Magnetic resonance image–guided focused ultrasound robotic system for transrectal prostate cancer therapy

Abstract Background A magnetic resonance image (MRI) guided robotic device for focussed ultrasound therapy of prostate cancer (PC) was developed. The device offers movement in 5 degrees of freedom (DOF) and uses a single‐element transducer that operates at 3.2 MHz, has a diameter of 25 mm and focuses at 45 mm. Methods The MRI compatibility of the system was evaluated in a 1.5 T scanner. The ability of the transducer to create lesions was evaluated in laboratory and MRI settings, on ex vivo pork tissue and in vivo rabbit thigh tissue. Results Cavitational and thermal lesions were created on the excised pork tissue. In vivo experiments proved the efficacy of the system in ablating muscle tissue without damaging intervening areas. Conclusions The MRI compatible robotic system can be placed on the table of any commercial MRI scanner up to 7 T. The device has the ability of future use for transrectal focal therapy of PC with the patient in supine position.

Focal therapy including laser ablation, cryotherapy and high-intensity focused ultrasound (HIFU) has been developed for low or intermediate risk patients so as to treat only the malignant tissue leaving the benign tissue (urethra, neurovascular bundles) unaffected. 4 It offers an alternative middle-point solution between whole gland radical treatment and AS. Laser ablation was introduced in 1990 for the treatment of benign prostate hyperplasia 5

