Experimental evaluation of the near-field and far-field heating of focused ultrasound using the thermal dose concept

Highlights

• Movement of a 1.1 MHz transducer for reduction of near and far-field heating.

• Evaluation of six motion algorithms and varied delays for sonications on agar phantom.

• Calculation of thermal dose on near and far-field regions.

• Delay of 30 s required for reducing thermal dose in far-field.

• Selection of optimal algorithm and delay can reduce treatment times.

Abstract

Background

Conventional motion algorithms utilized during High Intensity Focused Ultrasound (HIFU) procedures usually sonicate successive tissue cells, thereby inducing excess deposition of thermal dose in the pre-focal region. Long delays (~60 s) are used to reduce the heating around the focal region. In the present study the experimental evaluation of six motion algorithms so as to examine the required delay and algorithm for which the pre-focal (near-field) and post-focal (far-field) heating can be reduced using thermal dose estimations is presented.

Materials and Methods

A single element spherically focused transducer operating at 1.1 MHz and focusing beam at 9 cm, was utilized for sonication on a 400 mm² area of an agar-based phantom. Movement of the transducer was performed with each algorithm, using 0–60 s (10 s step) delays between sonications. Temperatures were recorded at both near and far-field regions and thermal dose calculations were implemented.

Results

With the algorithms used in the present study, a delay of 50–60 s was required to reduce heating in the near-field region. A 30 s delay induced a safe thermal dose in the far-field region using all algorithms except sequential which still required 60 s delay.

Conclusions

The study verified the conservative need for 60 s delay for the sequential plan treatment. Nevertheless, present findings suggest that prolonged treatment times can be significantly reduced in homogeneous tissues by selection of the optimized nonlinear algorithm and delay.

Introduction

The ablation of malignant tissue during a High Intensity Focused Ultrasound (HIFU) procedure is highly dependent either on the use of robotic devices or electronic steering for navigating the ultrasonic transducer so as to achieve utmost tissue ablation. Motion algorithms following predetermined rules, are utilized for maximal coverage of tissue areas with the majority of cases following a sequential [1], [2] or a spiral algorithm [3]. Successive sonications utilized by sequential movement do not allow cooling of proximal tissue cells, resulting in excess deposition of thermal energy in the pre-focal region (near-field). Near-field heating is a major drawback for a HIFU procedure since it limits the amount of ultrasonic energy supplied to the focal region affecting maximum tissue ablation and treatment time [4], impacting surrounding healthy tissues and inducing unwanted effects such as skin burns [5].

In order to allow diffusion of thermal energy, cooling periods should be introduced between successive sonications thereby reducing near-field heating. This was firstly introduced in 1993 [4] where a time delay of 20 s between sonications substantially decreased near-field heating. The authors [4] concluded that increasing time delay and transducer’s frequency or decreasing sonication time and transducer’s F-number (radius of curvature/diameter) greatly reduced near-field heating. However, later studies [6], [7] consistently show that the aforementioned delay was probably not sufficient. The effect of near-field heating’s dependency on F-number was later confirmed when two transducers having the same nominal frequency and diameter but varied focal length were used [6]. A variation of time delays between 30 or 90 s was needed for inducing the same amount of pre-focal heating among the two transducers, with the higher time delay required for the transducer having increased focal length (i.e. increased F-number). Near-field heating and the shape of the ablation area were investigated by McDannold et al. [7] during in vivo sonications using time delays of 11–60 s monitored by Magnetic Resonance Imaging (MRI). Time delays of up to 40 s induced increased pre-focal heating and necrosis area, with the optimal 60 s delay required for reduction of near-field heating and formation of uniform area of necrosis. Fennessy and Tempany [8] used 80–90 s delays. In the aforementioned studies, sequential algorithm was exclusively used for covering an entire treatment area. An appropriate time delay was allowed between the grid points but without investigating to reduce delay or following another treatment path scan so as to achieve a faster total treatment ablation.

Magnetic Resonance (MR) thermometry has been proposed and used for monitoring the relative temperature increase in the near-field region during volumetric in vivo ablations [5]. Substantial near-field heating and increased necrosis area were noted, with the former linearly related with energy density which can thus be a sign for possible induced necrosis. Although proton resonance frequency shift (PRFS) MR thermometry has since been utilized during in vivo soft tissue sonications for the monitoring of near-field heating [9], its utilization for monitoring near-field heating in fat tissue is unfeasible [10]. To compensate for this, T2-based temperature measurements were investigated and proven feasible for monitoring near-field heating in fat tissue although being much slower than the respective PRFS method [10]. Although near-field heating can be monitored with MR thermometry, undetected necrosis can still be induced if accumulation of thermal energy is not sufficiently accounted for by using appropriate cooling periods [11].

However, introduction of cooling periods between sonications significantly increases overall treatment time. The study by Ji et al. [12] proposed a new way of utilizing cooling periods by exploiting both in silico and ex vivo experiments. The authors used a linear algorithm for sonication of a 4 × 4 grid, where cooling periods were only utilized between each grid line. Additionally, they divided the volume in four 2 × 2 grids and used square movement for the sonication of each subsection, with cooling period introduced after each subsectional sonication. The nature of the square movement resulted in a decreased by 20 s treatment time with a 60 s time delay required for reduction of near-field heating.

