Studying the role of 10-2 visual field test in different stages of glaucoma.

OBJECTIVE: To assess the role of 10-2 visual field (VF) test in different stages of glaucoma. METHODS: In our prospective comparative study, 24-2 and 10-2 VF tests were done for 115 eyes with different stages of glaucomatous damage or glaucoma suspects. Optical coherence tomography (OCT) was performed in 79 eyes. We compared field changes of the central 10° on 10-2 and 24-2 tests and studied the correlation between the mean deviation (MD) measured by the two tests. RESULTS: In seven glaucoma suspects, glaucoma diagnosis was missed by 24-2 test but was detected by 10-2 test and confirmed by OCT. In the eyes with early damage, there was no correlation between 10-2 and 24-2 tests regarding the MD of the central 10º. In moderate and severe stages, there was a significant correlation between the results of 24-2 and 10-2 tests. CONCLUSION: We concluded that 10-2 test could help confirm glaucoma diagnosis in glaucoma suspects missed by 24-2 test before resorting to the more expensive OCT. In early glaucoma, we noted that 10-2, as confirmed by OCT, was a beneficial addition to 24-2 test for precise measurement of the MD and detection of defects of the central 10º missed by 24-2 test, where more intense treatment should be considered to preserve the threatened central visual function. In moderate and severe cases, the role of 10-2 test was not as pivotal as in early cases, but still it was useful for assessment of residual central visual function in severe cases with absolute central 10º defects on 24-2 test for proper management.

Steroid-induced ocular hypertension in the pediatric age group.

PURPOSE: Comparing the effects of topical Rimexolone versus Dexamethasone and Rimexolone versus Fluorometholone on the intraocular pressure in children <13 years. METHODS: A total of 40 patients (80 eyes) undergoing bilateral recession strabismus surgery were divided into two groups. Group A included 20 children (40 eyes); for each, one eye was randomized to receive 1% Rimexolone and the fellow eye received 0.1% Dexamethasone. Group B included 20 children (40 eyes); for each, one eye was randomized to receive 1% Rimexolone and the fellow eye received 0.1% Fluorometholone. Patients received eye drops for two consecutive weeks. Preoperative and postoperative intraocular pressure values for weeks 1, 2, 3, 4, and 6 were measured. The ocular-hypertensive response of all patients was categorized as either high, intermediate or low (Armaly-Becker Classification). RESULTS: After a 2-week treatment for both groups, peak and maximal intraocular pressure changes were reached. Changes were significantly higher in the Dexamethasone-treated eyes than in the Rimexolone- and Fluorometholone-treated eyes, which had a comparable change. (Week 2 intraocular pressure Group A: 14.15 ± 3.23 mmHg vs 17.95 ± 4.27 mmHg; Group B: 15.1 ± 2.27 mmHg vs 15.2 ± 2.73 mmHg). In both groups, the increase was statistically significant compared to the baseline intraocular pressure (preoperative intraocular pressure Group A: 13.2 ± 3.53 mmHg vs 13.1 ± 3.43 mmHg; Group B: 12.55 ± 2.98 mmHg vs 12.15 ± 3.31 mmHg). Intraocular pressure returned to near preoperative values over the following four consecutive weeks (Week 6 intraocular pressure Group A: 12.25 ± 2.67 mmHg vs 12.55 ± 2.95 mmHg; Group B: 12.15 ± 2.8 mmHg vs 12.00 ± 2.75 mmHg). None of the patients were high responders. CONCLUSION: Dexamethasone caused a higher elevation in intraocular pressure than Rimexolone and Fluorometholone in children. The ocular-hypertensive response was transient after the 2-week course.

A retrospective survey of childhood glaucoma prevalence according to Childhood Glaucoma Research Network classification.

