2.1.

Fiber Optic Sensor

An ultra-miniature fiber optic high-pressure sensor (Samba Preclin 360 HP, Västra Frölunda, Sweden) has been used (Fig. 1). It consisted of a silicon sensing head with 360 μm OD mounted on an optical multimode fiber with $\sim 400\mu \text{m}$ OD.

Fig. 1

Illustration of the Fabry-Perot sensor at the end of the lead fiber/cable (the sensing head and the nearby lead fiber were coated with a radiopaque material, which allowed knowing the position of the sensor inside the body).

Prior to ex vivo and in vivo experiments, the sensor was left at room temperature for $\sim 10\min$ in a solution of 4% Cidezyme (CIDEZYME^® Enzymatic Detergent Solution, Johnson & Johnson, Medical Inc., Irvine, California) in water and rinsed afterward in distilled and deionized water.

2.2.

Interrogation Unit

A purpose-built interrogation unit connected to a portable computer was used to interrogate the sensor. The basic functioning of the system is depicted in Fig. 2.

Fig. 2

Schematic representation of the interrogation unit and of the basic functioning of the system used to interrogate the sensor.

As can be observed, the power of the incoming light of a superluminescent diode was split 50/50 at the optical coupler/splitter. In such a way, part of this light was used as the signal of reference (at $\sim 1300\mathrm{nm}$) in order to account for source power fluctuations (measured in μW) and the remaining light was used to interrogate the sensor. At the sensor head, the light is back-reflected to the coupler/splitter and the corresponding power measured at the optical power meter (Fig. 2). The output signal $(\text{sensorsignal}/\text{referencesignal})$ was measured at a sampling rate of 17 Hz. A GPIB-USB controller (Prologix, LLC, Washington) was used to allow communication between the PC and the optical power meter (Fig. 2).

2.3.

Sensor Calibration and Data Acquisition

A purpose-built pressure device was used for calibration of the sensor (Fig. 3). The functioning of the pressure device consists of rotating the screw that is connected to the syringe plunger pushing the water inside the syringe into the acrylic chamber (previously filled with distilled water). This action increases the pressure and the rotation in the opposite direction decreases it. On both sides of the acrylic chamber a bolt and nut with passing holes allowed insertion of the sensor into the acrylic chamber (Fig. 3). The sensor was guided through the hole by means of a hypodermic needle that was retracted after sensor insertion. To seal the passing holes, a septum of silicone located in between the bolt and nut was used. Using the previous setup, three increasing and decreasing pressure cycles from 0.0 to 14.0 bar (step of 0.5 bar) were performed for sensor calibration. A manometer intended for medical applications (WIKA 111 series; EN 837-1) with a pressure range from $-1$ to 15 bar and accuracy class of 1.6 was used for pressure readings. Once the pressure is adjusted, the manometer indicator stays static, allowing signal readings without any type of screw/pressure adjustment being necessary for the measured range. A LabVIEW routine was implemented to control data acquisition and store an array of values ($n=10$) of the output signal at each calibration step. The averaged calibration data points were plotted and the calibration coefficients were calculated using fitting functions. These coefficients were used in another LabVIEW routine that was implemented for pressure readings during ex vivo and in vivo experiments.

Fig. 3

The pressure device used to calibrate the sensor.

The maximum error introduced by hysteresis of the sensor was calculated as a percentage of the maximum pressure used during calibration (14.0 bar). Maximum sensor drift during a measurement period of 30 min was also calculated for pressures of 0.0, 7.0, and 14.0 bar.

2.4.

Ex Vivo and In Vivo Experiments

Ex vivo and in vivo experiments were conducted by skilled veterinarians at the facilities of the Veterinary Hospital of the University of Évora. Experiments were authorized by competent national authorities and conducted according to the guidelines for animal care of the Federation of Laboratory Animal Science Association. Ex vivo experiments were useful to decide about the most appropriate in vivo surgical approach and to test the whole system and sensor performance prior to in vivo measurements.

For in vivo measurements, the Samba sensor was implanted in the fifth lumbar intervertebral disc (IVD) of a four-year-old female merino ewe with 45 kgf body-weight, under general anesthesia [Fig. 4(a)]. The animal was maintained on a radiolucent table in a lateral right recumbence position. Endotracheal intubation was performed and the anesthesia was maintained through isoflurane (2 to 3%) in oxygen with spontaneous ventilation. The heart and respiratory rates were controlled [Fig. 4(a)].

Fig. 4

(a) The anesthetized sheep was maintained on a radiolucent table in a lateral right recumbence position under control of heart rate and respiratory rate. (b) Implantation of the Samba sensor into the nucleus pulposus of the intervertebral disc under fluoroscopic control.

