Glaucoma surgery experiments using digital microscope-integrated optical coherence tomography and OCT-compatible instruments

1. Introduction

Glaucoma is one of the leading causes of irreversible blindness worldwide.¹ In traditional Glaucoma surgery, the operating microscope can only provide en-face images, limiting the surgeon’s depth perception and visualization of the subsurface anatomy. For example, trabeculectomy, one of the most familiar glaucoma surgeries around the world, requires surgeons to cut the trabecular meshwork and create aqueous humor drainage during surgery.^2,3 However, it is quite difficult to locate trabecular for surgeons using the traditional surgical microscope, which mainly depends on their experience. It was reported that the trabecular was untouched in 34% of cases,⁴ which should be responsible for the high failure rate (35.5%) of trabeculectomy.⁵ Canaloplasty,^6,7,8 another form of nonpenetrating glaucoma surgery, uses a microcatheter to perform a 360^∘ cannulation of Schlemm’s canal and leaves in place a tension suture providing an inward distension. However, 7.3–26% of the cannulation failed because the Schlemm’s canal cannot be seen during surgery.⁹

Optical coherence tomography (OCT)¹⁰ is a noncontact, high-resolution imaging technique that enables cross-sectional views of anterior and posterior segments of the eye. Combining the microscope and OCT is an ideal way that provide both OCT and microscopic images to guide eye surgery.^11,12,13,14 OCT visualization can be displayed on an external monitor,¹¹ or a built-in heads-up display.^15,16,17 Most (MIOCT) systems adopted the second way because the external display required surgical pauses during surgery. However, the built-in display has a limited resolution, field of view, and contrast to visualize the OCT images.¹⁸

In OCT-guided surgery, instrument shadow limits the visualization of instruments’ edges and underlying tissues, which is a main obstacle caused by the low transmittance of the metallic materials of the instruments. Several attempts of custom-made instruments using transparent materials have been made to solve this problem.^13,19 For example, Ehlers et al. designed custom-made corneal needles, surgical picks, and retinal forceps using different materials and measured the optical properties to determine whether they were consistent with the semitransparent nature within the OCT wavelength range.¹⁹ However, these instruments lacked an accurate feasibility assessment in real-time surgery.

Here in this paper, we demonstrate a digital MIOCT system including an external 3D display on which the microscopic and OCT images were projected simultaneously and an auxiliary monitor providing high-resolution multi-line OCT images. OCT-compatible instruments made of polymethyl methacrylate and quartz glass were also developed to solve the instrument shadowing problem. Then glaucoma surgery including trabeculectomy and canaloplasty was performed on porcine eyes using this MIOCT system and OCT-compatible instruments. Finally, a subjective feedback survey about the whole system was asked to take after surgery.

2. Materials and Methods

2.1. MIOCT system setup

An SS-OCT system was integrated into a custom-built surgical microscope, as shown in Fig. 1(a). The swept source operated at 1310nm wavelength with a swept range of 90nm and an A-Scan rate of 100KHz. In the sample arm, a super luminescent diode with a wavelength of 640nm was combined into the optical path used as a guide source to mark the imaging region. Beam direction and lateral scan were controlled by an X–Y galvanometer and a waveform generation card. The optical interference signal was detected by a balanced detector and then acquired by a data acquisition device. The optical power of OCT at the sample arm was 4.7mW, following the American National Standard for Light Hazard Protection for Ophthalmic Instruments (ANSI Z80.36-2016). The maximum depth and lateral range of the SS-OCT were 6mm and 14.3mm respectively. In this work, the OCT B-Scan lateral range was set to 10mm and consisted of 714 A-lines. There were three real-time imaging modes during surgery: single-line, three-line, and five-line. The average of three B-Scan was used for single-line mode and the nonaverage was used for three-line and five-line modes. 3D OCT imaging can also be obtained in our system.