PURPOSE: To evaluate the Childhood Glaucoma Research Network (CGRN) classification system and describe the prevalence of each subtype according to this classification. MATERIALS AND METHODS: Retrospectively, the medical records of 205 consecutive childhood glaucoma and glaucoma suspect patients at an urban tertiary care center were reviewed. The initial diagnosis and new diagnosis according to CGRN classification were recorded. RESULTS: All patients fit one of the seven categories of the new classification. Seventy-one percent of diagnoses were changed upon reclassification. Twenty-three percent of patients had primary glaucoma (juvenile open-angle glaucoma and primary congenital glaucoma [PCG]); 36% had secondary glaucoma (glaucoma associated with nonacquired ocular anomalies; glaucoma associated with nonacquired systemic disease or syndrome; glaucoma associated with acquired condition; and glaucoma following cataract surgery); and 39% were glaucoma suspect. Of the patients diagnosed with glaucoma, PCG was the most common diagnosis, seen in 32% of patients. CONCLUSION: The CGRN classification provides a useful method of classifying childhood glaucoma.

Twenty-four-hour pattern of intra-ocular pressure in untreated patients with primary open-angle glaucoma.

PURPOSE: To investigate and compare the 24-hr intra-ocular pressure (IOP) pattern among different severities of glaucomatous optic neuropathy (GON) together with different subtypes of primary open-angle glaucoma (POAG). METHODS: The IOPs of untreated patients with POAG were measured every 2 hrs over a 24-hr period. Subjects were divided into mild, moderate and severe groups according to the glaucoma staging system 2. The two POAG subtypes, normal tension glaucoma (NTG) and hypertension glaucoma (HTG) were classified according to peak IOP values throughout the 24-hr period. The 24-hr IOP curves were obtained based on the average IOP value at each time-point. The mean, peak, trough and fluctuation of IOP together with the diurnal-to-nocturnal IOP change were calculated and compared among different groups. RESULTS: A total of 160 POAG patients (84 NTG and 76 HTG) were enrolled in the study. Generally, IOP decreased in the diurnal period and increased progressively during the nocturnal period, with peak IOP occurring from 2:00 AM to 10:00 AM. The average IOP values at all time-points were significantly higher in the mild group compared to the moderate and severe groups (p < 0.05). Compared with the NTG subjects, the IOP parameters including mean, peak, trough and fluctuation were significantly higher in the HTG subjects (p < 0.001), and IOP increased more predominantly in the HTG subtype during the nocturnal period. CONCLUSIONS: There was some regularity in the 24-hr IOP pattern in POAG, but different severities of GON and different subtypes might present different characteristics. Other non-IOP factors may lead to pathological IOP fluctuation and could be correlated with GON.

Automated Detection of Glaucoma From Topographic Features of the Optic Nerve Head in Color Fundus Photographs.

OBJECTIVE: To describe and evaluate the performance of an automated CAD system for detection of glaucoma from color fundus photographs. DESIGN AND SETTING: Color fundus photographs of 2252 eyes from 1126 subjects were collected from 2 centers: Aravind Eye Hospital, Madurai and Coimbatore, India. The images of 1926 eyes (963 subjects) were used to train an automated image analysis-based system, which was developed to provide a decision on a given fundus image. A total of 163 subjects were clinically examined by 2 ophthalmologists independently and their diagnostic decisions were recorded. The consensus decision was defined to be the clinical reference (gold standard). Fundus images of eyes with disagreement in diagnosis were excluded from the study. The fundus images of the remaining 314 eyes (157 subjects) were presented to 4 graders and their diagnostic decisions on the same were collected. The performance of the system was evaluated on the 314 images, using the reference standard. The sensitivity and specificity of the system and 4 independent graders were determined against the clinical reference standard. RESULTS: The system achieved an area under receiver operating characteristic curve of 0.792 with a sensitivity of 0.716 and specificity of 0.717 at a selected threshold for the detection of glaucoma. The agreement with the clinical reference standard as determined by Cohen κ is 0.45 for the proposed system. This is comparable to that of the image-based decisions of 4 ophthalmologists. CONCLUSIONS AND RELEVANCE: An automated system was presented for glaucoma detection from color fundus photographs. The overall evaluation results indicated that the presented system was comparable in performance to glaucoma classification by a manual grader solely based on fundus image examination.

