Glaucoma is a leading cause of preventable sight loss and, thus, a major public health concern.1 Elevated intraocular pressure (IOP) is the key modifiable risk factor, and to ascertain the mechanism of raised IOP, it is important to visualize the anatomy of the iridocorneal angle (ICA), the location of the trabecular meshwork and primary site of aqueous outflow. Gonioscopy is essential for patients with raised IOP or suspected or confirmed glaucoma to assess the status of the ICA and determine the most appropriate treatment option.2,3 Examination of the ICA is also important for the diagnosis of secondary glaucomas such as neovascular glaucoma.
Cross-sectional imaging such as anterior segment optical coherence tomography (AS-OCT) also enables assessment of angle anatomy; however, despite the high resolution of these imaging techniques, this cross-sectional technique does not provide a real image. Gonioscopy remains the gold standard as it provides a 360-degree view of the iris root, angle, and peripheral cornea, in addition to allowing assessment of trabecular meshwork pigmentation and detection of abnormalities such as peripheral anterior synechiae, neovascularization, or cyclodialysis clefts.4 Nevertheless, gonioscopy is underutilized, with a lack of time and lack of confidence in potential explanations.5 Gonioscopy is also subject to intraobserver and interobserver variability, dependent on the level of experience of the user.6,7
The Gonioscope GS-1 (Nidek Co.) is a desktop camera that acquires 360 degrees automated color photographs of the angle using a 16-mirror gonioprism (Fig. 1).8 Photographs are obtained using a soft-contact method, using a coupling gel to overcome total internal reflection. The method is like slit lamp gonioscopy; however, due to the gel, the camera gonioprism does not make direct contact with the eye. Local anesthetic eye drops are, however, recommended. Each gonioprism mirror captures a 22.5-degree segment or sectoral image. The device automatically captures 17 images at different focal planes from each 22.5-degree sector. The software automatically aims to select the best-focused image, but all images can also be manually explored. In addition, 360-degree fused, panoramic circular, and linear images can be generated. All images are acquired in less than 1 minute per eye. We refer to gonioscopic photographs as “images”, but this is distinct from “imaging,” such as with AS-OCT, which produces a digitally reconstructed visualization, as opposed to the gonioscopic photographs, which are an image of a real object.
;)
FIGURE 1: Responses to 10-point Likert scale on subjective comfort of automated gonioscopy.
Several studies have examined agreement in the identification of angle structures between trained observers examining photographs and novel automatic grading using images acquired using the Gonioscope GS-1.6,9 However, as far as we know, none have looked into the feasibility of automated gonioscopy in a clinical setting, for example, by examining patient tolerability, ease of use, and image quality. This study examined the feasibility of using the Gonioscope GS-1 to image the ICA, comparing patient satisfaction to traditional gonioscopy, with image quality also studied.
PATIENTS AND METHODS
This was a prospective study involving 25 patients recruited from the glaucoma clinic at the Princess Alexandra Eye Pavilion, Edinburgh, Scotland, a tertiary referral center. Methods were prospectively approved by the NHS research ethics committee, and all participants provided written informed consent before enrollment. Inclusion criteria included patients with primary open angle glaucoma, angle closure glaucoma, suspect glaucoma, or ocular hypertension. Patients who had undergone intraocular surgery within 4 weeks and those who were unable to cooperate with gonioscopy were excluded.