The optomechanical design is shown in Figs. 1(b) and 1(c). All parts of the SS-OCT system were integrated into the microscope. OCT and microscope shared an optical path after a dichroic mirror using an objective lens with a 200mm focal length. The microscope had a zoom adjustment from $4.5\times$ to $21\times$ and an X–Y–Z displacement of $\pm 25$mm controlled by a foot pedal during surgery. Binocular microscopic images were separated by a beam splitter and recorded by two surgical video cameras (MCC-500MDC, SONY, Japan), then displayed on an HD (1080P, $1920\times 1080$pixel) 42-inch widescreen LCD 3D display (LMD-4251TD, SONY, Japan). The OCT real-time image was displayed on the right bottom of the 3D display. Surgeons need to wear circular polarized 3D glasses that can give depth perception in surgery. An additional monitor ($1920\times 1080$pixel) placed near the 3D display was capable of displaying up to six OCT images ($500\times 380$px for each). The first five display areas displayed multi-lime cross-section OCT images in real-time, while the last area displayed a screenshot of the 3D display that the surgeon thought was useful for subsequent manipulation. Surgeons can select one of the five real-time images displayed on the 3D display which saves the time of finding the target region during surgery.

Statistical analysis in this work was performed by SPSS (IBM, USA), and p-values of $<0.05$ were deemed statistically significant.

2.2. Custom-made OCT-compatible instruments

OCT-compatible instruments including iris repositor, forceps and trabeculotome were designed to avoid the shadowing problem of traditional metallic instruments. The criteria of the instruments were transmittance and reflectivity. Transmittance affects the visibility of tissues below instruments while reflectivity affects the visibility of the instrument surface. OCT-compatible instruments usually have a lower reflectivity than metallic instruments. The tips of the iris repositor and forceps in our studies were made of polymethyl methacrylate while the trabeculotome used quartz glass, as shown in Fig. 2.

Transmittance and reflectivity of OCT-compatible instruments were evaluated by quantifying specified areas in OCT images from six freshly enucleated porcine eyes. OCT images of the anterior chamber structure were first acquired. Then the iris repositor, forceps, and trabeculotome were placed on the iris, sclera, and collector vessels (corresponding to Schlemm’s canal in the human eye), respectively. Finally, the intensity of underlying tissues in OCT images with and without instruments was compared with paired samples t-test.

2.3. Real-time MIOCT imaging for trabeculectomy and canaloplasty

To evaluate the imaging performance of the MIOCT system and OCT-compatible instruments in glaucoma surgery, 16 ophthalmic surgeons from Fudan Eye & ENT hospital were invited to implement trabeculectomy and canaloplasty. Surgery experiments were conducted on 16 porcine eyes from eight live male Yorkshire pigs (10kg, 2 months old).

In trabeculectomy, the key operation procedure was shown as follows: After general anesthesia, the eyelids were opened first and then the conjunctiva’s capsule was gently dissected. A $5\times 5$mm limbal-based superficial scleral flap of one-half thickness was created subsequently using an artificial sapphire knife (Fig. 3(a)) to expose the trabecular. After that, a $4\times 2$mm (lateral) trabecular piece was removed as shown in Fig. 3(b). Finally, the iridectomy was performed.

In canaloplasty, the procedure was the same as trabeculectomy until the scleral flap was created. Then the collector vessel was located with the help of the MIOCT system and an incision perpendicular to it was made across the scleral bed. After that, the collector vessel was expanded with an OCT-compatible trabeculotome ($\mathrm{\Phi }=250\mu$m) (Fig. 3(c)). Finally, a microcatheter (iTrack 250, Ellex iScience, Inc., Freemont, CA, USA), which was an optical fiber ($\mathrm{\Phi }=200\mu$m) connected to red flickering laser light for identifying the location of the fiber tip, was threaded into the angular aqueous plexus (AAP), corresponding to the trabecular meshwork in human eye (Fig. 3(d)).

