1.

Introduction

Radio frequency (RF) lesioning has been used in functional neurosurgery since the 1950s.¹ With this interventional technique, permanent cell damage is achieved and it must therefore be performed with high accuracy and precision. Clinical applications are, e.g., symptom relief of movement disorders such as Parkinson’s disease^{2 3 4 5} and the reduction of severe chronic pain.⁶ During RF lesioning, Joule heating of the tissue in the proximity of the electrode tip creates the coagulation. To achieve an optimal outcome of the intervention, the localization of the target and the lesion’s size and shape must be considered. In the last few years, improvement of target localization has been achieved by the implementation of increased resolution in different imaging modalities such as magnetic resonance imaging (MRI) and computer tomography (CT) and their integration in preplanning surgical systems.⁷ Furthermore, microelectrode mapping of the neurophysiological activity^{8 9} together with an increased knowledge of the pathophysiology of the basal ganglia¹⁰ are also used to improve target localization.

Knowledge of the lesion size and shape is crucial to the surgical outcome. If the lesion is too large irreversible complications might occur and if it is too small the expected effect may be lost. Previous investigations have shown that the size and shape is influenced by parameters such as electrode configuration and temperature settings.¹¹ A linear relation between target temperature and lesion size has been found by using transparent homogenous albumin as a test solution in combination with a computer-assisted video system for documentation of the clot length, width, volume, and shape.^{11 12} Furthermore, the same electrode tests showed a high correlation with stereotactic thalamic lesions in the pig brain evaluated by MRI and postmortem histology sectioning.¹³ In living tissue, however, other parameters such as tissue type and blood perfusion in the target and its closest surroundings can also influence the size and shape.¹⁴ Up to now, the lesioning process has been clinically studied by observing the patient’s reactions during the surgical procedure as well as postoperatively by investigating the lesion’s localization and size using MRI or CT.¹⁵ Currently, there are no methods that can record the lesioning process intracranially parallel to the coagulation development with the necessary image resolution and scanning speed. The first part during RF surgery is most important to monitor, due to the major coagulation development is occurring during the first 10–20 s of lesioning. Image modalities such as intraoperative MRI, CT, or ultrasound is currently not appropriate for lesion size estimation during stereotactic RF lesioning, due to the relative low image resolution or scanning speed,^{16 17 18} however, there is a vast amount of research ongoing to improve these imaging techniques.

In order to monitor the optical changes in the surrounding media during lesioning, a RF electrode system was modified. Optical fibers were inserted along the electrode to deliver and collect backscattered laser light from the surrounding media. The backscattered photons were converted to an electrical signal, which was processed in order to detect unshifted and Doppler-shifted light before, during and after a coagulation process. The aim of this study was to evaluate the developed system’s performance by using an in vitro model and to investigate if optical variations could be used as markers for size estimation during experimental RF lesioning.

2.

Optical RF Electrode System

A schematic overview of the optical RF lesioning system is presented in Fig. 1(a). It comprises an optical RF electrode, Leksell^® Neuro Generator (LNG) (model 30-1, Elekta Instrument AB, Sweden), a low-power He–Ne laser at 633 nm, laser Doppler detection hardware (photodetector with filters), analog/digital converter and a computer for processing and presentation of collected data. The system has five temporal output signals: temperature, power, and impedance from the LNG, perfusion, and total backscattered light intensity from the laser Doppler unit.

Figure 1

System overview (a), flow model with the revolving disk (b).

2.3.

Evaluation of the Laser Doppler Unit

Measurements were conducted with the RF electrode in a Delrin^® piece of plastic with drilled holes of diameters between 2 and 12 mm, in order to investigate the system capability of detecting nonshifted photons. The optical RF electrode system’s ability to detect Doppler-shifted light and the system linearity were tested in a flow model. A bowl with a 2% microsphere diluted albumin solution was placed on top of a revolving disk, Fig. 1(b). The optical electrode was placed into the solution at a fixed distance from the rotational center. Via the angular frequency and electrode distance from the rotational center, the flow speed at the electrode tip was calculated. By changing the angular frequency, flow speeds between 0 and 9 mm/s were achieved. The solution passed perpendicularly to the electrode tip and the light from the electrode was thus transmitted into the moving solution. A fraction of the backscattered photons from the moving particles was detected, which gave rise to the perfusion signal. The calculated perfusion signals showed a linear relationship with the flow speed between 0 and 9 mm/s (r=0.995, p<0.00001, n=100), Fig. 2.

Figure 2

Linearity of the optical RF lesioning system.

3.

Experimental Setup

4.

Results

The developed optical RF electrode creates coagulations with similar size and shape as the standard RF electrode. An example of the coagulation shape at 80 °C in the standard albumin solution with the two electrodes is presented in Fig. 3. The calculated volume, width and length for the lesions created with the two RF electrodes are presented in Table 1.

Figure 3

Example of two 80 °C lesions created in a standard albumin solution with the optical electrode (a) and the standard electrode (b).

Table 1

Typical signal patterns of a coagulation procedure in the microsphere solution with a heating period of 60 s and a resting period of additional 60 s are presented in Fig. 4. A 10 s base line recording precedes the lesioning. After initiation of the coagulation the DC and perfusion signals display a steep increase, with a perfusion peak after 5 s and a slower DC increase rate after 15 s. The perfusion signal decreases as soon as the albumin protein begins to coagulate but the DC signal flattens out. In this example, a second pronounced peak is seen in the perfusion signal after 18 s. This peak is caused by boiling, often occurring during coagulation at high temperatures (here 90 °C) and is manifested as a bleb formation of the clot.¹¹ The DC signal is only slightly influenced by the boiling of the albumin solution. When the LNG is turned off (marked ended at the 120 s marker), both the DC and perfusion signals reach a stable but different level, compared to the base line.