For each participant, a drop of oxybuprocaine hydrochloride 0.4% was instilled into each eye, and manual slit lamp gonioscopy was performed using a 4-mirror gonio lens (Volk G-4 Mirror Gonio Lens, Volk Optical, Inc.). Immediately following traditional gonioscopy, photography of the ICA was performed using the GS-1 camera. Both gonioscopy photography and slit lamp gonioscopy were performed by 2 glaucoma specialists, neither of whom had previous experience with the GS-1 camera, except for undergoing training using standard model eyes (Nidek Co.). For each patient, the same clinician performed slit lamp gonioscopy and then acquired the gonioscopy photograph. If the clinician deemed the series of gonioscopic photographs to be poor quality, up to 2 further repeat acquisitions were permitted, with the best quality series of images selected. Immediately after the photographs had been acquired, participants were verbally asked to rate how comfortable they found automated gonioscopy, using a 10-point Likert scale (1–10), where ‘1’ was extremely comfortable and ‘10’ extremely uncomfortable. Participants were also asked whether they preferred undergoing manual or automated gonioscopy or whether they had no preference. At the same time, the clinician graded the ease of acquisition for each patient, also using a 10-point Likert scale where 1 was ‘extremely easy’ and 10 was ‘extremely difficult’ to obtain the photographs. Clinicians were asked to consider factors such as the ease of positioning the patient, use of the camera joystick and software, managing the patients’ lids, and maneuvering for accurate centration of image.
Following photography, all images were reviewed by a grader who had not been present during image acquisition. The Nidek GS-1 takes multiple color images of the ICA to cover the entire 360 degrees of the interface at many different focal planes with limited depth of field. Each sectoral image covers a 22.5 degrees wide sector. The software automatically selects the image assumed to be the most in focus at the ICA. Graders were able to access the entire exam acquisition and manually adjust any sectoral image where the automatic selection was poorly focused on removing the possibility of inaccurate automatic image selection by the software. Image quality was graded as clearly visible (grade 1), partially visible or slightly blurred (grade 2), or not visible (grade 3).10,11
RESULTS
The demographic and clinical characteristics of the groups are displayed in Table 1. Forty-six eyes of 25 participants were included. Participants had an average age of 69.8±14.6 y. 13 of 25 (52%) were female. Thirty-four of the 46 eyes (74%) had a diagnosis of primary open angle glaucoma, 10 (22%) had angle closure glaucoma, and 2 (4%) with pseudoexfoliative glaucoma.
TABLE 1 -
Characteristics of Participants
| Variable |
Mean (SD) |
Range |
Median |
Quartile 1 |
Quartile 3 |
| Age |
| All |
69.8 (14.6) |
28 to 85 |
73 |
62 |
80 |
| Male (50%) |
68.3 (15.3) |
28 to 85 |
72.5 |
62 |
77.2 |
| Female (50%) |
71.5 (13.5) |
50 to 93 |
78 |
62 |
81.5 |
| Clinic GAT (mm Hg), n=46 |
14.8 (4.8) |
7 to 26 |
14 |
11 |
18.7 |
| Mean deviation (db), n=46 |
−8.46 (7.5) |
0 to −27.06 |
−5.91 |
−1.58 |
−13.86 |
Seventeen of the 25 participants (68%) viewed automated gonioscopy as “extremely comfortable”, the remainder reporting it to be “comfortable” (Fig. 1). Ten of the 25 participants (40%) preferred automated gonioscopy, while 13 (52%) were equivocal and 2 (8%) reported a preference for slit lamp gonioscopy. These results were similar between the clinicians (Fig. 2).
FIGURE 2: Participants’ preference for manual or automated gonioscopy, with the outer circle representing patients where clinician 1 performed automated gonioscopy and the inner circle, clinician 2.
The ease of the use of the camera was graded by the clinicians and was variable but never reported as “difficult” or “extremely difficult”. Clinician 1 found image acquisition easy or somewhat easy from the first patient. Clinician 2 found image acquisition somewhat difficult initially but by the eighth patient was reporting the procedure to be easy (Fig. 3). There were no technical difficulties encountered when acquiring any of the photographs.
FIGURE 3: Plot of each clinician grading of ease of image acquisition (0-10) in each patient, chronologically.
Eleven of the 46 eyes (24%) required up to 2 further repeat acquisitions to obtain a satisfactory image. This was equally distributed between the 2 clinicians. Nine participants required 1 attempt in 1 eye, and 2 on the other, and of these participants all but 1 graded the comfort of automated gonioscopy as “extremely comfortable” with most showing no preference for slit lamp or automated gonioscopy. One participant required 2 acquisitions for each eye; however, this participant still rated the automated gonioscopy as ‘extremely comfortable’ and reported no preference between automated and slit lamp gonioscopy. Though these repeat acquisitions were performed to gain better quality images, the ease of acquiring the photographs were still rated as ‘very easy’.
