Toward more precise microwave ablation systems



Journal Title

Journal ISSN

Volume Title



Thermal ablation procedures aim to eradicate targeted tissue by heating to temperatures exceeding 55-60 °C, where cell death occurs through coagulative necrosis. In the context of tumor ablation, the goal of the procedure is to have the ablation zone encompass the pathological lesion and a surrounding 5-10 mm margin of normal tissue. Microwave ablation (MWA) is an image-guided, energy-based technology, typically operating at 915 MHz or 2.45 GHz, that is widely used for treating various solid tumors, including those in the liver, kidney, and lung. During an MWA treatment, monitoring the temperature and estimating the attributed to resultant thermal damage can provide valuable insights for clinicians. The absence of a widely available, reliable, low-cost, real-time imaging system is a recognized unmet clinical need, posing a barrier to wider clinical adoption of microwave thermal ablation procedures. This dissertation presents an electromagnetic transmission coefficient-based method for monitoring MWA procedures where the intended target is bracketed by two or more applicators. The electromagnetic transmission signal between the applicators depends on the geometry between the applicators and the medium between them. During ablation, the latter changes, and since the tissue dielectric properties change, this signal varies accordingly. A custom experimental ablation platform was developed to facilitate periodic switching between "heating mode," during which power from the generator was coupled to the applicators, and "monitoring mode," during which the applicators were connected to a network analyzer for broadband transmission coefficient (s21) measurements. The initial aim was to assess the feasibility of monitoring the transient spread of thermal ablation zones using the microwave transmission coefficient-based technique in ex vivo tissue. MWA was performed on ex vivo bovine liver using a pair of custom 2.45 GHz directional applicators. Experiments were conducted with applied powers ranging from 30 to 50 W per applicator for durations ranging between 53 and 1219 s. The transient s21 spectra collected throughout the ablations were analyzed to evaluate the feasibility of predicting the extent of ablation zones, and predictions were then compared against ground truth assessments derived from images of sectioned tissue. A linear regression-based mapping between the processed transmission data and the ground truth measurements was derived to predict the extent of ablation. The normalized average transmission coefficient initially decreased rapidly and then increased before asymptotically approaching a steady state. The transition time ranged from 53 s at 45 W to 109 s at 30 W. Analysis of the ground truth ablation zone images indicated that the time to complete ablation ranged from 230 to 350 s. The relative prediction error for the time to complete ablation, derived from the s21 data, was between 1.6% and 2.3% compared to the ground truth. Next, an experimental study was designed to determine the feasibility of distinguishing between contiguous and discontiguous ablation zones using broadband transmission signals from a pair of omnidirectional-directional MWA applicators in an in vivo large animal model. Custom-built hardware facilitated the measurement of transmission signals during dual-applicator MWA at predetermined 46 s intervals. A multimodal ground truth was established using contrast-enhanced computed tomography (CECT) of ablation zones, cross-sectional and transversal gross pathologic assessments, and histopathologic evaluations of thermal injury. The study included 15 experiments on the livers of four domestic swine, with ablation durations of either 200 s or 600 s. Statistical analysis, correlation assessments, and exploratory data analysis were employed to assess the system’s capability to distinguish between contiguous and discontiguous ablation zones. Results showed clearly distinctive transmission signal datasets describing contiguous and discontiguous ablation zones. Spatiotemporal spread of ablation appears to be strongly correlated with measured and processed transmission signals (Spearman correlation coefficient 0.87, p < 0.0001). Based on processed transmission signal, a threshold value of 53% was determined as optimal for indicating ablation evolution through ROC analysis (AUC 0.84, p =0.02, permutation test, 100,000 permutations). To summarize, the feasibility of monitoring the spatiotemporal spread of thermal ablation zones using broadband microwave transmission coefficient measurements in ex vivo tissue, as well as effectively distinguishing between contiguous and discontiguous ablation zones in MWA procedures in in vivo tissue, were demonstrated. The presented technique holds potential to address the clinical need for a reliable method to monitor the spread of thermal ablation zones in real-time, potentially improving the precision and effectiveness of ablation therapies.



Thermal Ablation, Transmission Coefficient-based Monitoring, Microwave Ablation, Image-Guided Interventions, Energy-based Medical Devices, Signal Processing

Graduation Month



Doctor of Philosophy


Department of Electrical and Computer Engineering

Major Professor

Punit Prakash