Agreement among graders on Heidelberg retina tomograph (HRT) topographic change analysis (TCA) glaucoma progression interpretation.

PURPOSE: To evaluate agreement among experts of Heidelberg retina tomography's (HRT) topographic change analysis (TCA) printout interpretations of glaucoma progression and explore methods for improving agreement. METHODS: 109 eyes of glaucoma, glaucoma suspect and healthy subjects with ≥5 visits and 2 good quality HRT scans acquired at each visit were enrolled. TCA printouts were graded as progression or non-progression. Each grader was presented with 2 sets of tests: a randomly selected single test from each visit and both tests from each visit. Furthermore, the TCA printouts were classified with grader's individual criteria and with predefined criteria (reproducible changes within the optic nerve head, disregarding changes along blood vessels or at steep rim locations and signs of image distortion). Agreement among graders was modelled using common latent factor measurement error structural equation models for ordinal data. RESULTS: Assessment of two scans per visit without using the predefined criteria reduced overall agreement, as indicated by a reduction in the slope, reflecting the correlation with the common factor, for all graders with no effect on reducing the range of the intercepts between the graders. Using the predefined criteria improved grader agreement, as indicated by the narrower range of intercepts among the graders compared with assessment using individual grader's criteria. CONCLUSIONS: A simple set of predefined common criteria improves agreement between graders in assessing TCA progression. The inclusion of additional scans from each visit does not improve the agreement. We, therefore, recommend setting standardised criteria for TCA progression evaluation.

The relationship between cup-to-disc ratio and estimated number of retinal ganglion cells.

PURPOSE: To investigate the relationship between cup-to-disc ratio (CDR) and estimates of retinal ganglion cell (RGC) number. METHODS: This cross-sectional study included 156 healthy eyes, 53 glaucoma suspects, and 127 eyes with glaucoma. All eyes had standard automated perimetry (SAP), Cirrus SD-OCT, and stereoscopic optic disc photography within 6 months. CDR was determined from stereoscopic photographs by two or more masked graders. The number of RGCs in each eye was estimated using a published model that combines estimates of RGC number from SAP sensitivity thresholds and SD-OCT retinal nerve fiber layer measurements. RESULTS: The mean estimated RGC count was 1,063,809 in healthy eyes; 828,522 in eyes with suspected glaucoma; and 774,200 in early, 468,568 in moderate, and 218,471 in advanced glaucoma. Healthy eyes had a mean vertical CDR of 0.45 ± 0.15 vs. 0.80 ± 0.16 in glaucomatous eyes. There was good correlation between stereophotographic vertical CDR and SD-OCT vertical CDR (R(2) = 0.825; P < 0.001). The relationship between estimated RGCs and vertical CDR was best represented using a third degree polynomial regression model, including age and optic disc area, which accounted for 83.3% of the variation in estimated RGC counts. The nonlinear relationship between RGC estimates and CDRs indicated that eyes with a large CDR would require loss of large RGC numbers for a small increase in CDR. CONCLUSIONS: The relationship between estimated RGC counts and CDR suggests that assessment of change in CDR is an insensitive method for evaluation of progressive neural losses in glaucoma. Even relatively small changes in CDR may be associated with large losses of RGCs, especially in eyes with large CDRs. (ClinicalTrials.gov numbers, NCT00221923, NCT00221897.).

The applicability of ganglion cell complex parameters determined from SD-OCT images to detect glaucomatous eyes.