After surgery, the experimental animals were euthanized by an intracardial injection of approximately 20ml 15% potassium chloride solution according to the 3Rs principle. All surgeons were asked to complete a quantitative subjective feedback survey about the user experience of the digital MIOCT system and OCT-compatible instruments (see supplementary material).

3. Results

3.1. Visualization of tissue and OCT-compatible instruments

In tissue and instruments imaging experiments, the tissue had a similar intensity with or without instruments above them (without forceps: 104.8, $\mathrm{\text{SD}}=25.0$, with forceps: 93.1, $\mathrm{\text{SD}}=21.4$, $P=0.226$; without iris repository: 77.9, $\mathrm{\text{SD}}=22.5$, with iris repository: 80.3, $\mathrm{\text{SD}}=18.4$, $P=0.611$; without trabeculotome: 67.5, $\mathrm{\text{SD}}=30.2$, with trabeculotome: 66.1, $\mathrm{\text{SD}}=32.5$, P = 0.839) as shown in Fig. 4. All the tissues and the edges of the instruments can be clearly identified.

3.2. Digital MIOCT-guided trabeculectomy and canaloplasty

Informative OCT images were successfully acquired using the MIOCT system and OCT-compatible instruments in surgery. During trabeculectomy, the assistant can draw a specified length line or measure the distance between two points on any area of the display to help the surgeon make a surgical manipulation. Surgeons can make the sclera flap more accurately with the help of the above function. The AAP could be easily located, as shown in Figs. 5(a) and 5(b). The scleral flap was created until the AAP was exposed. A new pathway for aqueous humor could be confirmed in the OCT image after a piece of AAP tissue was removed, as shown in Fig. 5(c).

In canaloplasty, the collector vessel was clearly visualized in the OCT image. The scleral flap was made until the collector vessel was reached. With the guidance of MIOCT, the collector vessel was expanded using the OCT-compatible trabeculotome, as shown in Fig. 6 and video 1 (Supplementary Information). Then an insertion of a microcatheter was performed as shown in Fig. 7 and video 2 (Supplementary Information). Surgeons can select one of the best images from (b) to (f) in Figs. 6 and 7 to display on the 3D display, which is quite efficient in surgery. The above two manipulations can also be identified in 3D-OCT imaging (Figs. 6(g) and 6(h) and Figs. 7(g) and 7(h)). The expanded collector vessel was also visualized by the OCT system at the end of the surgery.

3.3. Subjective feedback survey

Survey results about the custom-made instruments are shown in Fig. 8. The OCT-compatible iris repositor and forceps were the same as metallic instruments in terms of comfort and the ability of manipulation (iris repositor: 4.88, $\mathrm{\text{SD}}=0.342$; forceps: 4.56, $\mathrm{\text{SD}}=0.72$; between “Same” and “Slightly difference”). However, OCT-compatible trabeculotome showed worse performance (trabeculotome: 2.69, $\mathrm{\text{SD}}=0.60$; between “Little different” and “Mildly different”). The tip of the trabeculotome sometimes bent during the insertion into the collector vessel because of the low stiffness of the material. This may affect the expansion of the collector vessel. Surgeons were rarely affected by the OCT image deformation caused by the iris repositor and trabeculotome (iris repositor, 4.75, $\mathrm{\text{SD}}=0.45$; trabeculotome, 4.19, $\mathrm{\text{SD}}=0.66$; between “Sometimes” and “Never”), but forceps had a bad performance (2.94, $\mathrm{\text{SD}}=0.68$, between “Most of the time” and “About half of the time”). The refractive index of the forceps would cause deformation of the eye tissue under the instrument, which had a bad influence on the judgment of surgical operation. All OCT-compatible instruments were clearly recognized in digital microscopic images (iris repositor: 4.63, $\mathrm{\text{SD}}=0.62$; forceps: 4.44, $\mathrm{\text{SD}}=0.81$; trabeculotome: 4.50, $\mathrm{\text{SD}}=0.82$; between “Mildly easy” and “Very easy”).

