Glaucoma Fellows Research Articles

Diagnostic Performance of the Offline Medios Artificial Intelligence for Glaucoma Detection in a Rural Tele-Ophthalmology Setting

PurposeThis study assesses the diagnostic efficacy of offline Medios Artificial Intelligence (AI) glaucoma software in a primary eyecare setting, using non-mydriatic fundus images from Remidio's Fundus-on-Phone (FOP NM-10). AI results were compared with tele-ophthalmologists' diagnoses and with a glaucoma specialist's assessment for those participants referred to tertiary eyecare hospital. DesignProspective, cross-sectional study Participants303 participants from 6 satellite vision centers of a tertiary eye hospital MethodsAt the vision center, participants underwent comprehensive eye evaluations, including clinical history, visual acuity measurement, slit lamp examination, intraocular pressure measurement, and fundus photography using the FOP NM-10 camera. Medios AI-Glaucoma software analysed 42-degrees disc-centric fundus images, categorizing them as normal, glaucoma, or suspect. Tele-ophthalmologists who were glaucoma fellows with a minimum of 3 years of ophthalmology and 1 year of glaucoma fellowship training, masked to AI results, remotely diagnosed subjects based on the history and disc appearance. All participants labelled as disc suspects or glaucoma by AI or tele-ophthalmologists underwent further comprehensive glaucoma evaluation at the base hospital, including clinical examination, Humphrey visual field analysis (HFA), and Optical Coherence Tomography (OCT). AI and tele-ophthalmologist diagnoses were then compared with a glaucoma specialist's diagnosis. Main Outcome MeasuresSensitivity and Specificity of Medios AI ResultsOut of 303 participants, 299 with at least one eye of sufficient image quality were included in the study. The remaining 4 participants did not have sufficient image quality in both eyes. Medios AI identified 39 participants (13%) with referable glaucoma. The AI exhibited a sensitivity of 0.91 (95% CI: 0.71 - 0.99) and specificity of 0.93 (95% CI: 0.89 - 0.96) in detecting referable glaucoma (definite perimetric glaucoma) when compared to tele-ophthalmologist. The agreement between AI and the glaucoma specialist was 80.3%, surpassing the 55.3.% agreement between the tele-ophthalmologist and the glaucoma specialist amongst those participants who were referred to the base hospital. Both AI and the tele-ophthalmologist relied on fundus photos for diagnoses, while the glaucoma specialist's assessments at the base hospital were aided by additional tools such as HFA and OCT. Furthermore, AI had fewer false positive referrals (2 out of 10) compared to the tele-ophthalmologist (9 out of 10). ConclusionMedios offline AI exhibited promising sensitivity and specificity in detecting referable glaucoma from remote vision centers in southern India when compared with teleophthalmologists. It also demonstrated better agreement with glaucoma specialist’s diagnosis for referable glaucoma participants.

Complications Following XEN45 Gel Stent and Glaucoma Drainage Device Implantation During Glaucoma Fellowship.

Purpose XEN45 Gel Stent and glaucoma drainage device (GDD) implantation is safe and effective for glaucoma treatment and should be taught during glaucoma fellowship training. However, complications may still occur, with potentially sight-threatening consequences. The purpose of this study is to describe the management of complications following a series of XEN45 Gel Stent and GDDsurgeries performed over the course of glaucoma fellowship training. Methods This is a retrospective case series of XEN45 Gel Stent surgeries performed on 16 eyes and GDD surgeries performed on seven eyes. Patient demographics, disease characteristics, and complications are reviewed. The intra- and postoperative course of five select cases with complications are described in detail. Results The most frequent complications following XEN45 implantation were transient hypotony (10 eyes, 63%), reduced visual acuity (VA) (five eyes, 31%), choroidal effusion (three eyes, 19%), hyphema (two eyes, 13%), and intraocular pressure (IOP) spike (two eyes, 13%). Thirteen eyes (81%) required bleb needling, and three eyes (19%) required XEN45 replacement. Complications following GDD implantation included hypotony (three eyes, 43%), reduced VA (two eyes, 29%), choroidal effusion (two eyes, 29%), IOP spike (two eyes, 29%), implant exposure (two eyes, 29%), and shallow anterior chamber (one eye, 14%). Three eyes (43%) required revision or explantation with a secondary glaucoma surgery. One choroidal effusion following XEN45 surgery and one following GDD surgery were hemorrhagic choroidal effusions requiring surgical drainage. Conclusion Significant and potentially sight-threatening complications may occur following XEN45 Gel Stent and GDD implantation performed over the course of fellowship training. Glaucoma fellows should be ably equipped to recognize, diagnose, and manage these complications both intra- and postoperatively.

