Real-time in vivo monitoring of intraocular pressure distribution in the anterior chamber and vitreous chamber for diagnosis of glaucoma

Abstract

Glaucoma causes irreversible vision loss due to optic nerve damage and retinal cell degeneration. Since high intraocular pressure (IOP) is a major risk factor for glaucoma development, accurate IOP measurement is crucial, especially intravitreal IOP affecting the optical nerve and cells. However, conventional methods have limits in selectively and directly detecting local retina pressure. Here, we present continuous measurements of local IOP values in the anterior chamber and vitreous chamber of living animals using minimally invasive probes with pressure-sensitive transistors. After inducing glaucoma in animal models, we compared the local IOP distribution between normal and glaucomatous eyes. We also compared IOP values detected in the cornea using tonometry measurements. Our findings revealed that glaucoma induced higher IOP in the vitreous chamber than in the anterior chamber, indicating that measuring IOP in the vitreous chamber is key to the glaucoma model. This progress offers future directions for diagnosis and treatment of glaucoma.

Implantable probe-type transistors monitored intraocular pressure in the eye’s anterior and vitreous chambers in vivo.

INTRODUCTION

Glaucoma is an optic neuropathy that results primarily from damage to the axons of retinal ganglion cells (RGCs) as they exit the eye at the optic nerve head (ONH). The laminar region of the ONH, known as the lamina cribrosa (LC), is a major site of RGC vulnerability and the location of RGC axonal injury (1). Intraocular pressure (IOP) induces the deformation of the LC, and these changes promote damage to axons and their cell bodies by various mechanisms, including blocking the axonal transport and reducing the diffusion of nutrients from capillaries inside the laminar beam to the adjacent axons (2). Since IOP is the only modifiable biomarker of glaucoma (3, 4), accurate measurement of IOP is essential for the early diagnosis of glaucoma and the implementation of appropriate treatment strategies.

Among the various methods for IOP measurements, including smart contact lenses, Goldmann applanation tonometry is considered to be the gold standard (5–13). These methods, such as tonometry and smart contact lenses, indirectly calculate the values of IOP by measuring the deformation of the cornea (14–17). Thus, occasional misalignment of externally applied forces on the surface of the cornea, as well as discrepancies in the sizes of contact lenses with ocular curvature, may result in miscalculated IOP values. In addition, these methods are limited in selectively measuring local IOP distribution in the anterior chamber and the vitreous chamber. To better illustrate the change in IOP, Fig. 1A shows the structure of an eye and the flow of aqueous humor. The aqueous humor, which helps in maintaining the IOP, is generated in the ciliary body and it is drained through the anterior chamber into the trabecular meshwork (3, 18). Disruptions in this pathway can lead to an imbalance between the generation and the release of the aqueous humor (4), resulting in its accumulation and increased levels of IOP. Elevated IOP can cause damage to both the retina and the optic nerve (19). However, note that the cornea is located at the front of the eye, while the optic nerve is situated at the back. In studies involving rabbits and pigs (20), tonometer-measured IOP is well correlated with the IOP measured in the anterior chamber, but it tends to be underestimated (21). Moreover, some studies have reported the difference in IOP between the anterior chamber and the vitreous chamber, but these studies were focused on enucleated eyes (22, 23). Therefore, there is a need to confirm these differences in vivo. The biomechanical characteristics of the cornea and the sclera can vary substantially among individuals. For instance, in conditions such as pseudoexfoliative glaucoma, differences in scleral stiffness can affect IOP notably (24). These variations in biomechanical properties can affect the accuracy of the conventional methods in measuring the actual IOP applied to the optic nerve. Therefore, for an accurate diagnosis of glaucoma, it is very important to obtain real-time, in vivo local distribution of the IOP measured in the anterior chamber and the vitreous chamber, as well as the IOP values at the cornea.

Fig. 1. Probe-type pressure sensor for monitoring IOP distribution.

(A) Schematic illustration of eye structure and IOP generation due to the flow of the aqueous humor. (B) Schematic layouts of the probe-type pressure sensor. (C) Photograph of the probe-type pressure sensor. Scale bar, 5 mm. (D) Optical micrograph of the pressure-sensitive transistor. Source/drain electrodes were patterned on both sides of the isolated Si channel (left). The transistor was completed by assembling a dielectric layer and gate electrode (right). Scale bars, 100 μm. (E) The transfer characteristics of the transistor (V_D = 1 V). (F) The output characteristics of the transistor (V_G = −10 to 50 V). (G) Schematic illustration of IOP sensing by a probe-type pressure sensor in the vitreous chamber. PDMS, polydimethylsiloxane.

A highly sensitive and minimally invasive device also may be required to monitor IOP using an implant. While many highly sensitive pressure sensors are available, most of them are designed for monitoring external physical signals that occur outside the body (25–30). However, due to their bulky size, these sensors are not suitable for minimally invasive implantation into the body (31). Previous attempts have been made to insert pressure sensors into organs inside the human body, such as the brain or liver (32–34). However, categorizing these sensors as minimally invasive is misleading since they are several millimeters in size, and when they are used for monitoring physical signals in the human eye, potential damage to the eye and measurement errors can occur.

To address these limitations, we have developed a minimally invasive, probe-type pressure-sensitive transistor for selectively monitoring the local distribution of IOP in specific areas of the anterior chamber and the vitreous chamber. This field-effect transistor (FET) can be used as a pressure sensor to detect IOP directly, and it is designed in the shape of a sharp needle to minimize damage when it is inserted into the specific eye chambers of living animals and to reduce its impact on IOP distribution. This pressure-sensitive FET exhibits high sensitivity and fast response time. Moreover, its fine geometry minimizes insertion-related damage and ensures reliable IOP monitoring in vivo. Using this pressure sensor, we conducted real-time monitoring of the local IOP distribution in the anterior chamber and the vitreous chamber of live animals. Also, we induced glaucoma through cauterization and compared the differences in the IOP distribution between normal animals and animal models with glaucoma. These IOP values that were monitored in these two chambers using our pressure sensors also were compared with gold standard tonometry measurements that detect IOP in corneal deformation. With the induction of glaucoma, the vitreous chamber experienced relatively higher IOP than the anterior chamber. This indicated that measuring IOP in the vitreous chamber is key to the glaucoma model, and previous methods of calculating IOP from corneal deformation may have limitations in accurately measuring the IOP in the retina in the glaucoma model. Our study involved monitoring the IOP of normal animals and acute glaucomatous hyaluronic models, and these results can provide valuable insights for accurately diagnosing glaucoma.