For 21 of the 46 eyes (46%) clear images were obtained over 360 degrees of the ITA, meaning all 16 sectoral images provided clear visualization of the angle structures (Fig. 4). In the remainder of eyes, there were some segments of the ICA where structures were either not clearly visible or only partially visible. Only 1 eye (2.2%) had no parts of the ICA clearly visible. This participant with difficult-to-control angle closure had 2 attempts at automated gonioscopy in this eye and declined a further attempt.
FIGURE 4: Percentage of images graded as grade 1, 2, and 3 quality for each eye included in the study.
Thirty-four of the 46 eyes imaged (73.9%) had at least 1 clear 22.5-degree sectoral image where the ICA structures were clearly visible in each of the 4 quadrants, while 43 of the 46 eyes (93.5%) had at least 1 clear sectoral image in at least 3 quadrants. The mean number of images of good quality was 13.1±3.9 per eye out of a possible 16. In total, 736 sectoral images were obtained, with 602 (81.8%) Grade 1, 63 (8.6%) Grade 2, and 71 (9.6%) Grade 3. Figure 5 shows examples of varying image qualities from a single eye.
FIGURE 5: A, Grade 1, where ICA structures are visible. B, Grade 2, where ICA structures are partially visible. C, Grade 3, where ICA structures are not visible. D, A 360-degree stitched linear view, showing varying image quality in a single eye. Here, the inferotemporal ICA structures are not visible. A peripheral iridotomy is visible superiorly.
Participants were asked if they had any additional comments about their experience of automated gonioscopy. One of the patients who reported a preference for manual gonioscopy explained this was only due to it being a quicker process. Two participants commented that less pressure was felt during automated compared with manual gonioscopy, resulting in a preference for automated testing.
CONCLUSION
To the best of our knowledge, this is the first study examining the feasibility of automated gonioscopy in a clinical setting, where patients imaged reflect a typical glaucoma outpatient clinic. Patients found gonioscopic photography with the GS-1 camera comfortable, with most reporting it to be equivalent or preferred to slip lamp gonioscopy.
Though the clinicians performing gonioscopic photography were new to the technology, overall image quality was good, a key criterion for assessing the clinical utility of a new image acquisition modality. For almost 50% of the eyes imaged, good-quality images were obtained over 360 degrees, and on average, 13 of 16 images obtained for each eye showed the angle structures clearly. A previous study evaluating image quality with an older model reported that 36.9% of images acquired had no discernible ICA structures visible, which compares with only 9.6% in this study.10 The previous study assessed image quality from a prototype NIDEK GS-1 camera and found improvements were needed in order for the device to be used successfully in a clinical setting.10 Our results show a significant improvement in image quality, perhaps partly due to the device taking images from 17 different focal planes in each segment, compared with 4 in the prototype.
Automated gonioscopy has potential advantages over conventional gonioscopy, including that it can be operated by a technician and that it provides color images of the ICA. This enables a record of the ICA to be obtained for the electronic patient record without the clinician needing to be present and is, therefore, of potential value for virtual clinics and asynchronous telemedicine. Though AS-OCT also provides an image of the ICA, automated gonioscopy enables photography of 360 degrees of the ICA and allows assessment of ICA pigmentation and identification of other features such as iris and ICA neovascularization or Sampaolesi line, a feature of pigment dispersion or pseudoexfoliation. AS-OCT has not replaced gonioscopy as the gold standard method of ICA assessment, in part due to these limitations but also as there is a tendency for false positives when AS-OCT is used to detect angle closure. Angle closure disease is defined by the presence of irido-trabecular contact over 180 degrees or more on dark-room gonioscopy,12 meaning that assessment of 360 degrees of the ICA is important for diagnosis. While the agreement between dark-room gonioscopy and images captured using the GS-1 remains to be fully elucidated, the ability to obtain 360-degree photographs of the ICA allows assessment of whether the 180-degree irido-trabecular contact cut-off has been met.