Outcomes of Ahmed glaucoma valve surgery performed by glaucoma fellows - A large retrospective study from Eastern India.

Teaching and training of glaucoma fellows on the technique of Ahmed glaucoma valve (AGV) implantation surgery is very important. The purpose of this study was to describe the clinical outcomes and complications of AGV surgery performed by glaucoma fellows of a tertiary eye center from eastern India. This was a retrospective study based on electronic medical records. Thirty-five eyes operated with AGV by five glaucoma fellows from January 2016 to November 2020 were included. Best-corrected visual acuity, intraocular pressure (IOP), and number of antiglaucoma medications (AGMs) were the study outcome measures. The mean (±standard deviation) age of the patients was 40.5 (±19.7) years. The follow-up period was 24.37 (±13.01) months. Both complete success (IOP <21 mmHg without additional AGM) and qualified success (IOP <21 mmHg with an additional AGM) were achieved in eight eyes each (22.8%). Failure of the surgery was noted in 19 eyes (54.3%) as there was a need for repeat surgery for IOP control in 12 eyes and a persistent IOP spike (IOP >21 mmHg with AGM beyond 3 months) in seven eyes. Despite a high rate of failure of the primary surgery, with the consultants' intervention, there was statistically significant improvement in vision and IOP in all patients ( P < 0.01) at the last follow-up. AGV implantation is a challenging surgical skill to be acquired by the glaucoma fellows.

Glaucoma Fellows-in-Training Recent Surgery Trends

The volume of microinvasive glaucoma surgery (MIGS) has increased dramatically in recent years, from 31 059 in 2013 to 69 420 in 2018. We investigated the impact of this trend on trainees by determining the proportion of glaucoma surgeries performed by fellows-in-training comprised by MIGS, trabeculectomies, and aqueous shunts. Retrospective analysis. Fellows-in-training at Glaucoma Fellowship Programs certified by the Association of University Professors of Ophthalmology Fellowship Compliance Committee (AUPO-FCC) METHODS: We analyzed aggregate summaries of surgeries performed by fellows as reported to the AUPO-FCC from the academic years (AYs) beginning in 2014 through AY 2020. Each report lists the average number of procedures performed per surgery type per fellow. We combined these averages to create a sum "average number surgeries performed" across glaucoma surgeries and computed the proportion that each surgery type (MIGS, trabeculectomies, and aqueous shunts) represented within the total average number of surgeries performed per year. Average number of procedures performed for each surgery type as well as the proportion that each surgery type (MIGS, trabeculectomies, and aqueous shunts) represented within the total average number of procedures performed per year. Average number of MIGS performed is significantly greater in later years compared with earlier years (P < 0.001). The average number of trabeculectomies performed between AYs 2014 and AY 2020 ranged from 21.8 to 31.9 and decreased, on average, by- 0.80 year-to-year. The average number of aqueous shunts performed between AY 2014 and AY 2020 ranged from 44.7 to 49.5, with an average increase of+ 0.8 year-to-year. The total average number of procedures performed (across all 3 surgical subtypes) increased on average by+ 4.8 procedures each year. Since 2014, fellows are performing increasing numbers of MIGS procedures, whereas the total number of trabeculectomies and aqueous shunt surgeries performed each year remain similar, resulting in a net increase in total number of procedures performed per fellow each year. This suggests the increase in MIGS is not associated with a substantial decline in trabeculectomies or aqueous shunts performed by glaucoma fellows. Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Outcomes of "Complex" Cataract Surgeries Performed by Long-Term Glaucoma Fellows in a Tertiary Eye Centre from Eastern India.