RESULTS

Fabrication of an eye-implantable probe with a pressure-sensitive transistor

As illustrated in Fig. 1B, the eye-implantable probe with a pressure-sensitive FET is composed of two panels of biocompatible polyimide. A single-crystalline silicon layer (thickness: 205 nm) was used as the FET channel (channel length: 70 μm, width: 10 μm) and transferred from a silicon-on-insulator (SOI) wafer onto the bottom polyimide (PI) panel (thickness: 25 μm) using a polydimethylsiloxane (PDMS) stamp. Before this transfer, an SU-8 layer (thickness: 1.5 μm) was prespun as an adhesive on the bottom PI panel. After transferring the Si channel, source/drain electrodes (Cr/Au, 2 nm/300 nm) and interconnects were deposited using an e-beam evaporator, and then they were photolithographically patterned. Successively, an elastomeric partition spacer (thickness: 50 μm) of PDMS to define a local air gap was coated to form the air dielectric layer of this FET. Separately, the gate (Cr/Au, 2 nm/300 nm) and interconnects were deposited and patterned on the other PI panel (thickness: 25 μm). This top-gate electrode was brought into conformal contact with the PDMS layer and fully covered the top of the air-dielectric layer by laminating these two PI panels. Sealing the entire sidewalls of the resulting device using a biomedical-grade elastomer (SILASTIC MDX4-4210, Dow Corning) completed the fabrication of the eye-implantable probe with a pressure-sensitive FET. The fabrication process is described in detail in Materials and Methods, and it also is illustrated schematically in fig. S1. The height of the air gap was determined by the thickness of the elastomeric partition spacer between the Si channel and the top gate, and it was decreased by applying compressive pressure with increasing capacitance of the metal-air-channel structure. This pressure-sensitive capacitance change enables the individual FET to act solely as a single tactile pressure sensor.

Figure 1C shows a photograph of the resulting eye-implantable probe. This probe was 190 μm wide and 110 μm thick, and the pressure-sensitive FET was positioned at the tip. Figure 1D presents an optical micrograph of this FET. In the figure, the left inset shows the source/drain electrode with the channel, and the right inset shows their assembly with the gate electrode. Figure 1 (E and F) plots the transfer and output characteristics of this FET, respectively (without applying pressure). The mobility of the device was calculated as ~520 cm² V⁻¹ S⁻¹, and the on/off ratio and threshold voltage were 4.3 × 10³ and 12.3 V, respectively. The calculations of these characteristics are described in the Supplementary Materials. As plotted in fig. S2, this air-dielectric FET displayed negligible hysteresis due to the clean interface between the channel and the gate, which was suitable for rapid and reliable responses.

As illustrated in Fig. 1G, this long sharp probe with a pressure-sensitive FET was implanted directly into the anterior chamber or vitreous chamber of an eye in vivo. The insertion of this probe was performed after piercing a small hole in the eyeball using a 26-gauge needle. The outer diameter of this needle was only 474 μm, and the cross-sectional area of our probe (width: 190 μm, thickness: 110 μm) was small enough to be inserted through this hole. After this probe was inserted into the anterior chamber or the vitreous chamber (close to the surface of the retina), the hole in the eyeball was sealed completely by a tissue adhesive (Vetbond, 3M). The tissue adhesive rapidly cured within seconds. Following the sealing process, we allowed a stabilization period of 10 min to regulate the IOP before conducting the experiment. Then, we compared the change in IOP using a tonometer (ICare Pro, Tonolab, ICare) before the insertion of this probe and after stabilization to confirm that there was a negligible difference in the IOP before proceeding with the subsequent experiment. The small size of the hole and additional sealing using a tissue adhesive effectively prevented any leakage of the aqueous humor. The tip of the probe that was inserted was sufficiently small compared to the ocular volume, and its position was stationary without its bending inside the eye due to the isotropic intraocular forces caused by the aqueous humor inside the anterior chamber or vitreous chamber. Then, the pressure-sensitive transistor operated after connecting it to the source meter and power supply for measuring the changes in the drain current and in the IOP (drain voltage: 1 V, gate bias: 20 V). During the measurement of IOP using this probe, the animals were anesthetized by injection of ketamine hydrochloride and xylazine. For example, the rats were anesthetized via intraperitoneal injection (ketamine hydrochloride: 75 mg kg⁻¹, xylazine: 10 mg kg⁻¹), and the rabbits were injected intramuscularly with (ketamine hydrochloride: 50 mg kg⁻¹, xylazine: 5 mg kg⁻¹). In addition, the rabbits were deeply anesthetized by inhalation of 3% isoflurane for stable measurements.