While good-quality photographs of the ICA can be obtained using gonioscopy and slit lamp-mounted anterior segment cameras, this method requires considerable skill and is time-consuming. Moreover, obtaining photographs of 360 degrees of the ICA is challenging, meaning the technique is more commonly used for capturing images of isolated ICA abnormalities or of eyes post surgery. In contrast, automated gonioscopic photography could feasibly be used in a virtual clinic setting, for example, for referral refinement or for periodic assessment of the ICA during glaucoma monitoring. Virtual glaucoma clinics have been implemented in some settings to provide an efficient, high-throughput method of referral refinement or monitoring of low-risk patients. Though patients do not have a face-to-face consultation with an ophthalmologist, they report high levels of satisfaction with the virtual clinic experience.13 Virtual clinics are typically run by ophthalmic technicians, who perform tests, including tonometry, automated perimetry and optical coherence tomography of the retinal nerve fiber layer, optic nerve head, and macula. While some virtual clinics also capture AS-OCT images, gonioscopy is not performed. Automated gonioscopic photography offers an alternative option for assessment of the ICA in this setting.
Other potential applications of color ICA photography include use in community optometry, allowing identification of angle closure, peripheral anterior synechiae, and secondary glaucomas such as neovascular glaucoma. Community optometrists tend to use Van Herick’s technique to assess limbal chamber depth, and although this has a high sensitivity (84.9%) and high specificity (89.3%) for detecting occludable angles, it does not provide direct visualization of the ICA.14 Further potential advantages of automated compared with traditional gonioscopy include a role in teaching,5 and that the user can zoom in on any abnormalities, permitting a significantly higher magnification compared with that of slit lamp gonioscopy. Photographs also allow comparison over time, useful for longitudinal follow-up and are useful for postoperative documentation of microinvasive glaucoma surgical devices.3
Automated gonioscopic photography also offers the potential for reducing variation in the quality of ICA assessment. Inter-observer and intra-observer variability in the interpretation of gonioscopic images remains, at best, fair.6 Deep learning algorithms are currently being developed to identify structures in the ICA from goniophotographs, providing a means to automatically segment ICA anatomy and highlight areas of abnormality.15
Overall, we found clinicians performing automated gonioscopy reported high ease of use, despite having no previous experience with the automated gonioscopy camera. It is, though, important to emphasize that the clinicians were previously experienced in gonioscopy, and ophthalmic technicians who are unfamiliar with gonioscopy may have a steeper learning curve, producing a greater barrier to use. Previous studies have shown ophthalmic technicians are keen to embrace new technologies but value feedback on imaging findings.16 As technicians are unlikely to be familiar with normal ICA appearance, providing training on the anatomy of the ICA, its importance in glaucoma, and providing examples of ICA abnormalities, would likely improve technicians' engagement with the adoption of the new technology.
There are several limitations to this study. The sample size was relatively small; however, participants were consistent in reporting automated gonioscopy to be comfortable. The inter-clinician analyses are also subject to small sample size and were examined in this way to highlight any substantial difference in subjective evaluations by the participant and examiner. Secondly, there was the potential for selection bias as patients who declined to participate may have been more likely to report discomfort with eye examinations in general. Regardless, patient cooperation and ability to withstand gonioscopy are variable in the population, and factors such as poor posture, pain, and strong blink reflex can all contribute to how easy gonioscopic images are to acquire with both manual and automated techniques. In addition, ease of use and proportion of images of good quality may have been different if the study was performed using ophthalmic technicians. Overall, though gonioscopic photography provided good-quality images, it was often not possible to obtain clear images of the ICA over 360 degrees at the first attempt. However, as the procedure was well tolerated by patients, repeat imaging could be performed. In these circumstances, rather than repeat the entire series of image captures, any poor-quality images can be highlighted by the user, and the GS-1 camera can selectively photograph the sectors of the ICA that were not successfully imaged at the first attempt.