To evaluate the outcomes of operating on "complex cataracts" by the glaucoma fellows. This was a retrospective study done at a tertiary referral eye care centre in eastern India. After obtaining IRB approval, a retrospective chart review of all patients who underwent "complex" cataract surgery by one of four long-term (2 years) glaucoma fellows between January 2016 and November 2020 was conducted. 'Complex' was defined as cataracts complicated with pseudoexfoliation syndrome, phacodonesis with or without blunt ocular trauma, posterior polar cataract, small pupil, co-existent corneal opacity or uveal coloboma, post-glaucoma filtering surgery, post-vitreoretinal surgery, co-existent glaucoma or post-laser iridotomy and monocular patients. Out of a total of 677 eyes done by the glaucoma fellows during the study period, 83 eyes had complex cataract surgery and completed the six-week post-operative follow-up. Intraoperative surgical complications like posterior capsular rent or vitreous loss were noted in 36 of the cases. Thirty of the eyes were left aphakic. Despite a high rate of complications, the LogMAR best-corrected visual acuity (mean ± standard deviation) improved from the preoperative level of 1.7 (±0.5) to 1.0 (± 0.8) at post-operative six weeks, significant at p < 0.001. As far as the surgeon's experience was concerned-less than or more than a year since joining the fellowship-there was statistically no difference in the final visual acuity. The group with greater experience had shorter surgical time and lesser complications though this difference was not statistically significant. This is the first study in the literature reporting the outcomes of "complex" cataract surgery performed by glaucoma fellows. Though high rates of postoperative complications were noted in this study, the mean best-corrected visual acuity improved significantly in all eyes after the surgery.

Gonioscopy and Angle-Based Glaucoma Surgical Training on Human Eyes in the Wet Lab.

Human corneoscleral rims provide a low cost model and resource to enhance angle based glaucoma surgical training skills. To understand the feasibility, utility, and efficacy of a low-cost model for performing anterior chamber angle gonioscopy and micro-surgery on human corneoscleral tissues in wet lab on resident skill enhancement. Post-keratoplasty residual corneoscleral rims and/or the expired donor corneoscleral buttons were included for the purpose of this study. Initially, inverted rims were used to demonstrate the basic angle anatomy on a slit lamp bio-microscope. Central hazy cornea was then replaced with an artificial cornea and an artificial anterior chamber to simulate indirect gonioscopy, direct gonioscopy, goniotomy, and other angle-based surgeries. Direct gonioscopy and goniotomy exercises (as a means to simulate intraoperative gonioscopy and other angle-based surgeries respectively) were evaluated quantitatively, and the rest qualitatively. A total of 65 residents were included in the exercise. The mean age was 26.69±1.74 years. Amongst these, 55 ophthalmology residents performed the basic examination, that is, slit lamp mounted angle anatomy examination (n=55) and artificial chamber mounted indirect gonioscopic examination through artificial cornea (n=55). In the second exercise, 10 glaucoma fellows performed the direct gonioscopic and the goniotomy exercises on artificial chamber. The fellows demonstrated a statistically significant improvement in interpretation and surgical skill enhancement concurrent with reduction in time taken with increasing number of attempts ( P <0.05). In the third simulation exercise, micro-stent placement within the Schlemm's canal, subconjunctival space, and the suprachoroidal space was re-produced successfully on multiple occasions. The residual human corneoscleral rims proved to be an important resource for teaching both the basic examination skills and those required for transitioning to advanced intraoperative gonioscopy and goniotomy skills. These low cost, easy to adapt models can be incorporated within the curriculum for improved glaucoma training of both the residents and the glaucoma fellows.

Selective laser trabeculoplasty (SLT) performed by optometrists—enablers and barriers to a shift in service delivery

Background/objectivesTo explore the acceptability, training requirements, enablers and barriers of optometrist-delivered SLT.Subjects/methodsOptometrists, fellowship and consultant grade ophthalmologists, hospital managers and patients were interviewed using pre-defined topic guides. Interviews were audio-recorded, transcribed, and subjected to thematic analysis. Overarching themes were defined by the study aims and the topic guides; subthemes were derived from the interview data.ResultsSixty-six participants (three managers, eight glaucoma specialist consultant ophthalmologists, seven clinical glaucoma fellows, 12 optometrists (two of them performing SLT), two ophthalmic nurses and 34 patients) participated in the study. Overarching themes (and subthemes) were: necessity of non-medical SLT delivery, clinical practice and training, advantages, disadvantages, concerns, challenges, community delivery of SLT, patient values and other healthcare professionals that could also deliver SLT.ConclusionsCertain clinical pre-requisites, such as gonioscopy and independent prescribing rights, were perceived as necessary for undertaking SLT training. An optometrist-delivered SLT service was expected to benefit the NHS, but there was an identified need of a standardised training scheme and robust governance. Patients were accepting of an optometrist-delivered SLT service in the hospital eye service.