PURPOSE: To determine whether the thicknesses of the different parameters of the ganglion cell complex (GCC) and peripapillary retinal nerve fiber layer can be used to differentiate eyes with glaucoma from normal eyes. METHODS: Two hundred sixty-one eyes, including 68 normal eyes and 32 preperimetric glaucoma, 81 early glaucoma, and 80 advanced glaucoma were analyzed in the present study. The thicknesses of the GCC and retinal nerve fiber layer were measured using RTVue spectral-domain optical coherence tomographic (SD-OCT) images. The area under the receiver operating characteristic (AUROC) curve and sensitivities at fixed specificities were calculated for each parameter. A logistic regression analysis was used to determine the risk factors for glaucoma. RESULTS: The 2 largest AUROC curves for all glaucoma stages were those for the GCC parameters. The global loss volume (GLV) was always one of the 2 highest values of the AUROC curve. The GLV also had the highest sensitivity at a fixed specificity to identify glaucoma at early and advanced stage. The focal loss volume (FLV) had the largest AUROC curve value and the highest sensitivity at a fixed specificity for advanced glaucoma. The logistic regression analysis showed that the GLV was one of the factors that predicted preperimetric glaucoma [odds ratio (OR)=1.74] and early glaucoma (OR=1.22), whereas the FLV was useful for detecting advanced glaucoma (OR=2.32). CONCLUSIONS: The SD-OCT-derived macular GCC parameters can be used to detect preperimetric and perimetric glaucoma. The new GCC parameters, GLV and FLV, performed well in discriminating glaucomatous eyes from normal eyes.

The Portsmouth-based glaucoma refinement scheme: a role for virtual clinics in the future?

BACKGROUND: Glaucoma referrals continue to impart a significant burden on Hospital Eye Services (HES), with a large proportion of these false positives. AIMS: To evaluate the Portsmouth glaucoma scheme, utilising virtual clinics, digital technology, and community optometrists to streamline glaucoma referrals. METHOD: The stages of the patient trail were mapped and, at each step of the process, 100 consecutive patient decisions were identified. The diagnostic outcomes of 50 consecutive patients referred from the refinement scheme to the HES were identified. RESULTS: A total of 76% of 'glaucoma' referrals were suitable for the refinement scheme. Overall, 94% of disc images were gradeable in the virtual clinic. In all, 11% of patients 'attending' the virtual clinic were accepted into HES, with 89% being discharged for community follow-up. Of referrals accepted into HES, the positive predictive value (glaucoma/ocular hypertension/suspect) was 0.78 vs 0.37 in the predating 'unrefined' scheme (95% CI 0.65-0.87). The scheme has released 1400 clinic slots/year for HES, and has produced a £244 200/year cost saving for Portsmouth Hospitals' Trust. CONCLUSION: The refinement scheme is streamlining referrals and increasing the positive predictive rate in the diagnosis of glaucoma, glaucoma suspect or ocular hypertension. This consultant-led practice-based commissioning scheme, if adopted widely, is likely to incur a significant cost saving while maintaining high quality of care within the NHS.

Referral source, symptoms, and severity at diagnosis of ocular hypertension or open-angle glaucoma in various practices.

OBJECTIVE: To evaluate the severity of glaucoma at presentation and the relationship to referral source and presenting symptoms. DESIGN: Prospective, multicentre, cross-sectional, study. PARTICIPANTS: Patients newly diagnosed with ocular hypertension (OH) or open-angle glaucoma (OAG). METHODS: Eligible subjects underwent comprehensive ocular examinations, including medical histories and presenting symptoms. The referral source was also documented. Patients were classified as OH or mild, moderate, or advanced OAG. The proportion of subjects in each disease category was determined and compared for differences in demographics, ocular findings, referral sources, and presenting symptoms. RESULTS: We enrolled 404 subjects; 113 (28%) were diagnosed as having OH and 291 (72%) as having OAG. In the OAG group 151 (52%) had mild, 79 (27%) moderate, and 61 (21%) advanced disease at initial diagnosis. The majority of the referrals were from optometrists (82%), followed by family physicians (7%) and ophthalmologists (7%). Referral patterns and presenting symptoms were significantly different for advanced disease as compared to the other severity levels (p = 0.03 and p 0 .003, respectively). In more than half of the patients, the referral was made because of an abnormality noted by the referring source in an asymptomatic subject. Visual symptoms noted by the subject were more common in advanced OAG (23%) as compared to OH (9%), mild (4%), and moderate (9%) OAG. CONCLUSIONS: Nearly half of the patients with OAG had moderate or advanced disease at the time of initial diagnosis. Although optometrists were responsible for the majority of the referrals, the proportion was significantly less for advanced disease. These results raise the possibility of suboptimal detection services for glaucoma.