In conclusion, our results indicate that automated gonioscopy is feasible and that clinicians usually find images easy to obtain. Most of the acquired images were of good quality, with most eyes found to have over two-thirds of the angle structures clearly visible, despite the users being new to the GS-1 device. Those acquiring the images described a slight learning curve along with the normal variability in the cooperation of patients. Finally, patients found the procedure of image acquisition comfortable, with only 8% preferring traditional gonioscopy to automated gonioscopic photography.
REFERENCES
1. Tham Y-C, Li X, Wong TY, et al. Global prevalence of glaucoma and projections of glaucoma burden through 2040: a systematic review and meta-analysis. Ophthalmology. 2014;121:2081–2090.
2. Azuara-Blanco A, Bagnasco L, Barbosa Breda J, et al. European Glaucoma Society Terminology and Guidelines for Glaucoma, 5th Edition. Br J Ophthalmol. 2021;105:1–169.
3. Cutolo CA, Bonzano C, Scotto R, et al. Moving beyond the slit-lamp gonioscopy: challenges and future opportunities. Diagnostics. 2021;11:2279.
4. Desmond T, Tran V, Maharaj M, et al. Diagnostic accuracy of AS-OCT vs gonioscopy for detecting angle closure: a systematic review and meta-analysis. Graefes Arch Clin Exp Ophthalmol. 2021;260:1–23.
5. Feng R, Luk SMH, Wu CHK, et al. Perceptions of training in gonioscopy. Eye. 2019;33:1798–1802.
6. Matsuo M, Mizoue S, Nitta K, et al. Intraobserver and interobserver agreement among anterior chamber angle evaluations using automated 360-degree gonio-photos. PLoS One. 2021;16:e0251249–e0251249.
7. Riva I, Micheletti E, Oddone F, et al. Anterior chamber angle assessment techniques: a review. J Clin Med. 2020;9:3814.
8. Wong SH, Tsai JC. Telehealth and screening strategies in the
diagnosis and management of glaucoma. J Clin Med. 2021;10:3452.
9. Matsuo M, Pajaro S, De Giusti A, et al. Automated anterior chamber angle pigmentation analyses using 360° gonioscopy. Br J Ophthalmol. 2020;104:636–641.
10. Teixeira F, Sousa DC, Leal I, et al. Automated gonioscopy photography for iridocorneal angle grading. Eur J Ophthalmol. 2020;30:112–118.
11. Murakami Y, Wang D, Burkemper B, et al. A population-based assessment of the agreement between grading of goniophotographic images and gonioscopy in the Chinese-American eye study (CHES). Invest Ophthalmol Vis Sci. 2016;57:4512–4516.
12. Gedde SJ, Chen PP, Muir KW, et al. Primary angle-closure disease, preferred practice pattern. Ophthalmology. 2021;128:PP30–PP70.
13. Court JH, Austin MW. Virtual glaucoma clinics: patient acceptance and quality of patient education compared to standard clinics. Clin Ophthalmol. 2015;9:745–749.
14. Baskaran MDODNB, Oen FTSF, Chan Y-HP, et al. Comparison of the scanning peripheral anterior chamber depth analyzer and the modified van herick grading system in the assessment of angle closure. Ophthalmology. 2007;114:501–506.
15. Peroni A, Cutolo CA, Pinto LA, et al. A Deep Learning Approach for Semantic Segmentation of Gonioscopic Images to Support Glaucoma Categorization. In: Papież B, Namburete A, Yaqub M, Noble J, eds.
Medical Image Understanding and Analysis, Vol. 1248. Communications in Computer and Information Science. Springer, Cham; 2020:373–386.
16. Gunn PJG, Marks JR, Au L, et al. Virtual clinics for glaucoma care – Patients’ and clinicians’ experiences and perceptions: a qualitative evaluation. Eye. 2022;36:209–218.