Evolution of a Glaucoma Fellow's Surgical Training: Improvements in Tube Shunt Case Times during the Academic Year

Abstract Purpose The aim of the study is to report changes in tube shunt placement surgical case times for glaucoma fellows during the course of the academic year. Patients and Methods Electronic health records were retrospectively reviewed to determine patient demographics, surgical case times (defined as procedure start time to procedure end time), and glaucoma fellow involvement. Only cases with a glaucoma fellow as the primary surgeon were included. Operative case times were compared by first and second halves of the academic year (beginning in July and ending in June) using a two-tailed t-test. Results Five hundred and seventy-three individual tube shunt surgeries (385 Ahmed, 188 Baerveldt) performed by 28 glaucoma fellows (17 females, 11 males) at Duke University Eye Center and University of North Carolina Medical Center were included. Overall, case times were significantly shorter in the second half of the academic year as compared with the first (55.3 ± 17.1 minutes vs. 61.0 ± 17.4 minutes, p &lt;0.001). Both male (57.3 ± 16.8 minutes vs. 63.2 ± 18.6 minutes, p = 0.008) and female (53.5 ± 17.3 minutes vs. 59.3 ± 16.4 minutes, p = 0.003) fellows demonstrated shorter case times over the academic year; additionally, female fellows trended toward shorter case times than male fellows in both the first half (p = 0.072) and second half (p = 0.053) of the academic year. Fellows also exhibited shorter case times with both Ahmed implants (54.1 ± 16.2 minutes vs. 59.3 ± 15.8 minutes, p = 0.002) and Baerveldt implants (57.8 ± 18.9 minutes vs. 64.2 ± 20.0 minutes, p = 0.025) cases over the academic year. Baerveldt case times were significantly longer than Ahmed cases in the first half (p = 0.028) and trended toward being longer than Ahmed cases in the second half (p = 0.070). Conclusion Across 5 years at two academic institutions, glaucoma fellows had shorter primary tube shunt surgical case times in the second half of the academic year. These findings reflect improvement in surgical efficiency throughout glaucoma fellowship. These findings should be taken into consideration when scheduling trainee surgeries at academic medical centers at different points in the academic year.

Seven- and eight-year trends in resident and fellow glaucoma surgical experience.

PurposeThis study evaluated US resident and fellow glaucoma surgical experience over a 7- and 8-year period, respectively.MethodsUS ophthalmology resident glaucoma case logs from 2009 to 2016 and glaucoma fellow case logs from 2008 to 2016 were reviewed. The number of total, average, median, and minimum cases, along with percentile data for filtering surgery, shunting surgery, and novel procedures were analyzed for year-to-year trends.ResultsAmong residents training from 2009 to 2016, there was a decline in the average number of primary filtering surgeries by 20%, from 6.0±5.0 to 4.8±4.0 cases, with a concurrent increase in average primary glaucoma drainage implant (GDI) surgeries by 40%, from 4.5±4.0 to 6.3±5.0 cases, which represented an increase of 6.5% per year, P<0.001. Glaucoma fellow data from 2008 to 2015 demonstrated a decline in average primary trabeculectomy surgeries by 3.7%, from 30.1 to 29.0 cases, but then increased to 32.1 cases in 2015–2016. There was an increase in average GDI surgeries by 57.9%, from 30.2 to 47.7 cases for fellows over this 8-year period, which represented a 5.9% increase per year, P<0.001. There was no microinvasive glaucoma surgery (MIGS) or other novel procedures reported in the Accreditation Council for Graduate Medical Education (ACGME) resident case logs. From 2008 to 2016, Association of University Professors of Ophthalmology (AUPO) glaucoma fellow case logs demonstrated an increase in endocyclophotocoagulation (ECP) from 2.3 to 5.4 cases, a 15.5% increase per year, P<0.001.ConclusionsOver a 7–8-year period, trainee surgical experience with GDI surgery has steadily increased, whereas surgical experience with trabeculectomy has been fluctuating and may be on a downward trend. Educators should be aware of shifts in trainee surgical experience in order to maintain appropriate training experience.

‘Van Herick Plus’: a modified grading scheme for the assessment of peripheral anterior chamber depth and angle

AimTo evaluate the accuracy of a new, modified grading scheme involving a short vertical slit beam, at the inferior angle for peripheral anterior chamber depth (PAC) and angle estimation and...