Compared with traditional mental tools, most surgeons thought that the OCT-compatible instruments could provide more information during surgery. Nine of the 16 surgeons gave a high score for the custom instruments, as shown in Fig. 9(a). Out of 16 surgeons, 14 of them surgeons preferred using the OCT-compatible instruments in future practice, as shown in Fig. 9(b). The other two surgeons did not choose to use the custom instruments mainly due to the bad performance of the trabeculotome. They are concerned that there may be potential security issues.

Twelve of the 16 surgeons preferred using our digital MIOCT system over the traditional eyepiece microscope in glaucoma surgery. The most helpful part of the MIOCT system was the location of the APP and collector vessels during surgery. The reasons why 25% of surgeons preferred not to use MIOCT in the future are as follows. Surgeons performed the surgery through the eyepiece when using the traditional microscope. The visual direction was consistent with the surgical area. However, when using our MIOCT system, surgeons need to look up and observe the 3D display. The visual direction was perpendicular to the direction of the surgical area, which was different from surgeons’ usual practice and made them feel awkward.

4. Discussion

We introduced an external 3D display instead of the traditional eyepiece into the system which can provide higher resolution ($1920\times 1080$ pixel) and more surgery information than the built-in display.^20,21 Additionally, external 3D imaging as a new display method in eye surgery has been proven to be preferred by surgeons, which provides advantages in the field of view, image quality, and ergonomics compared with the traditional eyepiece and built-in display.^22,23,24 We will explore that in our future study.

We chose a 1310nm wavelength-swept source in our system mainly because of its better penetration depth than the 1060nm wavelength source in anterior segment imaging, inevitably having a lower imaging resolution. However, it has been proved in our studies that the system has adequate imaging resolution to locate important structures in surgery.

The visualization experiment provided a quantitative analysis to determine the impact on tissue imaging using different OCT-compatible instruments. In trabeculectomy and canaloplasty surgery, structures such as AAP and collector’s vessels can be easily located using our MIOCT system, which is critically important and requires much experience in traditional glaucoma surgery. Furthermore, surgeons suggested that the distance and angle measurement function was quite useful for the subsequent manipulation. Combined with a subjective feedback survey, it indicated that our MIOCT system and OCT-compatible instruments showed great advantages in glaucoma surgery. However, there was not an accurate value of success rate improvement compared with traditional glaucoma surgery. The hand feel of trabeculotome and refractive index information of forceps also need further optimization. We will focus on these questions in future studies.

In this study, we did not use real-time 3D OCT because: 1. Resolution, field of view, or frame rate of 3D OCT were limited by the swept speed of the laser; 2. Unlike vitreoretinal surgery, the instruments in canaloplasty and trabeculectomy were always surrounded by tissues, which decreased the visibility of tissues or instruments, as shown in Figs. 6(g) and 6(h) and 7(g) and 7(h).

5. Conclusions

Here, three kinds of OCT-compatible instruments were designed to conduct glaucoma surgery under the guidance of an MIOCT system. Glaucoma surgery experiments that include trabeculectomy and canaloplasty were successfully conducted. The surgeon feedback showed that OCT-compatible instruments greatly solved the shadowing problem in OCT-guided surgery and can provide more information during surgery. With the help of the MIOCT system and OCT-compatible instruments, surgeons can easily locate important tissue structures, confirm manipulations, and perform technique-challenging maneuvers.

Acknowledgments

The authors would like to acknowledge the financial support of the foundations: National Key R&D Program of China, Grant Nos. 2022YFC2404201; CAS Project for Young Scientists in Basic Research, Grant Nos. YSBR-067; The Gusu Innovation and Entrepreneurship Leading Talents in Suzhou City, Grant Nos. ZXL2021425; Jiangsu Science and Technology Plan Program, Grant Nos. BK20220263; National Key R&D Program of China, Grant Nos. 2021YFF0700503.

Conflicts of Interest

The authors declare “that there are no conflicts of interest relevant to this paper.