Pressure-sensing performances and in vitro IOP measurements

Figure 2A illustrates the sensing mechanism of the pressure-sensitive FET. When the IOP caused by the vitreous humor in an eye presses this FET, the thicknesses of the air gap and the PDMS partition spacer decreased as the capacitance of the gate-air-Si channel structure and the drain current (I_D) increased. Because of the elastic property of PDMS, this FET can act as a single pressure sensor solely with no integration of an additional component layer. To evaluate the pressure-sensing characteristics, a well-defined load was applied in the pressure range from 1.3 to 8 kPa using an experimental setup that consisted of a motorized z-axis stage (Mark-10 ESM 303) and a force gauge (Mark-10 M7-20). Figure 2B presents the real-time measurement of the relative change [ΔI_D/I₀ (%)] in I_D at V_G = 20 V and V_D = 1 V under different magnitudes of pressure, where I₀ is the current at zero Pascal and ∆I_D = I − I₀, which denotes the variation of the I_D during the stepwise pressure loading. The pressure increased stepwise from 1.3 kPa (equivalent to 10 mmHg) to 8 kPa (60 mmHg), with 60 mmHg being the maximum range that the human IOP typically exhibits. This graph shows that the response to the applied pressure was distinguished clearly as a step-like response. Figure 2C is a plot of the relative change in I_D with respect to the applied pressure, and ∆I_D/I₀ increased linearly within the pressure range. The sensitivity of this pressure sensor was obtained from the slope of this plot, and it was expressed as [I_D/I₀ (%)]/∆P, where ∆P denotes the applied pressure. The sensitivity of the pressure-sensitive FET was determined to be 0.44% kPa⁻¹ (equivalent to 0.059% mmHg⁻¹) in the pressure range below 60 mmHg. The signal-to-noise ratio (SNR) is defined as SNR = μ/σ, with μ representing the average value of relative I_D when 1-mmHg pressure is applied, and with σ indicating the SD of the noise levels when the pressure is released. The SNR measurement of our sensor at 1 mmHg was 32 (μ = 1.5215 nA, σ = 0.0491 nA at 1 mmHg). We calculated that the minimum detectable IOP, with an SNR of ~3, is ~0.094 mmHg. This exceeds the resolution of commercially available tonometers, which typically offer a resolution of 0.1 mmHg, as well as the previously reported IOP sensors (fig. S3) (35–37).

Fig. 2. Sensing properties of the pressure sensor.

(A) Schematic illustration of the sensing mechanism. (B) Real-time measurements of the relative change in drain current (ΔI_D/I₀) according to the various pressures (V_G = 20 V, V_D = 1 V). (C) The relative change in I_D with the applied pressures. S represents the sensitivity of the pressure sensor. (D) Relative change in the I_D of the pressure sensor that occurred after 1000 cycles of repeated compression to 60 mmHg. (E) Response and recovery time of the pressure sensor when a pressure of 60 mmHg is applied. (F) Comparison of sensing properties in a pressure sensor with one side fixed to the bottom and a pressure sensor floating on fluid. (G) The relative change in I_D with applied water pressure as the pressure is increased stepwise. S represents the sensitivity of the pressure sensor. (H) Correlation between the IOP measurements taken using the pressure sensor and the manometer. PI, polyimide.

Figure 2D shows that this pressure-sensitive FET operated stably; it had a negligible change in signals (∆I_D/I₀) after 1000 cycles of repetitive pressure loading at 8 kPa (60 mmHg). In addition, this pressure sensor exhibited a fast response (~41 ms) and recovery time (~54 ms) with the applied pressure of 60 mmHg, as plotted in Fig. 2E. For implantable electronic biomedical devices, the electrical leakage (that can be caused by environmental factors or insufficient insulation) is one of the biggest challenges. In particular, in the case of an eye, the leakage current generated by implantable devices can cause retinal damage by electrically stimulating optical neurons, which can be fatal. Therefore, the leakage current of an implantable electronic device should be less than 10 μA (38–40). To investigate this leakage, this probe-type device was immersed inside a balanced salt solution (BSS) for 7 days. BSS is the perfusate that is used during the surgery on human eyes, and it is the solution most similar in composition to the aqueous humor and the vitreous humor. As shown in fig. S4A, the pressure-sensitive FET preserved its sensing performance with negligible deviation in signals (∆I_D/I₀). Also, the leakage current was measured to be less than 1 nA, which was small enough for the leakage current of implantable ocular devices. Similarly, this device also exhibited negligible changes in signals when it was immersed in a phosphate-buffered saline (PBS) solution for 7 days (fig. S4B). In addition, there was negligible variation in the sensitivity within the temperature range of 30° to 45°C (fig. S5). In Materials and Methods, we conducted tests on human retinal pigment epithelial cells to assess the cytotoxicity of the pressure-sensing probe. According to fig. S6, the cell viability was 84.7 ± 5.8%, indicating no substantial difference compared to commercially available ophthalmic implantable devices (41, 42). Therefore, we concluded that our IOP-sensing probe is not substantially cytotoxic to the human eyes.

After immersing this probe inside a water tank, the pressure-sensing performance of this device was compared when it was attached to the bottom of the tank versus suspended stationarily in water (at the same height) without its attachment. As exhibited in Fig. 2F, this pressure sensor exhibited negligible signal differences in these two cases. The vitreous chamber and the anterior chamber are filled with aqueous humor and vitreous humor, which are liquids. Therefore, the pressure-sensing performance of our probe was compared with the measurement using a commercial pressure sensor (MS5803-14BA, TE Connectivity) when these two different devices were submerged in water. Our pressure-sensitive FET detected the same pressure values as the commercial device in the pressure range below 60 mmHg, and Fig. 2G shows that the signal (∆I_D/I₀) of our sensor increased linearly with the applied pressure even when immersed in water without substantial difference in sensitivity (percentage difference between air and water cases: 14%). Also, we compared the sensing performance of our probe with the measurement of a conventional manometer. As shown in fig. S7, our probe and a manometer were implanted together into a bovine eye to obtain their real-time IOP measurements. Figure 2H and fig. S8 indicate that these two different ways of measuring IOP, i.e., using either our pressure-sensitive FET or the manometer, had a good correlation [intraclass correlation coefficient (ICC) = 0.997) with the coefficient of determination (r²) of 0.99, which indicated good reproducibility for both IOP measurements.