Primary Trabeculectomy Outcomes by Glaucoma Fellows in a Tertiary Hospital in Brazil.

To examine outcomes of trabeculectomy with mitomycin C for uncontrolled glaucoma when performed by glaucoma trainee surgeons. Retrospective case series of patients who underwent trabeculectomy with mitomycin C. Primary outcome was to assess the rate of failure, which was defined as intraocular pressure in 2 consecutive visits >18 or <5 mm Hg or intraocular pressure reduction <30% from baseline, additional glaucoma surgery, or loss of light perception. The secondary outcome was to investigate risk factors for failure and surgical complications. One hundred forty-three eyes from 126 glaucoma patients underwent primary trabeculectomy between 2013 and 2014 at University of Campinas. Mean follow-up time was 1.47±0.96 years. Kaplan-Meier analysis showed cumulative survival rates of 70.1%, 62.5%, and 57.8%, after 1, 2, and 3 years, respectively. Multivariate analysis with generalized estimating equations revealed that neovascular glaucoma [odd ratios (OR): 5.05, P=0.043], suture lysis (OR: 8.89, P=0.011), and early bleb leak (OR: 4.63, P=0.011) were risk factors for failure. A success rate of approximately 60% was obtained 3 years after primary trabeculectomy performed by trainees. Patients with neovascular glaucoma, who underwent suture lysis and who presented an early bleb leak had an increased risk for failure during follow-up.

Real‐life Experience with glaucoma surgical training using SOS eyes

SummarySurgical training in Europe is very different according to each region. National regulations, different settings and allocated resources allow for a great heterogeneity in surgical opportunities. The equation of providing adequate training and performance levels to the surgeon while upholding the best possible practices of maximizing patient safety is a difficult task. Throughout Europe, surgeons are increasingly resorting to solutions that allow them to train in simulated scenarios before being confronted with a real life scenario. From wet‐labs to electronic simulations, there is a wide range of solutions that could be adapted to each reality. In this section of the symposium we look at real life experiences using specifically designed advanced ‘dry lab eye’ models (Simulated Ocular Surgery, or SOS, eyes) from the perspective of residents, glaucoma fellows and consultant trainers/supervisors. Videos and first hand experiences will be shared with the audience with useful tips, practical advice and potential drawbacks presented and described by a young qualified consultant trainer from one of the largest and oldest ophthalmic teaching hospitals in the U.K.

Comparing glaucomatous disc change using stereo disc viewing and the MatchedFlicker programme in glaucoma experts and trainees

Background/aimsThe objective of this study is to evaluate the accuracy and speed of trainees and experienced glaucoma specialists using the MatchedFlicker software against the manual examination of stereoscopic disc photographs...

Trabeculectomy Outcomes by Supervised Trainees in a Veterans Affairs Medical Center.

To evaluate the outcomes of trabeculectomy performed in an ophthalmology training program. Retrospective study. A total of 160 patients undergoing trabeculectomy performed by a resident or fellow under attending supervision. Trabeculectomy surgeries performed by a supervised resident or fellow surgeon between October 2000 and April 2010 were reviewed. Success was considered to be complete or partial if intraocular pressure (IOP)-lowering medications were not or were required to achieve IOP≤21 mm Hg, respectively. Failure was defined as IOP>21 mm Hg on 2 consecutive visits, loss of light perception vision, IOP≤5 on 2 consecutive visits with associated visual acuity loss of ≥2 lines, or need for surgical intervention. Trabeculectomy survival was determined using Kaplan-Meier analysis through 60 months of follow-up. Final IOP, success/failure rate. Complete success was achieved in 65 patients (41%). The average final IOP of this group was 9.1±3.7 mm Hg. Qualified success was achieved in 56 patients (35%). The average final IOP of this group was 11.5±6.4 mm Hg. At 60 months, the estimated cumulative probability of survival for complete and qualified successes was 28.9% and 63.7%, respectively. Among patients undergoing trabeculectomy by supervised residents or glaucoma fellows in a large Veterans Affairs Medical Center, IOP reduction was significant and similar to published studies. Trabeculectomy remains a successful intervention to lower IOP, with satisfactory success rates in the hands of trainee surgeons.

