Broadband electrical sensing and ablation devices for endoscopic cancer diagnostics and treatment

Cancer is currently the second most common cause of death in developed countries. Within this group, colorectal cancer is showing the highest increase in incidence, and it is the second deadliest at present. Therefore, novel techniques are constantly investigated to increase its detection rate in earlier stages where its survival rate is much higher. In the last century, since the discoveries of H. P. Schwan, the use of dielectric properties of tissues has gained in value. These novel insights have become useful techniques and applications in the war on cancer. This Ph.D. thesis follows in its track and tries to bring a new perspective to this broad research domain. Firstly, it makes suggestions towards a better-structured measurement protocol. Next, it applies said protocol for dielectric measurements of muscle tissue in a wide temperature range. Lastly, an applicator topology is proposed for dielectric detection and thermal therapeutics.

Various methods exist and are widely used to extract dielectric data in the frequency range from several Hz to THz. However, due to the lack of a standardized framework, comparative studies are challenging. The question remains whether the presented differences between different tissues originate from tissue characteristics, or whether the used measurement protocol introduces these discrepancies. Throughout the years, several confounders have been established that influence dielectric measurements. Nevertheless, this Ph.D. thesis presents two additional confounders that cannot be neglected. The first one considers dehydration of the tissue sample. This dehydration is especially important during measurements at body temperature or at hyperthermia or ablation temperatures. The presented research demonstrates that relative changes in the dielectric properties up to 9% were observed over 35 minutes at body temperature. However, reducing the air flow around the sample by storing it in a sealed container stabilizes the dielectric properties over time and makes a prolonged measurement period feasible.

Secondly, the influence of probe-to-tissue contact pressure of the most commonly used measurement probe was reported. When probing the tissue, an exponential decaying contact pressure was observed in which the tissue conforms to the probe. This tissue conforming process had time constants up to a minute. Therefore, a controlled lifting platform was introduced in the measurement setup to guarantee a stable pressure over time. In the contact pressure range from 7.74 kPa to 77.4 kPa, the average observed relative changes in the real and imaginary part of the complex permittivity are -0.31±0.09% and -0.32±0.14% per kPa, respectively. Given these significant changes, this work strongly advises a constant and reported contact pressure along with the dielectric data.

Furthermore, in order to design theranostic applicators, an adequate dielectric tissue model is required. In addition, tissue temperature will heavily influence the dielectric properties and the dielectric model should thus be extended to incorporate this temperature dependence. Therefore, this work measured the dielectric properties of porcine muscle tissue in the 0.5-40 GHz range for temperatures from 20°C to 45°C. According to the measurements, a single-pole Cole-Cole model was presented in which the five Cole-Cole parameters are given by a first order polynomial as function of temperature. The dielectric model closely agrees with the limited dielectric models known in the literature for muscle tissue at 37°C, which makes it suited for the design of in vivo applicators. Furthermore, the dielectric data at 41°C to 45°C is of great importance for the design of hyperthermia applicators. Alongside the dielectric model, a preferred operating frequency range of 2-6 GHz for applicators is suggested.

Given the existing differences in dielectric data between healthy and cancerous colorectal tissue, this Ph.D. thesis proposes a complementary split-ring resonator topology for diagnostic purposes that operates in the aforementioned frequency range. Three designs are manufactured on a 5 mm by 5 mm footprint and are evaluated regarding their sensitivity and penetration depth. A shift in the resonant frequency of 2.544 GHz is observed over a broad permittivity range from 1 to 80 and a penetration depth up to 7 mm is established. Furthermore, a deconvolution-based algorithm is presented that improves the spatial resolution of the sensor. Using a conformal mapping function and the sensor’s specific point spread function, objects smaller than the sensor itself are detected. A 2 mm spatial resolution is obtained according to the Rayleigh criterion, and multiple objects with varying permittivity values are successfully reconstructed.

Next, the complementary split-ring resonator topology is implemented onto a cap-assisted attachment for use throughout a colonoscopy. In this work, a flexible, tubular design with six separate segments on a thin polyimide substrate is presented. The segments expand such that the embedded sensors directly contact the tissue. Given the nature of the flexible design, different dilation levels are possible for varying colon diameters. When the sensor contacts the tissue, a sensitivity of up to 1.487 GHz is observed. Furthermore, multiple sensors are embedded while limiting the number of interconnects through frequency multiplexing. In addition, the bulky vector network analyzer is replaced with a compact and inexpensive measurement implementation without sacrificing measurement accuracy.

Lastly, the presented topology is used in a thermal therapy context. Here, a power amplifier supplies a 1.5 W signal to excite the complementary split-ring resonator. With an electronically controlled heating loop, the sample can be heated up to 50°C with an average heating rate of 0.72°C per second. Moreover, given the two resonant frequencies of the complementary split-ring resonator topology, a dual-mode operation is explored. Here, the first frequency band is used for heating, whereas the second one monitors the resonant frequency that depends on the permittivity of the heated sample. Since the temperature of the sample influences its dielectric properties, and thus the resonant frequency as well, the sample temperature can be derived from the resonant frequency. A positive correlation was measured in the observed temperature range from 20°C to 50°C between the sample temperature and the resonant frequency. Thus, thermal therapy and thermometry are possible while utilizing a single applicator, whereas other existing systems still require an additional thermometry system.