In vivo studies for monitoring IOP distribution in live animals

Using live rabbit models, our probe-type pressure sensors were used to monitor the local IOP distribution in both the anterior chamber and the vitreous chamber, and a commercially available tonometer (ICare Pro, ICare) was used to detect IOP in the eye based on the deformation of the cornea. Figure 3A illustrates the experimental setup, and Fig. 3B presents a photograph of the two pressure sensors that were implanted. Before conducting further experimentation, we ensured that the implantation of these probes caused negligible IOP differences, as measured by tonometry (Fig. 3C). For example, the rabbit’s IOP initially remained at 9.43 ± 0.39 mmHg (before insertion of the probe), and it was 9.49 ± 0.32 mmHg after the probe insertion. Also, after this probe insertion, the IOP in the contralateral control eye was recorded as 9.56 ± 0.66 mmHg, indicating a negligible difference. Subsequently, we started real-time monitoring of localized IOP distributions in the cornea, anterior chamber, and vitreous chamber. A comprehensive experimental setup, including the measuring instruments, is shown in fig. S9. Figure 3D and fig. S10 plot the IOP distribution for the normal rabbit model. The IOP values within these three areas were maintained similarly within the normal IOP range and no drastic variations occurred. Also, Fig. 3E presents the mean values and SDs of IOP in the three regions during the measurement period (cornea: 9.46 ± 1.24 mmHg, anterior chamber: 9.73 ± 1.47 mmHg, vitreous chamber: 9.67 ± 1.44 mmHg), and substantial differences were observed among the three areas. Comparing the IOP values detected by the tonometer (IOP_tono) at the cornea with those measured via our pressure sensors implanted in both chambers (IOP_chamber), Fig. 3F indicates a strong correlation (anterior chamber: r = 0.815 and ICC = 0.769, vitreous chamber: r = 0.754 and ICC = 0.729). The differences between the IOP values obtained at the cornea (by the tonometer) and those measured in the chambers are depicted in Fig. 3G. The variations in IOP in these regions were less than 0.5 mmHg, and the distribution of IOP among the cornea, anterior chamber, and vitreous chamber in a normal rabbit model indicates the reliability of conventional tonometry measurements in the normal IOP range.

Fig. 3. IOP distribution in a live rabbit model at normal IOP level.

(A) Schematic illustration of an experimental setup for measuring the IOP distribution with a tonometer and the probe-type pressure sensor on a live rabbit model. (B) Photograph of the pressure sensors inserted into the eye of a live rabbit. Scale bar, 5 mm. (C) The average IOP of the rabbit’s eye before and after the insertion of the device. A tonometer was used to measure the IOP. (D) Real-time measurements of the IOP distribution in the vitreous chamber, anterior chamber, and tonometer. (E) Normal IOP is measured with a tonometer and a pressure sensor. The normal IOPs of the anterior chamber and the vitreous chamber were measured using a probe-type pressure sensor. (F) Local IOP distribution of the anterior chamber and the vitreous chamber (IOP_chamer) compared to the tonometer (IOP_tono) in a normal IOP level. (G) The difference between the IOP (ΔIOP = IOP_chamber − IOP_tono) measured in the chambers of the rabbit’s eye and the IOP measured using a tonometer.

Hyaluronic acid is a naturally occurring polysaccharide that is used as a viscoelastic tool in ophthalmological surgery, and it also is included in artificial tears. It is used to increase IOP to prevent the collapse of the anterior chamber and to facilitate the manipulation of ocular tissues during surgery. As illustrated in Fig. 4A, we injected 0.2 ml of hyaluronic acid into the anterior chamber in a live normal rabbit model to increase the volume of aqueous humor in the anterior chamber and to raise the IOP level. Figure 4B is a photograph of this injection of hyaluronic acid through a syringe (needle outer diameter: 0.474 mm), and the photograph shows that there was no leakage of hyaluronic acid, ocular fluid, or blood. Figure 4C and fig. S11 present the real-time measurements of the local IOP distribution in the anterior chamber and the vitreous chamber (using our pressure-sensitive probes) as well as at the cornea (using a tonometer), after the injection of the hyaluronic acid. The injection of the hyaluronic acid induced an acute increase in IOP (within 30 s) in both chambers. Subsequently, the increased IOP decreased gradually and eventually returned to the normal IOP range. Figure 4D compares the probe-measured IOP in the vitreous chamber and anterior chamber (IOP_chamber) with the tonometer-measured IOP (IOP_tono) at the cornea. Despite the injection of the hyaluronic acid, the IOP distributions between the cornea, the anterior chamber, and the vitreous chamber were well aligned within the entire IOP range of 7 to 60 mmHg. For example, Fig. 4 (E and F) shows high concordance of IOP values in the anterior chamber (IOP_ant) and vitreous chamber (IOP_vit) in the normal IOP range (7 to 15 mmHg) and high IOP conditions (15 to 60 mmHg), respectively. These in vivo results in a live rabbit model are consistent with previous in vitro studies using enucleated porcine eyes (17, 22, 43). However, as shown in Fig. 4G, our in vivo experiments showed that the absolution difference in IOP between the vitreous chamber and the anterior chamber, i.e.,│IOP_vit − IOP_ant│, increased slightly in the high IOP condition (from immediately after the injection of hyaluronic acid until the IOP saturation to the normal range). This absolute difference was less than 3 mmHg in the normal IOP range, but it was more than 3 mmHg in 36% of the points detected at high levels of IOP (Fig. 4H). In addition, a small increase also was observed when the relative disparity was considered, as shown in fig. S12A. Specifically, the positions with a difference of 20% or more increased from 2.3% for normal IOP to 13.2% for the high IOP range (fig. S12B). In view of these results, IOP measurements using conventional tonometry are reliable in all ranges of IOP, but the transient difference (i.e., │IOP_vit − IOP_ant│) in high IOP ranges, due to temporary acute fluctuations in IOP in the anterior chamber, may not be reflected accurately in conventional tonometry because it detects IOP only through changes in the cornea that reflect IOP in the anterior chamber, not in the vitreous chamber.

Fig. 4. IOP distribution in a live rabbit model at a high IOP level.

(A) Schematic illustration of the hyaluronic injection method for inducing a high IOP level. (B) Photograph of hyaluronic injection into the anterior chamber. Scale bar, 5 mm. (C) Real-time measurements of IOP distribution with pressure sensors and tonometer. The green part indicates where the high IOP level was induced by hyaluronic injection. (D) Local IOP distribution of the anterior chamber and the vitreous chamber (IOP_chamber) compared to the tonometer (IOP_tono) in high IOP level. (E) Local IOP distribution between the anterior chamber (IOP_ant) and the vitreous chamber (IOP_vit) at the normal IOP level. (F) Local IOP distribution between the anterior chamber (IOP_ant) and the vitreous chamber (IOP_vit) at a high IOP level. (G) IOP differences (IOP_vit − IOP_ant) between the anterior chamber and the vitreous chamber. (H) The distribution of IOP difference (│IOP_vit − IOP_ant│) values in the normal IOP level and the high IOP level.