Morphometric analysis and classification of glaucomatous optic neuropathy using radial polynomials.

PURPOSE: To quantify the morphological features of the optic nerve head using radial polynomials, to use these morphometric models as the basis for classification of glaucomatous optic neuropathy via an automated decision tree induction algorithm, and to compare these classification results with established procedures. METHODS: A cohort of patients with high-risk ocular hypertension or early glaucoma (n=179) and a second cohort of normal subjects (n=96) were evaluated for glaucomatous optic neuropathy using stereographic disc photography and confocal scanning laser tomography. Morphological features of the optic nerve head region were modeled from the tomography data using pseudo-Zernike radial polynomials and features derived from these models were used as the basis for classification by a decision tree induction algorithm. Decision tree classification performance was compared with expert classification of stereographic disc photographs and analysis of neural retinal rim thickness by Moorfields Regression Analysis (MRA). RESULTS: Root mean squared error of the morphometric models decreased asymptotically with additional polynomial coefficients, from 62±0.5 (32 coefficients) to 32±5.7 µm (256 coefficients). Optimal morphometric classification was derived from a subset of 64 total features and had low sensitivity (69%), high specificity (88%), very good accuracy (80%), and area under the receiver operating characteristic curve (AUROC) was 88% (95% confidence interval, 78%-98%). In comparison, MRA classification of the same records had a comparatively poorer sensitivity (55%), but had higher specificity (95%), with similar overall accuracy (78%) and AUROC curve, 83% (95% CI, 70%-96%). CONCLUSIONS: Pseudo-Zernike radial polynomials provide a mathematically compact and faithful morphological representation of the structural features of the optic nerve head. This morphometric method of glaucomatous optic neuropathy classification has greater sensitivity, and similar overall classification performance (AUROC) when compared with classification by neural retinal rim thickness by MRA in patients with high-risk ocular hypertension and early glaucoma.

Effect of signal strength on reproducibility of circumpapillary retinal nerve fiber layer thickness measurement and its classification by spectral-domain optical coherence tomography.

PURPOSE: To assess the effect of signal strength (SS) on the reproducibility of circumpapillary retinal nerve fiber layer (cpRNFL) thickness measurement (measurement agreement) and its color-coded classification (classification agreement) by Cirrus spectral-domain optical coherence tomography (OCT). METHODS: Seven hundred and eighty-nine participants, consecutively enrolled from December 2009 to August 2010, underwent two repeated Cirrus OCT scans to measure cpRNFL thickness (optic disc cube 200 × 200). Intraclass correlation (ICC) and linear-weighted kappa coefficient (κ) were calculated as indicators for cpRNFL measurement and classification agreement. The difference in cpRNFL thickness measurements between repeated OCT scans (inter-scan measurement difference) was correlated with mean SS, SS change or difference between repeated OCT scans (inter-scan SS difference), and average cpRNFL thickness. RESULTS: Mean SS was negatively correlated with inter-scan SS difference and was positively correlated with the inter-scan measurement difference. Repeated scans with SS change (inter-scan SS difference = 1 or 2) showed larger variability than scans without SS change (inter-scan SS difference = 0) in total average cpRNFL and all quadrant maps except for the nasal quadrant. Multivariate analysis revealed that inter-scan measurement differences increased in subjects with thinner cpRNFL thicknesses and in scans with lower mean SS and higher inter-scan SS differences. Measurement agreement (ICC) in groups with larger inter-scan SS differences (=2) was lower than those with smaller inter-scan SS differences (=0 or 1) in their average cpRNFL and inferior quadrant maps. Classification agreement (κ) in groups with larger inter-scan SS differences (=2) was lower than those with smaller inter-scan SS differences (=0 or 1) in their average cpRNFL and quadrant maps (superior and inferior quadrant) although statistically not significant. CONCLUSIONS: Special attention should be paid when comparing the cpRNFL thickness measurement and color-coded classification of 2 or more Cirrus OCT scans as the agreements may be susceptible to SS differences.