A dorsolateral transforaminal approach into the center of the NP, similar to that used in discography and percutaneous nucleotomy, was followed for sensor implantation.³² The lumbar region was prepared for needle puncture with a povidone-iodine solution. A guiding needle was inserted percutaneously in the dorsolateral intervertebral disc space under fluoroscopic control [Fig. 4(b)]. Once positioned in the NP, the stylet point of the needle was taken out and the pressure sensor was passed fully through the cannula and introduced into the NP. The cannula was retracted from the IVD and data collection started.

3.1.

Sensor Calibration

The average results of the three calibration cycles are presented in Fig. 5. The differences on the output signal between consecutive steps of calibration of 0.5 bar were calculated. On average, these differences were of $(5.754\pm 0.472)\times {10}^{-4}$. The mean standard deviation of each step of calibration was $(2.022\pm 0.262)\times {10}^{-4}$, which represents $\sim 35\%$ of the average difference between consecutive steps of calibration. The previous data were used to calculate the calibration coefficients using a linear regression model (Fig. 5).

Fig. 5

Calibration points of three pressure calibration cycles. A detail on the average of the differences between consecutive steps of calibration was shown. Y-error bars represent one standard deviation (SD). A linear regression model was applied to calculate the calibration coefficients.

Maximum hysteresis of the sensor was 0.46%. Maximum sensor drift during a measurement period of 30 min for pressures of 0.0, 7.0, and 14.0 bar was $\pm 0.002$ bar.

3.2.

In Vivo Experiments

The percutaneous approach under fluoroscopic control seems to be the adequate technique for animal experiments and to ensure a transition to human in vivo applications. Moreover, compared to an open approach, where inner organs and tissues are exposed, the technique is less invasive. In fact, in the present study, the animal was able to recover and released to its natural environment in $<4\mathrm{h}$. The complete operation from the beginning to the end of anesthesia lasted for $\sim 2\mathrm{h}$.

The intradiscal pressure variation pattern observed during spontaneous breathing under general anesthesia was observed once the sensor was in situ. A section of this pattern was plotted in Fig. 6.

Fig. 6

Intradiscal pressure pattern in the fifth lumbar disc during spontaneous breathing. Average pressure and time were detailed as well as maximum and minimum pressures.

On average, the signal periodicity was $2.81\pm 0.12\mathrm{s}$ (time peak to peak), which corresponds to $\sim 21.30\pm 0.12$ pressure cycles per minute (Fig. 6). The previous rate was similar to the mean respiratory rate under spontaneous ventilation, which was 20.5 breaths per minute. Sato et al.¹⁰ found that the pressure wave pattern was synchronized with the number of respirations but did not present the corresponding rates. The effects of breathing rate and volume on disc pressure were studied by Keller et al.³⁰ in lumbar discs of anesthetized pigs. In that case, instead of promoting spontaneous ventilation, a ventilator was used to control the previous parameters. Breathing had a significant effect on the intradiscal pressure, which seems to decrease with breathing rate and increase with breathing volume.³⁰

In the present study, pressure fluctuations ranged between 2.31 and 3.45 bar with a maximum amplitude of 1.14 bar (Fig. 6). On average, the resting intradiscal pressure corresponding to lateral right recumbence position was $2.78\pm 0.28$ bar (Fig. 6). Few studies have registered the possible effect of breathing on intradiscal pressure. Moreover, in most of them, the phenomenon was only presented graphically along with the calculation of the mean resting (physiologic, intrinsic, or baseline) pressure.^10,24,26,27 In the previous studies, the mean resting pressure ranged from 0.7 (Ref. 26) to 2.0 bar.²⁷ These pressures were measured in anesthetized pigs and are lower than the mean resting pressure registered in the present study (2.78 bar). On the other hand, higher mean resting pressures were also reported. For example, the resting pressures found in rabbit lumbar discs ranged between 2.2 and 4.2 bar (mean was 3.6 bar).²⁸ In the study of Nachemson and Morris,⁷ the mean pressure obtained for human subjects in the reclining position was 5.4 bar (ranging between 1.4 and 8.3 bar). Finally, in thoracic discs of human subjects, which have the same kyphotic curvature as the lumbar spine of a sheep, the mean resting pressures found were closer to those of the present study, ranging from 2.0 to 3.4 bar.¹³ In the previous study, it was also found that the pressure depends on the resting position and disc level. For example, for the same disc levels (T9 to T10, T10 to T11), the resting pressure was higher for the lying on side position (3.0 bar) than in the lying prone position (2.0 bar). Ekström et al.²⁴ also suggested the resting pressure can be influenced by the pre-tension in the ligaments and the AF, being also highly dependent on the angulation of the vertebrae. The above factors may explain the differences found between studies. While some of them seem difficult to control, such as the pre-tension in the ligaments and the AF, further research seems necessary to better understand these phenomena. Using FOS also seems an excellent contribute to study them because they are minimally invasive and should not affect the natural behavior of the IVD.