In vivo studies for monitoring IOP distribution in normal and glaucomatous models

The aqueous humor is secreted by the ciliary epithelium and it flows through the trabecular meshwork into Schlemm’s canals, which are located between the iris and the cornea. When the eye makes too much aqueous humor or the drainage system does not work properly, the accumulated fluid increases the pressure in both the anterior and vitreous chambers, and it is commonly linked to elevated IOP. In glaucoma models, the pressure in the vitreous chamber may differ from the pressure in the anterior chamber due to various factors, such as the presence of the lens and variations in fluid dynamics and drainage pathways. Thus, measuring the pressure in the vitreous chamber is crucial because it can provide valuable information about the condition of the retina, which helps in early diagnosis of problems, such as glaucoma, and it enables effective management to prevent the loss of vision. In addition, this measurement of both sides may be important for understanding the pathophysiological progression of disease in glaucoma. We utilized this pressure-sensitive FET to monitor the distribution of IOP in the eyes of rats. Glaucoma was induced in rats by cauterizing the two episcleral veins located in the upper part of the eye (Fig. 5A and fig. S13). The induction of glaucoma was confirmed by hematoxylin and eosin (H&E) staining of the retina. Figure 5B presents the degradation of the uppermost RGC layer in the glaucoma model, compared to the normal model case. The RGC density in glaucoma-induced rats was only 60.2% compared to the normal group (fig. S14). Then, we monitored the local IOP values in both the vitreous and anterior chambers of these rats’ eyes continuously after inserting pressure-sensor probes inside these chambers. Simultaneously, for comparison, we also measured IOP in the cornea using a tonometer (Fig. 5C). Also, Fig. 5D shows a photograph of these two pressure-sensor probes, each inserted into the anterior chamber and the vitreous chamber, without any bleeding or substantial leakage of ocular fluid. In addition, fig. S15 indicates that the change in the IOP caused by the probe we inserted was negligible.

Fig. 5. IOP distribution in normal and glaucoma rats.

(A) Schematic illustration of glaucoma induction by cauterization of the episcleral veins. (B) Hematoxylin and eosin staining of the normal model and glaucoma model. GCL, ganglion cell layer; INL, inner nuclear layer; ONL, outer nuclear layer; IS/OS, photoreceptor inner and outer segments; RPE, retinal pigment epithelium. Scale bar, 100 μm. (C) Schematic illustration of an experimental setup for measuring the IOP distribution in rats. (D) Photograph of the probe-type pressure sensors inserted into the eye of live rats. Scale bar, 2 mm. (E) Real-time measurements of the IOP distribution in the vitreous chamber, anterior chamber, and tonometer with normal and glaucoma models. (F) Changes in IOP measured by tonometer and in the vitreous chamber and anterior chamber, caused by glaucoma. (G) Local IOP distribution of the anterior chamber and the vitreous chamber compared to the IOP measured with the tonometer. (H) Local IOP distribution of the anterior chamber and the vitreous chamber in a normal eye and glaucoma-induced eye. (I) The IOP differences between IOP in the anterior chamber and vitreous chamber compared to tonometer-measured IOP in normal and glaucoma models. (J) The percentage of time at which the difference in IOP in the vitreous chamber compared to values in other regions (top inset: tonometer; bottom: anterior chamber). (K) The Pearson’s correlation coefficient between the tonometer and the vitreous chamber (top inset) and the Pearson’s correlation coefficient between the anterior chamber and the vitreous chamber (bottom) that measured in both normal and glaucoma models.

Real-time IOP measurements were conducted for 30 min in three eyes each from the normal model and the glaucoma model (Fig. 5E and figs. S16 and S17). Snapshot IOP measurements using a commercial tonometer (Tonolab, ICare) were performed at 150-s intervals. As shown in Fig. 5F, in the normal model, the tonometry measurement showed an average IOP value of 10.33 ± 0.65 mmHg, pressure-sensor detection exhibited 10.62 ± 0.62 mmHg in the anterior chamber, and 10.87 ± 0.69 mmHg in the vitreous chamber. This normal model presented a similar IOP distribution in the cornea, the anterior chamber, and the vitreous chamber. However, in the glaucoma model, the corneal IOP (tonometry detection: 15.54 ± 0.81 mmHg on average) was almost similar to the anterior chamber’s IOP (pressure-sensor measurement: 15.43 ± 0.76 mmHg), but the vitreous chamber’s IOP (pressure-sensor measurement: 19.51 ± 1.29 mmHg) was noticeably higher. This glaucoma model exhibited substantially higher IOP in all three regions (i.e., the cornea, the anterior chamber, and the vitreous chamber) compared to the case of the normal model case, indicating a successful increase in IOP through cauterization. Three mice were tested for each model, and figs. S18 and S19 show the average IOP of each subject for normal and glaucoma-induced eyes, respectively. Figure 5G compares the IOP of each chamber to the corneal IOP for the normal and glaucoma models. In particular, Fig. 5H plots all IOP values detected in the anterior chamber and vitreous chamber of each subject (in the normal and glaucoma models). The red dotted line indicates the value at which IOP in the vitreous chamber and the anterior chamber are identical. In normal eyes, most values were around the dotted line (in both the anterior and vitreous chambers). However, while IOP values in the anterior chamber of the glaucoma-induced eye were still around this red dotted line, all values in the vitreous chamber of the glaucoma model were above this baseline. Figure 5H indicates that, with the induction of glaucoma, the vitreous chamber experienced relatively higher IOP than the anterior chamber experienced. All other subjects presented the same trend, as shown in fig. S20.

