This study was conducted on a total of two swines, in compliance with animal health regulations, Biovivo’s Standard Operating Procedures and the principles of Good Laboratory Practice. The protocol was submitted to the VetAgro Sup Ethics Committee (number 1934).
Study endpoints
The primary study endpoints were accuracy and safety of needle insertion. Needle insertion accuracy was defined as the distance between the needle tip and the centre of the targeted fiducial. Needle inserted within 5 mm to the targeted fiducial defined the procedure as accurate. The safety evaluation was obtained from procedure-related complications assessed by an interventional radiologist who reviewed the post insertion CT and by the clinical evaluation of a veterinarian. Adverse events were defined according to Common Terminology Criteria for Adverse Event 5.0 [9].
The secondary endpoints were feasibility and adjustment of needle placement. Feasibility was defined as the percentage of successful needle insertions, exempt of problems encountered with the medical device tested during the intervention. A successful placement then allows the procedure to be carried on without reiteration of planning and insertion, after guide release and imaging control. Needle placement adjustment was defined as the number of needle repositioning needed to reach the target.
Robotic device
The EPIONE® robotic device (Quantum Surgical, Montpellier, France) consists of a mobile cart that carries a robotic arm bearing a needle guide, a mobile display cart, a patient reference (Quantum Surgical, Montpellier, France) attached to the animal skin, that monitors the patient’s motions including breathing, with the help of a mobile navigation camera (NDI Polaris Vega, Ontario, Canada) (Fig. 1a).
Preparing the animals
Two animals weighing approximately 50 kg were placed under general anaesthesia and received pre-medication with morphine at 0.1 mg/kg (subcutaneous route). Anaesthesia was induced with xylazine at 1.5 mg/kg (intramuscular route) and tiletamine + zolazepam at 3.75 mg/kg (intramuscular route). The animals were then intubated and placed under general anaesthesia using Isoflurane-O2 and injection of curare cisatracurium (0.1 to 0.4 mg/kg by intravenous route) renewed as many times as necessary, in order to constrain the breathing movements of the animals. Each animal was surgically implanted with 8 radiopaque fiducials (4 per kidney) measuring 1 mm × 5 mm (GTP 1820≠2, Geotek Medical, Ankara, Turkey) 1 or 2 days before the experiment. Fiducials were surgically inserted to make sure that the robot could target all fiducials wherever they were within the kidney, and not just those that were percutaneously implantable. At the end of the surgery, the animals were awakened.
CT protocol
All CT scans were performed using a 16-slice Brightspeed CT scanner (General Electric Healthcare, Chicago, IL). Animals were in ventral position and abdomino-pelvic acquisitions were performed with a 35-cm field of view. The scanning parameters were 120 kVp, 150 mA, 0.8 s gantry rotation time, and 0.938 pitch. Scans were reconstructed with a 0.625-mm slice thickness using the ‘SOFT' convolution kernel also provided by General Electric, a standard soft tissue convolution kernel.
Percutaneous robotically guided needle insertions
Needle trajectory planning
A pre-puncture CT volume was acquired under apnoea and loaded into the robotic device software. Then, the needle trajectory was defined in order to reach the centre of a fiducial and to avoid bone and critical structures in the abdomen (Video 1 in Supplementary Material). Among the four fiducials implanted in each kidney, two fiducials were selected in each kidney and thus the procedure was repeated twice per kidney.
Robotically-assisted needle insertion
All procedures were performed by one experienced interventional radiologist (more than 10 years of experience in image-guided procedures). The robot arm was sent to its predefined ‘HOME’ position. The needle guide was attached to the robot arm. Then, the software automatically located the ‘patient’ reference markers in the images. The position of each marker was verified and manually adjusted whenever necessary. A 17-gauge disposable coaxial biopsy needle (Bard Medical, Covington, USA) with the appropriate needle length (from 10.6 to 14.7 cm), i.e., a needle with a length slightly greater than the trajectory length, was selected into the planning software. The robot arm was manually brought close to the area of interest in order to place both the ’patient’ reference and the needle guide within the optical camera’s field of view. Then, the robot arm automatically positioned the needle guide in line with the planned trajectory. After the skin incision was made at the needle entry point, the needle was manually inserted inside of the needle guide and pushed until the end position (Supplementary material: Video 2 and Fig. 1b). The needle was then released from the needle guide and the robot arm was manually moved away from the puncture site.
Needle placement verification
A CT scan image was acquired under apnoea to assess needle placement accuracy. The procedure was deemed complete when the operator judged the needle tip as being ≤ 5 mm of the target. Otherwise, an additional attempt was allowed with the same planning. A third attempt was not permitted.
Safety assessment
Five minutes after needle removal, a contrast-enhanced CT scan of the abdomino-pelvic area was performed 60 s after the injection of 30 mL of ioxehol (Omnipaque®350 mg/mL, General Electrics Healthcare, Chicago, IL) at a rate of 2.2 mL/s. Any procedure-related complication that can be depicted on this CT scan was reported.
Apnoea repeatability test
To verify that the kidney was repositioned at the same location under repeated apnoea, a repeatability test was performed on all the fiducials inserted into the two swines, using the respiratory monitoring module. Apnoea was induced during all CT scan acquisitions and robotically-assisted needle insertions by turning off the ventilation at the end of expiration.
Two successive unenhanced CT scans were acquired during apnoea to assess their repeatability. In-between, animal breathing was resumed. The location of each fiducial was determined on the two CT scans, using Digital imaging and communication in medicine (DICOM) coordinates with a standard DICOM image viewer software (Myrian v2.4, Intrasense, Montpellier, France; https://intrasense.fr/fr/). The coordinates of each fiducial relative to the patient reference were determined and the three-dimensional (3D) Euclidean deviation between the two acquisitions was computed to determine the displacement of the fiducial under repeated apnoea. Apnoeas were considered as repeatable if the displacements of the 8 radiopaque fiducials were below 2 mm in-between the two subsequent CT scans. The displacement of the fiducials under subsequent apnoea was calculated in relation to the ‘patient’ reference in order to strictly reproduce the operating principle of the robotic system.
Statistical analysis
For this preliminary study, no specific sample size calculations were performed. The description of the relevant study variables was performed using, mean, standard deviation, median, min and max for quantitative variables and using frequency and percentage for qualitative variables.
The needle placement accuracy was determined by calculating the 3D Euclidean deviation between the needle tip position and the centre of the fiducial (targeted point). The Myrian® Image viewer (Myrian v2.4, Intrasense, Montpellier, France; https://intrasense.fr/fr) was used to obtain the coordinates of the points in the DICOM images and to evaluate the 3D Euclidian deviation, the lateral deviation and the depth deviation. These deviations were described as quantitative variables.
All measurements were performed blindly by an experienced interventional radiologist. 3D deviation, lateral deviation and depth deviation least square mean estimates (and 95% upper limit) were provided using a random effect model for the animal; p value testing for the inferiority of mean to 5 mm was calculated.
Exploratory analyses were performed to identify potential factors influencing the needle placement accuracy (angulations, skin entry to target distance).
All attempts were prospectively included in the statistical analysis. There was no missing data imputation. All statistical analyses were produced using SAS® 9.4 software (Cary, NC, USA).