(BPH) and
has since been used in several clinical trials for the treatment of PC with promising results. 6,7 Cryoablation is a more frequently used technique, introduced in 1966 for the treatment of BPH, 8 and later used for prostatic neoplasms 9 although, with minor tumour ablation and severe complications. Advancements in the technology have since resulted in widespread use with high survival rates and low complication rates. 10 HIFU exists as a technology since 1940 and uses high-intensity ultrasound waves focused on a single point to cause high increases in tissue temperature. This causes cell necrosis thus disrupting their infinite neoplastic proliferation. Ultrasound (US) and MRI are used during treatments to provide anatomical image feedback.
Commercially, there are two U.S.-guided FDA-approved HIFU systems for PC treatment: Sonablate (Sonacare) and Ablatherm (Edap-TMS). Ablatherm incorporates two individual transducers operating at 3 and 7.5 MHz used for therapy and imaging respectively, 11 while Sonablate uses a single transducer operating at 4 MHz for both therapy and imaging. 12 Additionally, differences in patient position, treatment protocols and planning exist between the two devices. 11,12 The two devices have been extensively used in clinical trials to prove their efficacy, with a majority of them performed using the Ablatherm device. The first use of HIFU on malignant PC was done in 1992 on ablating Dunning R3327 adenocarcinoma on rats using an Ablatherm prototype. A complete ablation was achieved in 64 % of the cases, resulting in higher efficacy rates than other modalities reported in literature. 13 Madersbacher et al. 14 were the first to use the Sonablate device on 29 human prostates before RP, where they concluded that the ablation of 20 ml of prostatic tissue was safe without unwanted changes in the intervening rectal wall. While HIFU is not recommended for high-risk patients, when combined with transurethral resection of the prostate (TURP) it can offer an alternative treatment 15. The adverse side effects induced by the two systems include fistula and urethral stenosis and incontinence, with the former rarely being reported [16][17][18][19] and the latter more commonly being indicated. 16 Incontinence rates slightly increase after multiple HIFU treatments; 20 however, significant improvements in bladder function are achieved 6 months posttreatment. 21 When compared to brachytherapy, 5-year cancer-specific and survival rates do not differ between the two modalities, with HIFU achieving PSA nadir at a shorter time. 22 Cancer specific survival rates after HIFU treatment of 100% and 98% at 5-year and 8-year respectively have been noted. 23 The use of MRI as a guidance modality optimises the efficacy of tissue ablation by providing real-time feedback of in situ temperature increase as well as providing higher tissue contrast. Robotic systems used inside MRI environments need to be constructed of materials that do not result in hazardous projectiles due to the high magnetic field, do not interfere with image acquisition nor is their functionality affected by the strong electromagnets. The majority of robotic systems are actuated by hydraulic, piezoelectric or pneumatic motors. Piezoelectric motors are advantageous due to their small size however, their motion is obstructed during imaging, contrary to pneumatic motors which are the most MRI compatible with no loss in signal to noise ratio (SNR). 24 Nevertheless, the source of pressure required for hydraulic and pneumatic motors demands large and complex robotic systems.
The first MRI guided HIFU robotic system was patented by General Electric in 1993 and was developed with hydraulic actuators. 25 Thenceforth, there has been a gradual evolution of MRI compatible HIFU robotic systems for a variety of applications. 26 The majority of those systems utilize piezoelectric actuators, firstly introduced by the Israeli company Insightec 27 to compensate for accuracy and MRI interference problems induced by hydraulic motors. Robotic systems for MRI-guided focussed ultrasound (MRgFUS) treatment of PC were motivated by MRI compatible systems for other prostate interventions such as biopsy and brachytherapy. The majority of the systems offer motion in more than 3 degrees of freedom (DOF), with some of them employing piezoelectric actuators for both biopsy and brachytherapy 28,29 or pneumatic actuators for sole brachytherapy use. 30 The widespread use of piezoelectric motors has led to the recent development of 6 DOF robotic system for both biopsy and brachytherapy as well as laser ablation of the prostate. 31 However, the aforementioned systems cannot be used for focused therapy since they do not provide the spatial requirements for addition of water coupling necessary for ultrasound propagation.
There are two novel Conformité Européenne (CE)-marked MRguided ultrasound systems for the treatment of PC. The ExAblate (Insightec) system 32 offers a transrectal probe with 990 phased array focused elements for PC ablation while the TULSA-PRO (Profound Medical) offers a transurethral probe with a linear array of 10 single unfocused elements. 33 The ExAblate system was first evaluated in 2009 in the preclinical setting on canine prostate model, 34 while the first clinical experience on 5 patients with PC before RP was performed in 2012. 32 While no thermal necrosis was observed in the preclinical study due to limitations on dog size, histopathology specimens revealed complete necrosis at the sonication site in the clinical study. Further studies confirmed device efficacy, with minor complications. 35,36 Clinical trials have also been performed with the TULSA-PRO achieving low PSA levels but with increased rates of haematuria and urinary tract infections 33   The X axis has a motion range of 50 mm. Figure 1 shows the computer-aided design (CAD) of the X axis. Motion is achieved through a pinion gear attached to the ultrasonic motor through a motor holder. The pinion gear was coupled with two spur gears each attached to a jackscrew mechanism. The X axis jackscrews were coupled to the Θ axis motor holder. The use of two jackscrews enables the application of uniform force on both sides of the Θ axis.
The Θ axis was the most complex mechanism of the device since it uses a two-stage planetary gear mechanism. Figure 2A shows the cross section of the motion mechanism of the axis whereas Figure 2B shows the CAD design of the complete Θ axis. A sun gear is coupled to the motor and placed at the centre of the mechanism. Two plan-   Figure 3B shows the rear view whereas Figure 3C

| HIFU system
The system consists of a RF generator (AG Series Amplifier, model

| MR compatibility
The transducer and robotic device were individually evaluated for MR compatibility in a 1. The MR compatibility was evaluated by measuring the SNR utilising a method in the National Electrical Manufacturers Association (NEMA) standard. 40 Two images of the phantom (agar or MR) are acquired under the same conditions. The SNR was calculated by dividing the signal of the first image (S image1 ) with the standard deviation of the pixel-by-pixel difference of the two images (SD| image1-image2 |) as shown in the following Equation (1): The compatibility of the transducer and robotic device was was taken with only the agar phantom in the MR bore. Images were then acquired for both amplifier and transducer deactivated, for activated amplifier and deactivated transducer and for both amplifier and transducer activated. Hence, the standard deviation of the F I G U R E 2 (A) Cross section of the Θ axis motion mechanism. (B) CAD design of the complete angular Θ axis. CAD, computer-aided design difference between the baseline, and images taken in each condition was found, and thus SNR was calculated. For the T2-W FRFSE sequence, activation of the transducer was not followed, since this sequence is not intended for MR thermometry.
Concerning the robotic device, the baseline image was taken with only the MR phantom in the bore. Images were then obtained for the robotic device in the bore (no cables attached), the robotic device wired but with deactivated electronic system, and the connected device with powered electronic system. The SNR was calculated for each of the aforementioned conditions.
Given that all components require the use of electricity for activation, the device is classified as MRI-conditional according to the American Society for Testing and Materials (ASTM) standards (F2503, F2052, F2213, F2182, F2119).

