Progression of patterns (POP): a machine classifier algorithm to identify glaucoma progression in visual fields.

PURPOSE: We evaluated Progression of Patterns (POP) for its ability to identify progression of glaucomatous visual field (VF) defects. METHODS: POP uses variational Bayesian independent component mixture model (VIM), a machine learning classifier (MLC) developed previously. VIM separated Swedish Interactive Thresholding Algorithm (SITA) VFs from a set of 2,085 normal and glaucomatous eyes into nine axes (VF patterns): seven glaucomatous. Stable glaucoma was simulated in a second set of 55 patient eyes with five VFs each, collected within four weeks. A third set of 628 eyes with 4,186 VFs (mean ± SD of 6.7 ± 1.7 VFs over 4.0 ± 1.4 years) was tested for progression. Tested eyes were placed into suspect and glaucoma categories at baseline, based on VFs and disk stereoscopic photographs; a subset of eyes had stereophotographic evidence of progressive glaucomatous optic neuropathy (PGON). Each sequence of fields was projected along seven VIM glaucoma axes. Linear regression (LR) slopes generated from projections onto each axis yielded a degree of confidence (DOC) that there was progression. At 95% specificity, progression cutoffs were established for POP, visual field index (VFI), and mean deviation (MD). Guided progression analysis (GPA) was also compared. RESULTS: POP identified a statistically similar number of eyes (P > 0.05) as progressing compared with VFI, MD, and GPA in suspects (3.8%, 2.7%, 5.6%, and 2.9%, respectively), and more eyes than GPA (P = 0.01) in glaucoma (16.0%, 15.3%, 12.0%, and 7.3%, respectively), and more eyes than GPA (P = 0.05) in PGON eyes (26.3%, 23.7%, 27.6%, and 14.5%, respectively). CONCLUSIONS: POP, with its display of DOC of progression and its identification of progressing VF defect pattern, adds to the information available to the clinician for detecting VF progression.

Comparación del rango de medida de defectos entre la perimetría estándar blanco/blanco y la perimetría Pulsar / Comparing the ranges of defect measured with standard white on white and Pulsar perimetries

Objetivo: Los umbrales normales de la perimetría Pulsar caen más rápidamente en el campoperiférico que los estándar. Se han realizado dos estudios relacionados, en primer lugar seha investigado la distribución de frecuencias de los defectos glaucomatosos en perimetríaautomática estándar (SAP), y la relación de los periféricos con los centrales (estudio A). Acontinuación se han tratado de definir los límites de examen Pulsar (estudio B).Material y métodos: Estudio A: las frecuencias se calcularon en 78.663 perimetrías SAP (G1-TOP, Octopus 1-2-3, Haag-Streit). Estudio B: 204 ojos con defecto medio (MD-SAP) inferiora 9 dB se examinaron 8,92±4,19 veces con SAP (TOP-32, Octopus 311) y con perimetría demodulación temporal (T30W, Perímetro Pulsar, Haag-Streit).Resultados: Estudio A: el 50,7% de los estudios SAP presentaron valores de MD inferiores a9 dB y el 32,7% inferiores a 6 dB. La correlación delMDde los 20◦ centrales con respecto alMDde los más periféricos fue de r = 0,933. Estudio B: en los casos con valores de MD-TOP-32 inferioresa 6 dB, SAP alcanzó sus posibilidades máximas de detección de defecto en el 0,02% delos puntos y Pulsar en el 0,29%. En los sujetos con MD-TOP-32 situado entre 6 y 9dB las frecuenciasfueron 0,38% en SAP y 3,5% en Pulsar (5,1% para excentricidades superiores a 20◦).Conclusiones: Pulsar permite detectar defectos, sin limitación de rango, en la mitad inicialde las frecuencias de defecto SAP esperables en el paciente glaucomatoso. Para estudiar laprogresión de defectos más profundos el análisis deberá centrarse en los puntos centrales,donde el rango dinámico de ambos sistemas es más equivalente(AU)