Furthermore, Fig. 5I shows the difference between the corneal IOP and each chamber’s IOP in the normal and glaucoma models. Although the normal model exhibited negligible difference in IOP, the glaucoma model presented a substantial difference in IOP (3.84 ± 1.24 mmHg) between the vitreous chamber and the cornea, indicating a higher IOP in the vitreous chamber. This emphasizes that the development of glaucoma resulted in substantially higher IOP in the vitreous chamber than it did in the anterior chamber and cornea. Figure 5J shows the temporal percentage during which the vitreous chamber’s IOP was higher than the IOP values of the other two regions (i.e., the cornea and the anterior chamber). In the case of normal eyes, the difference between the vitreous chamber’s IOP and the corneal IOP is maintained below 10% for 71.8% of the total measured time. (Similarly, the difference between the vitreous chamber’s IOP and the anterior chamber’s IOP was maintained below 10% for 93.4% of the total time). On the other hand, in the case of glaucomatous eyes, the period in which the vitreous chamber’s IOP was more than 10% higher than the corneal IOP was 97.5% of the total measurement time. (For the anterior chamber of the glaucoma model, this period was 98.9%.) Figure 5K compares the Pearson’s correlation coefficient before and after the induction of glaucoma induction, and all three subjects in the glaucoma model had decreases in this coefficient compared to the cases involving the normal model. Our results indicated that glaucoma alters the local distribution of IOP in the eye, with a marked increase of the IOP in the vitreous chamber, where RGC damage occurs. This is thought to be due to differences in the biomechanical properties between the anterior chamber and the vitreous chamber. In particular, the lens occupies a larger proportion of the total eye in the rat than in the rabbits. In addition, glaucoma causes tissue remodeling of the ocular components such as the cornea and sclera, leading to a change in biomechanical properties. This change can maximize the IOP difference between the anterior chamber and the vitreous chamber. This means that it may be difficult to use conventional tonometry for measurements in glaucoma models because it only measures IOP through changes in the cornea which primarily reflects the IOP in the anterior chamber, rather than in the vitreous chamber.

DISCUSSION

This in vivo study presents a minimally invasive, probe-type pressure-sensitive FET for monitoring the local IOP distribution in the anterior chamber and the vitreous chamber of living animals. To use this FET as an IOP sensor, it was designed with a long, sharp, and needle shape to facilitate its direct insertion into specific local areas of an eye. High sensitivity, as well as the fast response of this air-dielectric sensor structure, enabled reliable and accurate monitoring of acute changes in the IOP in animal models. Also, this implantation did not cause any bleeding or leakage of ocular fluid. A comparison of IOP values measured with a tonometer before and after the implantation of our probe confirmed that this insertion did not alter the IOP.

With the sensor that was fabricated, first, we confirmed the difference in IOP distribution in both normal and glaucoma models of living animals. When comparing the IOP distribution between normal and glaucomatous eyes in vivo models after the induction of glaucoma, both the vitreous chamber and the anterior chamber had distinct differences in the IOP, demonstrating that the IOP can vary depending on its intraocular location. The normal model showed an alignment of the distribution of the IOP among the cornea, the anterior chamber, and the vitreous chamber, but the glaucoma model presented notably higher IOP in the vitreous chamber, where RGC damage occurred. This means that it may be difficult for conventional tonometers to measure IOP accurately in glaucoma models because they measure IOP only through changes in the cornea, which reflects IOP in the anterior chamber rather than in the vitreous chamber. Furthermore, when hyaluronic acid was injected into the anterior chamber in a living normal model to increase the volume of aqueous humor in this chamber, our continuous monitoring sensor observed transient and acute IOP differences between the anterior chamber and the vitreous chamber in high IOP ranges. Given that the IOP spikes that are caused by radical fluctuations can accelerate glaucoma (44), these deviations are essential factors to consider in the management of glaucoma. However, these transient and temporary fluctuations in IOP may not be reflected accurately by snapshot measurements of conventional tonometry. This occurrence is due to the biomechanical characteristics of the eye. The fluids that fill the anterior and vitreous chambers have differences in viscosity and properties. These two areas, separated by the lens, display volume discrepancies. There are also variations in the modulus of the cornea and sclera that surround these areas. The IOP difference observed in these two compartments is expected to be due to these factors. Moreover, conditions like glaucoma can cause reorganization-induced changes in ocular tissue characteristics, which can worsen the IOP imbalances.

Although the distributions of IOP measured using our probe-type pressure sensors indicate a potential approach for the analysis of glaucoma, our experiments were limited to anesthetized animal models due to the invasiveness of our device. Therefore, assessing IOP distribution in non-anesthetized subjects during everyday activities or study factors that alter IOP (such as posture or exercise) would provide future strategies for managing glaucoma. However, our current device form with wired interconnections requires the use of large, external measuring equipment, and its invasive nature poses a risk of contagion in everyday situations. Consequently, there is a need for wireless operation of devices by integrating our sensor with wireless IC chips and wireless antennas. Then, a secure fixation method for such wireless devices can be achieved by inserting them into the suprachoroidal space.

The suprachoroidal space, which is situated between the sclera and the choroid, provides an appealing opportunity for the placement of a wireless device, which would eliminate the necessity for direct intraocular penetration (45). This space displays potential for multiple applications, including the delivery of drugs or serving as an engraftment site for ocular prosthetics (46). Positioned right under the sclera, the suprachoroidal space offers enough room for the implantation of a device similar in size to an eyeball. The suprachoroidal space, which is a void between membranes, impedes external devices from penetrating the chamber, thereby reducing the risk of infection. This method presents substantial benefits in terms of biocompatibility and long-term stability. Although various difficulties remain to be resolved, such as wireless functionality and miniaturization, the implementation of these techniques in the future is anticipated to allow the assessment of IOP distribution in daily life.

