were measured at 1 month follow up in the anterior posterior (AP), transverse (TR), and craniocaudal (CC) directions.
Results :
The cohort consisted of, 123 (123/170, 72.4%) men and 57 (57/170, 27.6%) women with an average age of 64 ± 11.3 years. The mean percentage of predicted for AZ using the in vivo model was 96.5 ± 26.1%, 89.1 ± 31.5%, and 93.1 ± 28.2% in the AP, TR, and CC directions respectively. While the mean percentage of predicted in the ex vivo model was 96 ± 25.4%, 82 ± 28.6%, and 85.9 ± 26.4% in the AP, TR, and CC directions respectively.
The mean percentage of predicted for 100 watt treatments of ≤5.5 minutes in the in vivo model was 95.4 ± 18.6%, 87.3 ± 26.5% and 89.6 ± 17.6% in the AP, TR, and CC directions respectively. The mean percentage of predicted for 100 watt treatments of ≤5.5 minutes in the ex vivo mode was 103.8 ± 19.8%, 99 ± 26.2% and 102.4 ± 18.6% in the AP, TR, and CC directions respectively. The ex vivo model was significantly more accurate in the AP (103.8 ± 19.8% vs 95.4% ± 18.6%, p=0.03), TR (99 ± 26.2% vs 87.3 ± 26.5%, p=0.03), and CC (102.4 ± 18.6% vs 89.6 ± 17.6%, p=0.002) than the in vivo model in this setting.
The mean percentage of predicted for 100 watt treatments >5.5 minutes in the in vivo model was 98.3 ± 25.9%, 91.6 ± 31.9%, and 95.2 ± 28.2% in the AP, TR, and CC directions respectively. While the mean percentage of predicted for 100 watt treatments >5.5 minutes in the ex vivo model is 94.7% ± 24.8%, 78.2 ± 27.2%, and 81.3 ± 24.2% in the AP, TR, and CC directions respectively. There was no significant difference in the ability of the in vivo and ex vivo to predict the AP size of the AZ (98.3 ± 25.9% vs 94.7% ± 24.8%, p=0.13), however, the TR (91.6 ± 31.9% vs 78.2 ± 27.2%, p=0.001) and CC (95.2 ± 28.2% vs 81.3 ± 24.2%, p< 0.001) were both significantly more accurately predicted by the in vivo as compared to the ex vivo model.
Conclusions :
The in vivo model, which more realistically recreates the clinical scenario, was more accurate than the ex vivo model in many situations. However, surprisingly the ex vivo model outperformed the in vivo model at some time settings. These findings can help guide users as to which model is best for a given treatment.

Associated Sessions

University of Minnesota
University of Minnesota
University of Minnesota
University of Minnesota
Minneapolis VA

Similar Abstracts by Type

Submission ID
Submission Title
Submission Topic
Submission Type
Primary Author
GEST2021-85
Liver
Live - Oral Presentation
Mr. Courtney Henry
GEST2021-172
Live - Oral Presentation
Research Committee
GEST2021-173
Live - Oral Presentation
Research Committee
GEST2021-174
Live - Oral Presentation
Research Committee
GEST2021-175
Live - Oral Presentation
Research Committee
74 hits