Studies have shown that the reduction of near-field heating and consequently increase of the energy absorption in the focal region can also be achieved by using pulsed waves [9], cooling of the transducer with cold water for approximately half an hour prior sonication [13], or exterior tissue cooling through the means of a cooling mat [14]. The latter was utilized for abdominal ablation since higher powers are used so as to compensate for the increased perfusion of organs thereby resulting in higher near-field heating [14]. The use of perfluorocarbon agents such as microbubbles and nanodroplets, has also been proven able in reducing heating in the pre-focal region and thereby increasing temperature deposition at the focal region [15]. Compared to microbubbles, nanodroplets resulted in decreased near-field heating and double energy deposition in the focal region thereby concluding that they can possibly reduce treatment time by a factor of 3 [15].

Despite the fact that phased arrays are not dependent on the motion algorithm but rather on the electronic steering of the focal point for treating large volumes of tissue, the volumetric ablation utilized entails extended ultrasonic exposures resulting in greater induced near-field heating than their non-phased counterparts, thereby requiring longer delays between sonications [7], [16]. Further studies also confirmed that electronic steering of the phased array transducer significantly increases the accumulated thermal dose in the near-field region, compared to mechanical steering [17]. Volumetric ablation was introduced more than a decade ago and electronically maneuvers the focus of the transducer, for sonicating points located in concentric circles of increasing diameter [16]. The method was later improved by development of an algorithm that dynamically controls the ultrasonic duration on each concentric circle through MR thermometry feedback from already sonicated points, thereby resulting in decrease of the near-field heating [18]. The algorithm was further improved through development of a dynamic control of activation or deactivation of the individual ultrasonic elements inducing further reduction of near-field heating [19].

In order to reduce near-field heating, a simulation study [20] and later experimental evaluation [21] for a phased array transducer operating at a frequency of 500 kHz was performed for potential ablation of fibroids. Although the near-field heating of the transducer was reduced, its low operating frequency increased heating in the far-field (post-focal region) since it induces lower attenuation compared to clinically available systems that usually utilize frequencies of 1 MHz [20]. According to the author’s proposal for use on fibroid ablation, the transducer could result in unnecessary potential heating in the spinal area [20].

The introduction of varied cooling times between sonications for reducing the near-field heating effect results in prolonged treatment times particularly for large areas. Previously published works have recommended possible treatment paths without considering the increase in treatment time [6], [17], [22]. A fixed 60 s delay for sequential treatment strategy was used to investigate different planar paths [6] and heuristic paths with the greatest possible distance between sequential small rapid scanning volumes [22]. Moreover, Payne et al. [17] allowed 60 s cooling time for sequential algorithm, however the spacing between the grid points was set at 1 cm which was considered relatively large and led to underrated results. In another study [23], the treatment time was calculated for three suggested path scans using a 1 MHz phased array transducer. The cooling times were selected to retain the cumulative near-field heating beneath 5 cumulative number of equivalent minutes at 43 °C (CEM43°C) for the whole treatment. However, the results of this study [23] were based on simulations models.

The use of conventional sequential movement for treatment, resulted in the utilization of other algorithms (spiral, square) [3], [12] that arrange the spatial distribution of successively sonicated cells in a way that allows diffusion of thermal energy so as to reduce high accumulation of energy in the near-field region and thereby decrease the cooling times used between sonications. Development of novel algorithms that avoid sonications of successive cells in order to result in lower treatment times, stimulated the introduction of new algorithms by Yiannakou et al. [24]. In their study, Yiannakou et al. [24] used simulation for evaluating the induced heating and thermal dose on the pre-focal region with each of the six proposed algorithms. Nevertheless, no temperature data was acquired for supporting the modelling. Varying time delays were introduced between sonications in order to reduce near-field heating and treatment time. The authors reached the conclusion that half of the number of proposed algorithms, significantly reduced near-field heating.

In this paper we present the experimental evaluation of the six algorithms previously proposed by Yiannakou et al. [24], by sonicating an agar-based phantom doped with wood powder [25]. The phantom was used as a soft tissue mimicking material since it presents with the same ultrasonic, thermal and MR properties [25]. The effect of increasing time delay between sonications on the induced temperature in the near and far-field regions of the transducer using each algorithm was examined. The recorded temperature increase was utilized in order to calculate the thermal dose, as defined by Sapareto and Dewey [26], induced in both the near and far-field regions. Although there have been a number of simulation studies [17], [24] examining the thermal dose induced by the respective candidate transducer in the near-field region, existing literature does not include any experimental assessment of the induced dose on either near or far-field. Thereby, the proposed study is novel since it experimentally assesses induced thermal dose on both near and far field regions so as to find the optimal time delay and algorithm. It is worth noting that the results are only valid for the proposed transducer since it has been shown that the deposition of thermal energy in the near-field region is dependent on the transducer’s structural characteristics as well as the sonication parameters [4].