Comparison of two retinal nerve fibre layer thickness measurement patterns of RTvue optical coherence tomography.

BACKGROUND: To evaluate the agreement and repeatability between operator-dependent centring and automatic centring retinal nerve fibre layer thickness measurement patterns (RNFL3.45 and NHM4) of RTvue optical coherence tomography in normal and glaucomatous eyes. DESIGN: Cross-sectional observational study. PARTICIPANTS: A total of 153 eyes from 149 normal subjects and subjects with glaucoma were analysed. METHODS: The retinal nerve fibre layer thickness was measured using RNFL3.45 and NHM4 three times on the same day to determine the repeatability and agreement between the two scan patterns. MAIN OUTCOME MEASURES: Student's paired t-testing, intra-class correlation coefficient, coefficient of variation, test-retest viability, Pearson's correlation coefficient and Bland-Altman analysis of retinal nerve fibre layer thickness measurements. RESULTS: The difference between RNFL3.45 and NHM4 measurements was statistically significant by paired t-testing (P=0.003) only in severe glaucoma group. The Pearson's correlation test showed a high degree of correlation of the mean retinal nerve fibre layer thickness (r=0.949). Bland-Altman plots showed that the differences between RNFL3.45 and NHM4 were smaller at thicker retinal nerve fibre layer values, but larger at thinner retinal nerve fibre layer values. The intra-class correlation coefficient for RNFL3.45 (and lower 95% confidence interval) in normal and glaucomatous eyes was 0.990 (0.985 confidence interval) and 0.997 (0.995 confidence interval), respectively. The intra-class correlation coefficient for NHM4 in normal and glaucomatous eyes was 0.989 (0.983 confidence interval) and 0.995 (0.992 confidence interval), respectively. The test-retest variability for RNFL3.45 and NHM4 ranged from 3.98 to 9.75µm in normal eyes, and from 4.10 to 13.34µm in glaucomatous eyes. CONCLUSION: Measurements of retinal nerve fibre layer thickness by RNFL3.45 and NHM4 are in good agreement.

GDx staging system: a new method for retinal nerve fiber layer damage classification.

PURPOSE: To provide a reliable and easy-to-use method for classifying retinal nerve fiber layer (RNFL) damage using the parameters obtained by the scanning laser polarimeter with variable corneal compensator (GDx VCC). METHODS: GDx Staging System is a new method that uses the superior and inferior RNFL thickness values plotted on an x-y diagram to classify GDx VCC results. RNFL defects are classified into 6 stages of increasing severity and 3 classes of defect localization (superior, inferior, or diffuse defect). The diagram was created based on 320 GDx VCC tests from 84 healthy controls and from 236 patients affected by ocular hypertension or chronic open-angle glaucoma. Sensitivity and specificity of the method were assessed in a different cohort that included 161 patients with either ocular hypertension or open-angle glaucoma, and 34 normal participants. The results were correlated with both a clinical classification of the GDx VCC results and with visual field defects classified with the Glaucoma Staging System 2. RESULTS: Sensitivity and specificity of the GDx Staging System were, respectively, 88.2% and 100%. Correlations with the clinical classification and the Glaucoma Staging System 2 results were statistically significant (Spearman correlation coefficient 0.92 and 0.57, respectively; P<0.0001). CONCLUSIONS: The GDx Staging System is an easy and quick method for interpreting GDx VCC results. It can be clinically useful, especially for nonexpert ophthalmologists.