Objectives: Normal thresholds on Pulsar perimetry fall faster than those of standard perimetryin the peripheral visual field. Two related studies were performed. Firstly, the frequencydistributions of glaucoma defects on standard automated perimetry (SAP) and the relationshipof the centre and periphery (Study A) were studied first, followed by an attempt toestablish the limits of pulsar perimetry (Study B).Material and method: A: frequency of defects was calculated in 78.663 SAP perimetries (G1-TOP, Octopus 1-2-3, Haag-Streit). Study B: 204 eyes with mean defect (MD-SAP) lower than9 dB were examined 8.92±4.19 times with SAP (TOP-32, Octopus 311) and temporal modulationperimetry (T30W, Pulsar Perimeter, Haag-Streit).Results: StudyA: 50.7% of the SAP examinations showedMDvalues lower than 9 dB and 32.7%bellow 6 dB. The MD correlation of the central 20◦ with the MD of the most peripheral pointswas r = 0.933. Study B: in cases with MD-TOP-32 lower than 6 dB, SAP had the maximumpossibility of detecting defect in 0.02% of points and Pulsar in 0.29%. In subjects with MDTOP-32 between 6 and 9 dB frequencies were 0.38% in SAP and 3.5% in Pulsar (5.1% foreccentricities higher than 20◦).Conclusions: Pulsar allows detecting defects, without range limitations, in the initial half ofSAP frequencies expected on glaucoma patients. In order to study the progression of deeperdefects the examination should focus on the central points, where the dynamic range ofboth systems is more equivalent(AU)

Pattern recognition can detect subtle field defects in eyes of HIV individuals without retinitis under HAART.

OBJECTIVES: To use machine learning classifiers (MLCs) to seek differences in visual fields (VFs) between normal eyes and eyes of HIV+ patients; to find the effect of immunodeficiency on VFs and to compare the effectiveness of MLCs to commonly-used Statpac global indices in analyzing standard automated perimetry (SAP). METHODS: The high CD4 group consisted of 70 eyes of 39 HIV-positive patients with good immune status (CD4 counts were never <100/ml). The low CD4 group had 59 eyes of 38 HIV-positive patients with CD4 cell counts <100/ml at some period of time lasting for at least 6 months. The normal group consisted of 61 eyes of 52 HIV-negative individuals. We used a Humphrey Visual Field Analyzer, SAP full threshold program 24-2, and routine settings for evaluating VFs. We trained and tested support vector machine (SVM) machine learning classifiers to distinguish fields from normal subjects and high and CD4 groups separately. Receiver operating characteristic (ROC) curves measured the discrimination of each classifier, and areas under ROC were statistically compared. RESULTS: Low CD4 HIV patients: with SVM, the AUROC was 0.790 ± 0.042. SVM and MD each significantly differed from chance decision, with p < .00005. High CD4 HIV patients: the SVM AUROC of 0.664 ± 0.047 and MD were each significantly better than chance (p = .041, p = .05 respectively). CONCLUSIONS: Eyes from both low and high CD4 HIV+ patients have VFs defects indicating retinal damage. Generalized learning classifier, SVM, and a Statpac classifier, MD, are effective at detecting HIV eyes that have field defects, even when these defects are subtle.

Staging functional damage in glaucoma: review of different classification methods.

Classification of glaucomatous visual field defects for different severity levels is important. The reasons for this are numerous, and include: to distinguish between healthy and diseased individuals, to have homogeneous grouping criteria when perimetry is used to define the severity of glaucoma, to adjust therapy on the basis of disease severity, to describe visual field conditions in a short and simple format, to monitor the progression of the disease, and to provide a common language for both clinical and research purposes. Many severity classification methods have been proposed, although none have had widespread use in clinical practice. Other methods, like the cumulative defect curve (Bebie curve), can be used to distinguish the type of visual field loss as diffuse, localized, or mixed. This article provides a review of the main classification methods that have been proposed in the past 40 years.

Classifying patterns of localized glaucomatous visual field defects on automated perimetry.

AIM: To classify the classic patterns of glaucomatous visual field defects on automated perimetry and to study their proximity to fixation. STUDY DESIGN: Cross-sectional observational study. MATERIALS AND METHODS: About 1120 full threshold 30-2 reliable visual fields of glaucoma patients were analyzed by 2 glaucomatologists. Classically described patterns of visual field defects were identified on the pattern deviation plot and definitions proposed. Interreader agreement between 3 independent (not involved in the classification) readers was determined. Proximity to fixation of the different patterns was assessed. RESULTS: Interreader agreement with 3 readers was found to be 93% or more between any 2 readers using the present system of classification. Central fixation was seen to be involved in 45% of the glaucomatous visual field defects studied overall. CONCLUSIONS: The proposed definitions of topographical glaucomatous field defects based on the pattern deviation probability plot are simple to use in clinical practice with good interreader agreement.