| Laboratory and MRI ex vivo evaluation of the FUS system
The

| MR thermometry
MR thermometry data were produced using the proton resonance frequency shift method 41 which relates the measured phase shift with the change of temperature (ΔT). This relationship is given by: where φ(T) and φ(T 0 ) are the phases at starting and final temperature T and T 0 respectively, γ is the gyromagnetic ratio, α is the proton

| In vivo evaluation of the FUS system
The   Figure 7 shows the MR images acquired using EPI sequence for different configurations of the robotic device while Figure 8 shows the SNR as calculated from the MR images of the T2-W FRFSE, FSPGR and EPI. The SNR was not affected upon introduction of the robotic device in the bore, for all three sequences.

| MR compatibility
There was an SNR reduction upon connection of the robot, with FSPGR resulting in the greatest decrease. On the other hand, EPI resulted in the largest SNR reduction during activation of the electronic system. Figure 9 shows the thermocouple-measured temperature change versus the depth in the ex vivo tissue. The expected focus of the transducer was at 20 mm. However, there was a large increase of temperature recorded at the tissue interface (5 mm distance). Sonication was then performed at an acoustical power of 30 W for a sonication time of 60 s. Figure 10 shows the rate of change of temperature at the focus. This sonication created a lesion at the interface. Figure 11 shows the 12 mm wide and 13 mm long lesion formed at the interface.

| Creation of discrete lesions on ex vivo tissue
Lesions on ex vivo porcine were created in a laboratory environment using movement of the transducer. Figure 12 shows   Figure 14A shows the PD MR image with fat suppression obtained on coronal plane. Figure 14B shows the sliced tissue with the lesions formed on a plane parallel to the ultrasonic

| In vivo evaluation of the FUS system
Sonication of the right thigh was executed at acoustical power of 22.5 W for a sonication time of 30 s at 2 cm focal depth. Figure 15A shows the MR temperature map acquired on coronal plane while   Figure 17A shows the lesion formed on a plane perpendicular to the beam while Figure 17B

| DISCUSSION
A MRgFUS robotic device with a transrectal probe has been developed for localized therapy of PC. The proposed system has been an evolution of previous robotic prostate devices developed by some researchers from this group [37][38][39] and has been inspired by existing FUS devices featuring an endorectal probe. 11  diameter of the probe was reduced compared to previous designs [37][38][39] and therefore the pressure to the rectum is reduced. Furthermore, motion along the axis of the rectum (X) is achieved through two jackscrew mechanisms compared to the single mechanism previously used. 39 Consequently, there is a uniform force application, resulting in a smoother and more accurate motion. Moreover, addition of the enclosure permits placement of the whole device in a sterilization chamber for disinfection between procedures. Taking into consideration the materials used, ethylene oxide sterilization is recommended to achieve the recommended sterility assurance level of 10 -6 . 42 The use of radiation is not recommended due to possible colour degradation nor is steam sterilisation due to deformation of the ABS enclosures attributable to the high temperatures used in autoclaves.
In future designs, the transrectal probe can be designed to be detachable so as to allow for easier sterilization procedure.
The device was tested in a 1.5 T scanner for MR compatibility of the transducer and robotic system by measuring the SNR for a variety of activation conditions. The SNR was measured for T2-W In a clinical setting, pre-treatment, MR images will identify the tumour location. Therefore, based on this information, the physician will insert the probe accordingly. During the operation, the physician might manually correct this depth based on MR images.
Most likely, due to the long range of the X axis, it is possible that targeting of the tumour is achieved without any manual correction.
In case the patient moves during a future clinical trial, a sensor might be needed to interrupt the treatment. A similar approach is applied in radiotherapy. The proposed device was designed ac-

ACKNOWLEDGEMENT
The study was co-financed by the European structural and investment funds and the Republic of Cyprus through the Research and GIANNAKOU ET AL.