Sensitivity and specificity of the Heidelberg Retina Tomograph II Version 3.0 in a population-based study: the Tajimi Study.

PURPOSE: To evaluate the sensitivity and specificity of the 3 glaucoma classification programs, the FS Mikelberg discriminant function (FSM), Moorfields Regression Analysis (MRA), and Glaucoma Probability Score (GPS) of version 3.0 of the Heidelberg Retina Tomograph (HRT) II (Heidelberg Engineering, Dossenheim, Germany), in a population-based setting for the first time. DESIGN: Population-based cross-sectional study. PARTICIPANTS: One randomly chosen eye of each subject without glaucoma, subject with glaucoma, and subject with suspected glaucoma with reliable HRT II measurements from the Tajimi study (2297 eyes of 2297 subjects) were included for analysis. METHODS: Glaucoma was diagnosed by the optic disc and visual field findings according to the criteria of the International Society of Geographical and Epidemiological Ophthalmology. The sensitivity and specificity of FSM, MRA, and GPS were calculated. Characteristics of erroneously diagnosed glaucoma (false-negative) eyes and factors that influenced specificity with the 3 programs were investigated. MAIN OUTCOME MEASURES: Sensitivity and specificity of FSM, MRA, and GPS. RESULTS: Sensitivity and specificity varied significantly among the 3 programs: 59.1%, 39.4%, and 65.2% (P = 0.02 approximately 0.003, chi-square test), and 86.7%, 96.1%, and 83.0% (P<0.0001) with FMS, MRA, and GPS, respectively. MRA gave the lowest sensitivity but the highest specificity. Positive predictive values for these programs ranged between 0.10 and 0.23, whereas negative predictive values ranged between 0.98 and 0.99. False-negative eyes had significantly better visual field indexes (P<0.01 approximately 0.002, Mann-Whitney U test) and smaller cup and larger rim parameters compared with true-positive glaucoma eyes. Older age and hyperopia were negatively correlated with the specificity of GPS but not with that of FMS and MRA. Larger disc area was significantly associated with decreased specificity of all programs. CONCLUSIONS: In a population-based setting, the sensitivity of the HRT II was unsatisfactory with any of the classification programs, whereas specificity was satisfactory. A significant percentage of the glaucoma discs were labeled as normal, and eyes in the earlier stage of the disease appear to be more likely to be misdiagnosed as normal. Factors such as age, refraction, and disc area had an influence on specificity, but the degree of its influence was different for each classification program.

Glaucoma detection and evaluation through pattern recognition in standard automated perimetry data.

BACKGROUND: Perimetry remains one of the main diagnostic tools in glaucoma, and it is usually used in conjunction with evaluation of the optic nerve. This study assesses the capability of automatic pattern recognition methods, and in particular the support vector machines (SVM), to provide a valid clinical diagnosis classification of glaucoma based solely upon perimetry data. METHODS: Over 2,200 patient records were reviewed to produce an annotated database of 2,017 eyes. Visual field (VF) data were obtained with HFA II perimeter using the 24-2 algorithm. Ancillary information included treated and untreated intraocular pressure, cup-to-disk ratio, age, sex, central corneal thickness and family history. Ophthalmic diagnosis and classification of visual fields were provided by a consensus of at least two glaucoma experts. The database includes normal eyes, cases of suspect glaucoma, pre-perimetric glaucoma, and glaucoma with different levels of severity, as well as 189 eyes with neurologic or neuro-ophthalmologic defects. Support vector machines were trained to provide multi-level classifications into visual field and glaucoma diagnosis classes. RESULTS: Numerical validation indicates 70-90% expected agreement between multi-stage classifications provided by the automated system, using a hierarchy of SVM models, and glaucoma experts. Approximately 75% accuracy for the classification of glaucoma suspect and pre-perimetric glaucoma (which by definition do not exhibit glaucomatous defects) indicates the ability of the numerical model to discern subtle changes in the VF associated with early stages of glaucoma. The Glaucoma Likelihood Index provides a single number summary of classification results. CONCLUSIONS: Automatic classification of perimetry data may be useful for glaucoma screening, staging and follow-up.