Comparing machine learning classifiers for diagnosing glaucoma from standard automated perimetry.

PURPOSE: To determine which machine learning classifier learns best to interpret standard automated perimetry (SAP) and to compare the best of the machine classifiers with the global indices of STATPAC 2 and with experts in glaucoma. METHODS: Multilayer perceptrons (MLP), support vector machines (SVM), mixture of Gaussian (MoG), and mixture of generalized Gaussian (MGG) classifiers were trained and tested by cross validation on the numerical plot of absolute sensitivity plus age of 189 normal eyes and 156 glaucomatous eyes, designated as such by the appearance of the optic nerve. The authors compared performance of these classifiers with the global indices of STATPAC, using the area under the ROC curve. Two human experts were judged against the machine classifiers and the global indices by plotting their sensitivity-specificity pairs. RESULTS: MoG had the greatest area under the ROC curve of the machine classifiers. Pattern SD (PSD) and corrected PSD (CPSD) had the largest areas under the curve of the global indices. MoG had significantly greater ROC area than PSD and CPSD. Human experts were not better at classifying visual fields than the machine classifiers or the global indices. CONCLUSIONS: MoG, using the entire visual field and age for input, interpreted SAP better than the global indices of STATPAC. Machine classifiers may augment the global indices of STATPAC.

[Visual field defects due to antiepileptic drugs]. / Gesichtsfeldstörungen bei Antiepileptikabehandlung.

Within the last years several reports concerning visual field defects, associated with antiepileptic drugs, have been published. In addition to antiepileptic drugs several other causes (e.g. retinopathy or chloroquine, phenothiazine etc.) may induce visual field disturbances. Visual field defects have been observed during vigabatrine, tiagabine, gabapentine, diazepam, phenytoine, and carbamazepine treatment. In 13 to 46% visual field defects are reported to be linked with epilepsies. In addition to general population based studies concerning visual field defects and prospective etiological studies in epilepsies, preclinical studies for the examination of the pathomechanism of visual field defects are necessary.

[Modification of color perception during the course if glaucoma]. / Modification du sens coloré chez le glaucomateux suivant l'évolutivité de sa maladie.

Direct heterochromic comparisons were made in 334 examinations of 148 patients with glaucoma. This test evidenced the classical breaches of blue and also alterations in the red spectrum which were observed mainly in cases with an unfavorable course or during the most severe stages of the disease. This examination technique is easy to repeat and gives an accurate assessment of disease course. It is useful for guiding treatment.

Automated perimetry. American Academy of Ophthalmology.

Automated perimetry is one product of the computer revolution that has had a dramatic impact on the practice of ophthalmology, affecting the quality of both the perimetric test (perimetric technique) and the interpretation of the test results. Before automation, the quality of visual field technique in many practices was poor, partly because perimetrists lacked the volume of patients requiring the test to provide training sufficient to maintain a high level of experience and expertise. At first, automated perimetry was developed simply to make available to all practitioners visual field testing that was nearly as good as the most skilled manual perimetry. The data provided by first-generation automated perimeters lacked the accuracy, sensitivity, and reproducibility obtainable from a highly skilled manual perimetrist. As automated perimeters evolved, perimetric techniques and the software that analyzes data for clinical significance were improved. The use of automated perimetry to evaluate routinely various visual disorders that require testing of the visual field has increased significantly. Statistical analysis of the test results aids interpretation, both to determine if the field is abnormal and if there is change in the field from one occasion to another, especially in the context of glaucoma. The test is excellent; it is now standardized, with better test algorithms, and in many cases it achieves better results than the best manual perimetrist. However, attention to detail and skill in administering the automated test are still essential. The future will hold further improvements. Nevertheless, as each manufacturer enhances computerized perimetry, it may become more difficult to compare the test results among instruments or even between two test algorithms on the same instrument. In addition, clinical knowledge and experience will still be necessary to reach accurate diagnostic conclusions.