The measured distribution of IOP suggests the need for direct IOP measurements applied to the retina within the vitreous chamber in a clinical setting. Specifically, the measurement of IOP in the vitreous chamber is crucial for glaucoma patients because it can damage the retina. The aforementioned strategy of inserting devices into the suprachoroidal space can be utilized in this regard. Currently, efforts are being made to transplant nonelectronic devices, such as stents, into the suprachoroidal space for the treatment of glaucoma (47, 48). With advancements in wireless capabilities and miniaturization, it will be possible in the future to insert devices into this space. This approach has the potential to minimize damage to the vitreous humor and the lens, thus preserving the natural function of vision without obstructing the field of view. The device also will help people deal with concerns about portability and the inconsistent measurements that commonly are associated with conventional tonometers. This allows for the direct IOP assessment on the retina. Continuous monitoring of the IOP permits the identification of increased pressure, which, in turn, permits the use of drugs to avert injury to the RGCs and the optic nerve. Ultimately, the adoption of patient-centered interfaces for measuring IOP enables the patients’ participation in prevention and timely therapeutic interventions.

This FET-based pressure sensor probe is expected to be used for various other purposes. As an example, the real-time measurement of intracranial pressure (ICP) is another suitable application scenario for our probe-type pressure sensor. ICP can fluctuate due to brain-related conditions such as brain hemorrhage and tumors, so, after brain surgery, continuous monitoring may be required to confirm the recovery and detect potential side effects (49, 50). Furthermore, ICP shares a similar range with IOP and also can indicate locally different pressures (51). A FET-based pressure probe is highly suitable for monitoring ICP because it has minimal invasiveness and the capability for precise pressure measurement within the specified pressure range. Real-time monitoring of ICP allows for the confirmation of the progression of the disease and the postsurgical recovery situation because it provides various information about ICP and brain diseases. In this way, our probe has notable potential and will provide advantages for biomedical purposes.

MATERIALS AND METHODS

Fabrication process of a soft probe-type pressure sensor

The pressure sensor is fabricated by dividing it into three parts, i.e., (i) a top film with channel and source/drain electrodes, (ii) a bottom film with gate electrodes, and (iii) a deformable dielectric layer. A PI film (25 μm) is attached to the cleaned Si wafer with spin-coated PDMS (Sylgard 184, Dow Corning, base:agent = 10:1) at 3000 rpm for 30 s as an adhesive layer. The PI substrate for the source/drain electrodes was spin-coated by a negative photoresist (SU-8 2002, MicroChem) at 4000 rpm for 40 s for the transfer of the Si channel. After the SU-8 coating was applied, it was cured by exposure to ultraviolet (UV) for 30 s. The Si channel for the transistor was acquired from the top Si layer (205 nm) of the SOI wafer (205-nm Si on 400-nm buried oxide layer, Soitec). The channel was patterned using a positive photoresist (S1818, MicroChem) with photolithography on the SOI wafer. S1818 was spin-coated at 4000 rpm for 40 s, and it was exposed to UV (365 nm) for 13 s. The adhesion promoter (Surpass 4000, Dis-Chem) was used to prevent the failure of the channel in the reactive ion etching (RIE) process. The channel (length: 10 μm, width: 70 μm) was etched by RIE using sulfur hexafluoride [SF₆, 25 standard cubic centimeters per minute (SCCM); Ar, 55 SCCM/300 W/40 s]. After the channel was etched, the residue of the photoresist was removed completely with Piranha solution (sulfuric acid:hydrogen peroxide = 3:1) for 15 min. The buried oxide layer was etched anisotropically in a 49% hydrogen fluoride (HF) solution for 9 min and 20 s to detach the Si channel from the SOI wafer. Then, the channel was transferred to the PI substrates coated with the SU-8 layer using the PDMS stamp (base:agent = 4:1). The metal layers (Cr/Au, 2 nm/300 nm) were deposited for electrodes on the PI substrates by e-beam evaporation, and the source/drain electrodes were patterned using S1818 (3000 rpm, 30 s) photolithographically. For patterning electrodes, S1818 was exposed to UV for 15 s, and it was removed by a developer (AZ 300 MIF, Merck) for 45 s. Then, the electrodes were wet etched by Au etchant (Gold ETCH TFA, Transene) for 30 s and Cr etchant (Cr Etch 905 N, Transene) for 10 s. First, the photoresist residue was removed through UV exposure (5 min) and developer (15 min); then, it was cleared completely by RIE (O₂, 40 SCCM/200 mTorr/100 W/40 s). Except for the channel and the electrode pads for interconnection, the SU-8 layer (thickness: 1.5 μm) was spin-coated at 4000 rpm for 40 s for the passivation of the electrodes. The PDMS film (thickness: 50 μm) for the dielectric layer was pattered by using a laser ablation method (CO₂ laser, Epilog Laser, Inc). Then, the top film and bottom film were patterned for a needle shape using RIE (SF₆, 25 SCCM; Ar, 55 SCCM; O₂, 90 SCCM/200 W/35 min). The copper film (thickness: 3 μm), which was made in the shape of a needle, was used as the mask to protect the films from the RIE, and it was deposited and patterned by photolithography. Then, the top film, bottom film, and dielectric layer were assembled using an optical microscope (OM), and the side of the pressure sensor was sealed by a biomedical grade elastomer (SILASTIC MDX4-4210, Dow Corning).

Characterization of pressure-sensitive transistor

The characteristics of the pressure-sensitive transistor, including the transfer and output characteristics, were measured through a probe station with a parameter analyzer (Keithley 4200-SCS). The current between the source electrode and the drain electrode was measured using a three-terminal measurement system. Then, to assess the performance of pressure sensing, an experimental setup was configured using equipment that was capable of applying precise pressure. The pressure was applied up to 60 mmHg by using a movable z-axis stage (Mark-10 ESM303) and a highly accurate force gauge (Mark-10 M7-20). At this time, the change in the drain current in the pressure sensor was measured through source meters (2400 series, Keithley). One source meter was used to apply a constant voltage to the gate electrode, and another source meter was connected to the source electrode and drain electrode to measure the drain current.

