An example of the applicator design is shown in Figure 1.  Ultrasound (US) from a piezoelectric disk was fed via a  stainless steel multistage (three to six) conical velocity transformer (acoustic horn) into a spinal or hypodermic needle.  Several disk, 12.7 - 25.4 mm, and needle, 0.5 - 1.22 mm, diameters were tried in various experiments.   The transformer was glued to the aluminum casing of the piezoelectric transducer with a silicone RTV coating.  A length of the needle was clad in a polyolefin plastic tubing.  An air gap between the cladding and the needle provides for acoustic insulation. The cladding needle system behaves as a waveguide.  The needle 15 mm to 20 mm long exposed tip in acoustic contact with the medium forms an antenna.  US transducers were controlled from a synthesizer through an rf power amplifier. We have used 0.75 - 2.25 MHz frequency range  in the experiments. The experiments showed that ~1 MHz is the optimum frequency for the design of Fig.1.  The typical rms voltage at the transducer required to produce expected temperature elevation varied from 13 to 23 V rms.

It was found early on experimentally [1] confirmed later by calculations [2] that a single applicator can produce required temperature elevation in ~1-cm diameter almost cylindrical volume with the cylinder height slightly more than the antenna's length.  Our collaboration with a neurosurgeon led to a conclusion that there will be a need to use three- to four-applicator array in a typical procedure. Research of a multi-applicator array showed that the region at the required temperature is not only a function of the array geometry, but also a function of a given applicator proximity to the boundary at an ambient temperature. Because of the latter finding, we decided to carry out further research in a specific organ, human brain.

The need for a multi-applicator array helped to solve an important concern of controlling heating inside the organ.  While in our phantom and animal experiments we use several (up to thirteen) temperature sensors, implantation of so many devices into a patient's brain becomes impractical. We re-designed then the applicator mounting a small, 230-µm bead size micro-thermistor as the sensor at the antenna's tip as shown in Fig. 1.  Interrupting sequentially power delivery to the array's applicators [3] we can interrogate about temperature in the heated volume to a precision of about a degree.  This procedure together with FEA computations may enable the heating control.

While our initial computations of heating effects for a single applicator were done analytically, temperature patterns produced by the array lack the cylindrical symmetry observed for a single applicator.  Also, as explained above, we had to develop computational model that reflects complicated geometry of the organ of interest. We decided to use FEA for our computations since this approach provide very accurate solutions.  Our FEA computations are based on a mixed bioheat-transfer/effective- conductivity equation. Using a commercial software we build 3-D FEA models and find transient solutions of the equation.
 

One of the most important questions arising in thermal therapy treatment is local blood vessel cooling. Our interstitial methodology can not be used for heating in vicinity of large blood vessels because of the danger on hemorrhage during implantation.  However, significant, 0.3 - 0.5-mm and larger diameter blood vessels may also play role in local cooling.  Figure 2 shows a detail of an FEA model in which we considered three pairs of adjacent vessels around array applicators. The bar at the bottom of the figure represents 1 cm.  Two inner circles show to scale outside and inside diameter of the applicator's needle.  Colored outlines give lumen of the blood vessels and in this model they were 0.37 and 1.1 mm. Blood vessels with smaller than 0.37-mm diameter didn't affect temperature distribution.  In the calculations we used observed dependence of blood flow velocity on the lumen diameter.