Comparing Glaucomatous Disc Change Using Stereo Disc Viewing and the MatchedFlicker Software Program in Ophthalmologists-in-Training

To compare the accuracy and speed of using the computerized MatchedFlicker software program (EyeIC Inc, Narberth, Pennsylvania, USA) to evaluate glaucomatous optic disc change against the traditional gold standard of manually examining stereoscopic disc photographs. A prospective evaluation of diagnostic technology. Two resident ophthalmologists and 1 glaucoma fellow at the University of Florida independently evaluated 140 image pairs from 100 glaucomatous/ocular hypertensive patient eyes using a handheld stereo viewer and the MatchedFlicker program. Fifty had progression to glaucoma as determined by the Ocular Hypertension Treatment Study (OHTS) Optic Disc Reading Group and the OHTS Endpoint Committee in the OHTS, and 50 more had photographs taken a few minutes apart, which were negative controls with no progression. Twenty photograph pairs from each group were duplicated to determine reviewer variability. Photographs were examined in alternating blocks of 70 photograph pairs for each method, with the starting viewing method randomized. Reviewer accuracy and time to review for each method were measured. Using the handheld stereo viewer, the reviewers correctly identified progression or nonprogression in 76.0% of the slide pairs. Using the MatchedFlicker software, 87.6% were correctly identified (P= .011). Evaluator speed averaged 34.1 seconds per image pair with the stereo viewer vs 24.9 seconds with the MatchedFlicker program (P= .044). Overall, Flicker was significantly more specific but less sensitive than stereo slides. Trainees appeared more reluctant to identify glaucoma progression from slides than from Flicker. For the 2 less experienced trainees Flicker was significantly more accurate. The MatchedFlicker software had a greater accuracy and was quicker to perform than using a handheld stereoscopic viewer.

Introducing David Epstein, the 2013 Recipient of the Weisenfeld Award

Introducing David Epstein, the 2013 Recipient of the Weisenfeld Award

Three-dimensional Nature of Retinal Nerve Fiber Layer Defects

To demonstrate the 3-dimensional nature of wedge-shaped retinal nerve fiber layer (RNFL) losses. To study the impact of training glaucoma fellows in this identification technique by evaluating their rate of RNFL loss detection on stereo disc photographs. Wedge-shaped RNFL defects were identified in stereo disc photographs of early-glaucoma patients. The 3-dimensional nature of RNFL defects and overlying retinal blood vessel contour changes were identified. Four glaucoma fellows were given stereophotographs to identify the presence or absence of RNFL defects. After training in identification of RNFL defects by noting their 3-dimensional nature and overlying blood vessel contour changes, the same set of photographs were reevaluated by the fellows. We determined whether there was a difference between the before and after assessments in the probability of a correct response for each of the fellows. Loss of retinal thickness in the form of grooves and the resulting 3-dimensional nature of the RNFL defects were visualized on stereoscopic examination. Overlying retinal blood vessels coursed into and out of the RNFL defects. Three case examples of such wedge-shaped RNFL defects are presented. For each fellow, the proportion of RNFL defects correctly identified after training was greater than the proportion before training (P<0.001). Identifying the 3-dimensional nature of wedge-shaped RNFL defects may improve rate of detection of RNFL defects during clinical examination.

Automated Segmentation of the Optic Disc from Stereo Color Photographs Using Physiologically Plausible Features

To evaluate a novel automated segmentation algorithm for cup-to-disc segmentation from stereo color photographs of patients with glaucoma for the measurement of glaucoma progression. Stereo color photographs of the optic disc were obtained by using a fixed stereo-base fundus camera in 58 eyes of 58 patients with suspected or open-angle glaucoma. Manual planimetry was performed by three glaucoma faculty members to delineate a reference standard rim and cup segmentation of all stereo pairs and by three glaucoma fellows as well. Pixel feature classification was evaluated on the stereo pairs and corresponding reference standard, by using feature computation based on simulation of photoreceptor color opponency and visual cortex simple and complex cells. An optimal subset of 12 features was used to segment all pixels in all stereo pairs, and the percentage of pixels assigned the correct class and linear cup-to-disc ratio (LCDR) estimates of the glaucoma fellows and the algorithm were compared to the reference standard. The algorithm was able to assign cup, rim, and background correctly to 88% of all pixels. Correlations of the LCDR estimates of glaucoma fellows with those of the reference standard were 0.73 (95% CI, 0.58-0.83), 0.81 (95% CI, 0.70-0.89), and 0.86 (95% CI, 0.78-0.91), respectively, whereas the correlation of the algorithm with the reference standard was 0.93 (95% CI, 0.89-0.96; n = 58). The pixel feature classification algorithm allows objective segmentation of the optic disc from conventional color stereo photographs automatically without human input. The performance of the disc segmentation and LCDR calculation of the algorithm was comparable to that of glaucoma fellows in training and is promising for objective evaluation of optic disc cupping.