Artificial neural network-based glaucoma diagnosis using retinal nerve fiber layer analysis.

PURPOSE: To develop, train, and test an artificial neural network (ANN) for differentiating among normal subjects, primary open angle glaucoma (POAG) suspects, and persons with POAG in Asian-Indian eyes using inputs from clinical parameters, optical coherence tomography (OCT), visual fields, and GDx nerve fiber analyzer. METHODS: One hundred eyes were classified using optic disc examination and perimetry into normal (n=35), POAG suspects (n=30), and POAG (n=35). EasyNN-plus simulator was used to develop an ANN model with inputs including age, sex, myopia, intraocular pressure (IOP), optic nerve head, and retinal nerve fiber layer (RNFL) parameters on OCT, Octopus 30-2 full threshold visual field, and GDx parameters. RESULTS: With two outputs (POAG or normal), specificity was 80% and sensitivity was 93.3%. Ninety percent of POAG suspects were labeled as abnormal in this analysis. ANN assigned the highest importance to Smax/Imax RNFL on OCT followed by cup-area (OCT) and other RNFL parameters (OCT) for two outputs. With three outputs (normal, POAG, and POAG suspect), ANN gave an overall classification rate of 65%, specificity of 60%, and sensitivity of 71.4% with a target error rate of the training set at 1%. The parameters for three outputs, in decreasing order of relative importance, were Savg, vertical cup-disc ratio, cup-volume, and cup-area on OCT. CONCLUSIONS: An ANN taking varied diagnostic imaging inputs was able to separate POAG eyes from normal subjects and POAG suspects. The network had reasonable sensitivity with three outputs; however, it had a tendency to mislabel POAG suspects as POAG.

Relationship between Humphrey 30-2 SITA Standard Test, Matrix 30-2 threshold test, and Heidelberg retina tomograph in ocular hypertensive and glaucoma patients.

PURPOSE: To evaluate the relationship between global indices of Humphrey standard automated perimetry (SAP, 30-2 SITA standard test), Humphrey Matrix frequency doubling technology (FDT, 30-2 threshold test), and Heidelberg Retina Tomograph (HRT II) parameters and measure the level of agreement among these 3 tests in classifying eyes as normal or abnormal. METHODS: The study included 1 eye of 29 ocular hypertensive and 56 glaucoma patients with a mean age of 60.9+/-10.5 years. All subjects had reliable visual fields and HRT measurements performed within a 2-week period. The eyes were classified as normal/abnormal according to visual field criteria and Moorfields regression analysis (MRA). Correlations between visual field indices (mean deviation and pattern standard deviation) and HRT parameters were analyzed using Spearman correlation coefficient (r) and the agreement between the tests in classifying eyes was defined with kappa value. RESULTS: FDT Matrix mean deviation and pattern standard deviation parameters were found to be highly correlated with those of SAP (r=0.66 and 0.69, respectively). Visual field indices showed statistically significant correlations with cup area, rim area, cup/disc (C/D) area, linear C/D, cup shape, mean retinal nerve fiber layer thickness and retinal nerve fiber layer area parameters (P<0.05). Fifty-eight patients (68.2%) had abnormal results at least with 1 of the tests and 21 subjects (24.7%) had abnormal results with all 3 tests. The kappa values were 0.6 for SAP and Matrix (P<0.001), 0.33 for SAP and MRA (P=0.002), and 0.31 for Matrix and MRA (P=0.004). CONCLUSIONS: FDT Matrix results are highly comparable with SAP in the assessment of glaucoma. Visual field global indices show statistically significant, but low-moderate correlations with most of the HRT parameters. The agreement among MRA and visual fields for abnormality is fair. Either HRT or visual fields may show the first evidence of glaucomatous damage; therefore, the combination of optic nerve head parameters and visual field results could improve glaucoma diagnosis and follow-up.