Cell culture and cell viability assay

ARPE-19 cells (retinal pigment epithelial cell line, American Type Culture Collection, Manassas, VA, USA) were cultured in Dulbecco’s modified Eagle’s medium (DMEM)/F12 (Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with GlutaMax, 10% heat-inactivated fetal bovine serum (FBS), and penicillin (50 U ml⁻¹)/ streptomycin (50 μg ml⁻¹) (both from Welgene Inc., Gyeongsangbuk-do, Korea). The culture media were replaced with fresh media every 3 days and grown at 37°C in a humidified 5% CO₂. For the cell viability assay, the cells were plated at a density of 1 × 10⁵ cells ml⁻¹ in 96-well plates. The fabricated pressure sensor was sterilized with 70% ethanol for 3 min, rinsed in H₂O for 3 min, and then exposed to UV for 30 min. The sensor was incubated with DMEM/F12 1% FBS supplemented GlutaMax for 24 hours, and the sensor-immersed medium (SIM) was collected and stored at −80°C until it was used. Each well was treated with 100 μl of SIM or only media as a control group for 24 hours, and viability was analyzed using Quanti-MAX WST-8 Cell Viability Assay Kit according to the protocol provided by Biomax (Seoul, Korea). The optical density was measured at 450 nm using Victor X5 2030 Multiable Plate Readers (PerkinElmer).

Bovine eyeball model of ocular hypertension

In vitro tests were conducted by using a bovine eyeball. The IOP of the bovine eyeball was controlled using a syringe pump by inserting the PBS solution into the vitreous chamber. A commercial manometer (EM201B, UEi) was inserted into the vitreous chamber of the eye to measure the IOP of the vitreous chamber in the bovine eyeball. Also, the fabricated pressure sensor was inserted into the vitreous chamber to measure the IOP. The change in the drain current of the pressure sensor was characterized using source meters (2400 series, Keithley). One source meter was connected to the source/drain electrode to apply the drain voltage, and the other source meter was connected to the gate electrode to apply the gate voltage. The change in drain current due to the IOP was measured in real time, and the IOP through the manometer was measured every 15 s.

Animal models

All of the research in this paper complies with all relevant ethical regulations. Animal procedures were conducted on the basis of the guidelines of the Institutional Animal Care and Use Committee of the Daegu-Gyeongbuk Medical Innovation Foundation (DGMIF, DGMIF-21090302-01). The Institutional Animal Care and Use Committee of the Daegu-Gyeongbuk Medical Innovation Foundation was the ethics review committee. The experimental procedures were carried out in rabbits (New Zealand white rabbit, male, 3.0 kg) and rats (Orient Bio, male, 150 g). The animals were maintained on a 12/12-hour light/dark cycle in a temperature-controlled room (23° ± 2°C).

Rat models of glaucoma for IOP measurement

The chronic IOP elevation model (glaucoma model) was implanted surgically in the left eyes of three male Sprague-Dawley rats (weighing >150 g). Before surgery, the rats were anesthetized with an intraperitoneal injection of ketamine hydrochloride (75 mg kg⁻¹) and xylazine (10 mg kg⁻¹) before proceeding to any procedures. The two episcleral (dorsal and temporal) veins of the left eye were caused by hand-held cautery (Change-A-Tip, Aaron, USA), while the right eye did not receive surgical procedures and was used as a control. After cauterization, the rats were returned to the cage. Three weeks after the chronic glaucoma animal model, the IOP was measured with two probe-type pressure sensors.

Rabbit models for IOP measurement

Rabbits (New Zealand white male rabbits) that weighed 2.5 to 3.0 kg were intramuscularly preanesthetized with ketamine hydrochloride (50 mg kg⁻¹) and xylazine 10% (5 mg kg⁻¹) and anesthetized deeply by inhalation of 3% isoflurane (HanaPharm Co. Ltd., Seoul Korea). Topical anesthesia (0.4% oxybuprocaine drops) was applied to the eyes. The IOP of the live rabbits was measured at the cornea by using the tonometer (ICare Pro, ICare). In addition, the probe-type pressure sensors were inserted into the anterior chamber and vitreous chamber. In addition, hyaluronic acid was used to induce high IOP. Hyaluronic acid is a viscous substance that is included in artificial tears, and it is used for surgery or eye treatment by injecting it into the eye. To increase IOP, 0.2 ml of hyaluronic acid was used per session.

IOP measurement with the pressure sensor in the animal model

The pressure sensor was inserted after a hole was made using a 26-gauge needle. The outer diameter of the 26-gauge needle was 0.474 mm, and it can make a hole sufficient for the pressure sensor to enter. The pressure sensor worked by connecting to the source meter and power supply. The source electrode and drain electrode were connected to both ends of the source meter (2400 series, Keithley) to measure the change in drain current. A drain voltage of 1 V was applied through the source meter. The gate electrode was connected to the power supply (2260b series, Keithley), and a voltage of 20 V was applied. The drain current was measured in real time through software. The measured drain current was calculated through sensitivity and converted into pressure.

Tissue processing and pathological damage assay

The rat’s retinas were enucleated, and the anterior segments were removed. Retinas were fixed in Davidson’s fixative (Hartmann’s fixative; 32% ethanol, 2.2% neutral buffered formalin, and 11% glacial acetic acid in distilled water) (Merck, Darmstadt, Germany) for 4 hours at room temperature and then in 4% paraformaldehyde overnight at 4°C. Eyecups were washed with PBS several times and embedded in paraffin. Five-micrometer sections were cut and placed in a heat plate at 56°C for 30 min. Following deparaffinization and rehydration, sections were stained with an H&E stain kit according to the protocol provided by Vector, Brussels, Belgium). Then, they were dehydrated in alcohol and cleared in xylene. The image of the section was captured by an Olympus BX53M microscope in three random fields per section and used to assess pathological damage. All photo adjustments were carried out equally across the sections. The number of RGCs was counted.

Statistical analysis

The data were expressed as means ± SD. Statistical analysis of the P value was conducted using an open-source MATLAB code. Significant differences were assessed using the unpaired Student’s t test and indicated as n.s. P > 0.05, *P > 0.01, **P > 0.001, ***P > 0.0001, and ****P < 0.0001.

Supplementary Materials

This PDF file includes:

Supplementary Text

Figs. S1 to S20