Knowledge of chronology of optic disc stereophotographs influences the determination of glaucomatous change

To determine the influence of knowledge of the chronology with which stereoscopic photographs of the optic disc were obtained on how the photographs are interpreted. Case series. Disc photographs of patients enrolled in the Advanced Glaucoma Intervention Study and Collaborative Initial Glaucoma Treatment Study studies from Wills Eye Hospital. Two sets of interpretable optic disc stereo slides were included for evaluation. All patients had a follow-up of at least 5 years with baseline and follow-up photographs. Three observers (1 glaucoma attending physician and 2 glaucoma fellows) evaluated the stereophotographs in chronological order to identify the clinical impression of glaucomatous change (glaucomatous deterioration or improvement) based on information such as neuroretinal rim width, nerve fiber layer defects, rim color, peripapillary atrophy, disc hemorrhages, and blood vessel alterations. Three months later, the slide pairs were shuffled so the observers were masked to the chronological sequence. The photographs were reevaluated to determine if there was glaucomatous change. Agreement between the 2 readings was measured for each observer individually and for the 3 observers collectively. The difference between the number of photographic pairs identified as worse when chronological order was known versus unknown was evaluated using a chi-square test. Intraobserver and interobserver agreement were assessed in terms of kappa values. Intraobserver agreement between chronologically masked and unmasked readings was 61%, 64%, and 71% for the 3 observers, respectively. Average intraobserver agreement between masked and unmasked readings was 65% (kappa, 0.35). Agreement in identifying change (either deterioration or improvement) was 59%, 70%, and 68%, respectively, for each observer. Agreement in identifying stability was 63%, 62%, and 73%, respectively. The number of cases identified as having deteriorated was significantly higher (101 vs. 54) when the observer knew the chronological order with which the photographs were taken (P=0.007). Interobserver agreement was higher when the chronology of photographs was known (kappa, 0.68+/-0.05 for unmasked evaluation; kappa, 0.30+/-0.05 for masked evaluation). When disc photographs are read with knowledge of the chronology with which they were obtained, the observations differ considerably from when the readings are made without this knowledge.

Relationship between structural abnormalities and short-wavelength perimetric defects in eyes at risk of glaucoma

PURPOSE: To determine the relative prevalence of blue-yellow perimetric defects and structural abnormalities of the optic nerve and nerve fiber layer in eyes at risk of glaucoma. METHODS: Seventy-two eyes (of 72 patients) at risk of glaucoma, with normal white-on-white full threshold perimetry, were examined prospectively with blue-yellow full-threshold perimetry (Humphrey). Structural evaluations were conducted with qualitative assessment of stereoscopic color optic disk photographs and monochromatic nerve fiber layer photographs performed independently by three masked examiners (a glaucoma specialist and two glaucoma fellows), and statistical analysis of summary parameters was obtained with scanning confocal laser tomography (abnormal defined as values outside 95% confidence limits established in normal control subjects). RESULTS: Kappa values for interobserver agreement were 0.64, 0.88, and 0.79 for optic disk evaluation and 0.59, 0.60, and 0.61 for nerve fiber layer evaluation. Thirteen (18%) of 72 eyes had blue-yellow abnormalities. A total of 30 eyes (42%) were identified as having a structural abnormality; 29 (40%) had qualitatively determined optic disk abnormalities, 21 (29%) had qualitatively determined nerve fiber layer defects, and 26 (36%) had statistically significant structural abnormalities. Twelve of 13 eyes with blue-yellow defects had a detectable structural abnormality; all 12 had abnormalities identified with disk photography, nine with nerve fiber layer photography, and 12 by scanning laser tomography. CONCLUSIONS: Clinically detectable structural abnormalities frequently coexist with blue-yellow perimetric defects in patients with ocular hypertension. A substantial proportion of patients with ocular hypertension with normal blue-yellow perimetry has early detectable glaucomatous structural abnormalities.