Figure 4

An example of boiling during coagulation at 90 °C, the signal above is perfusion and the signal below is DC.

An example of the recorded DC and perfusion signal changes during coagulations at 70, 80, and 90 °C is shown in Fig. 5. To emphasize the signal changes pre- and postcoagulation, the DC and perfusion signals were normalized to their corresponding base lines. After a short heating period, the perfusion signal for all three temperatures rose to a high peak value, Fig. 5(a). When the LNG procedure was ended three different stable perfusion levels were detected with all values situated below the base line. Larger clots (higher temperature) showed lower perfusion values, Fig. 5(a). The changes in perfusion levels before and after coagulation were significant for all three of the temperature settings (p<0.001, n=10). The coagulation development can also be seen in the increased DC signals Fig. 5(b). When the LNG was turned off three different stable DC levels were detected with all values situated above the base line. Distinct signal levels were found between the 70, 80, and 90 °C clots. Larger clots (higher temperature) showed larger DC levels. The DC signal changes before and after lesioning were significant for all three temperatures (p<0.05, n=10 for 70 °C and p<0.001, n=10 for both 80 and 90 °C).

Figure 5

An example of perfusion (a) and DC (b) signals during lesioning at temperatures of 70, 80, and 90 °C.

The calculated signal changes, according to Eqs. [2(a)–2(b)] with corresponding standard deviations, were plotted against the created three different volumes in Fig. 6. The changes for the DC values are presented as values above 100% due to the increase of the DC signal during coagulation. There was a high significance (p<0.0005, n=30) between the different signal changes for both the DC and perfusion signals for all tested temperatures.

Figure 6

Relative changes for DC and perfusion signals at 70, 80, and 90 °C.

A linear relationship between the measured coagulation sizes and the calculated signal changes was found with a linear regression model in the albumin solution with microspheres. The correlation between Q _DC and volume, length, and width of the clots, respectively, was at least r=0.88 (p<0.0001, n=30). Also Q _perf was correlated with the clot volume, length, and width. The correlation coefficient was at least r=0.81 (p<0.0001, n=30).

5.

Discussion

In this study, a system for the investigation of optical measurements during radio frequency thermolesioning has been presented and evaluated. A brain lesioning electrode with integrated optical fibers inserted towards an optical window at the electrode tip, was developed. The changes in both static and Doppler-shifted light in the vicinity of the tip during lesioning were investigated. These experiments show that perfusion and DC changes can be used as markers for lesion size estimation in vitro.

The perfusion signal, originates from the backscattered Doppler-shifted light processed according to the laser Doppler theory, and is proportional to the average number of moving scatterers times their mean velocity in the sampling volume.^{19 20} During coagulation of the tissue (here modeled as albumin with moving microspheres) the number of moving scatterers close to the electrode tip is reduced when the clot size is increased. An inverse relationship between the Doppler signal and clot size is thus expected. The sampling depth, however, is one limiting factor. Previous studies have shown that a 90 °C clot created with a 2 mm in diameter electrode does not expand more than 2 mm from the electrode surface,¹³ which is considered to be within reach of the sampling depth of laser Doppler signals in the brain.^{22 23} In addition, Giller and colleagues^{24 25} have performed both phantom studies and Monte Carlo simulations in order to investigate the “look-ahead distance” for an near infrared probe used in diffuse reflectance spectroscopy. Depending on the optical scattering and absorption coefficients of the media and the used equipment the look-ahead distance varied between 1 and 3 mm. These simulations support our results of DC measurements within a 2 mm distance light source to coagulation border. A way of increasing the sampling depth in our application may be to change the used light source to a longer wavelength.

Electrical properties (resistance, delivered current, etc.) for both electrodes used in the first part of the study were measured and had the same values. The comparison between protein clots generated in 37 °C albumin without scatterers by a regular brain lesioning RF electrode and the optical RF electrode showed practically no difference in size and shape. The standard deviation of the volume, width, and length within one temperature setting was small for both electrodes, which implies good reproducibility.

The developed albumin model with microspheres has some features that have to be taken into consideration. A 24 °C microsphere diluted albumin solution was used in the rotating disk setup. The change in temperature between the experiments was performed due to difficulties to heat the solution without causing turbulent flow in the liquid. An agiated solution has a forced convective cooling at the electrode tip which results in slightly smaller coagulations, however, the investigated signal changes were compared to the created coagulations in the model and not the absolute sizes created in the stationary 37 °C experiment. A limitation of the model was a sediment effect of the microspheres due to their low buoyancy, which could affect the Doppler part of the measurements with a slightly increased perfusion base line level. For the experiments performed, the model was found to be adequate to monitor perfusion and DC changes around the electrode tip during coagulation. When the albumin solution starts to coagulate two distinct effects were expected and did occur. The optical reflectivity increases and the perfusion decreases, due to the reduced amount of freely moving scatterers close to the electrode tip. A characteristic signal pattern and a significant change of signal levels pre- and postlesioning between 70, 80, and 90 °C were found. The correlation between lesion size and DC signal changes were slightly higher than for perfusion signals. This implies that the DC signal might be a better choice than the perfusion signal for estimating lesion sizes in vitro. But in vivo in a more turbid media, this result might be different. The DC and perfusion signal represents different properties in the measured volume and is therefore both important.

The presented system is currently evaluated in vivo using an animal model. The next step in the development towards a system for on-line estimation of lesion size after the animal evaluation is adaptation of the system for human use and clinical evaluation. Our hope is that the use of an optical RF electrode will become a valuable tool during functional neurosurgery, both for monitoring the lesion development and to give an indication of the actual created lesion size.