IPS 2004: 16th Perimetry & Imaging Symposium

Barcelona, Spain
June 29-July 2, 2004
|ABSTRACTS

pdf version of this document

IMPROVING GCP LIMITS USING PATIENT AND PROCEDURE KNOWLEDGE

A Turpin¹, AM McKendrick², BC Chauhan³. Curtin University of Technology, Perth, Australia¹, University of Western Australia, Perth, Australia², Dalhousie University, Halifax, Canada³.

Purpose: To show how the accuracy of the Glaucoma Change Probability (GCP) can be improved if individual patient variability and the characteristics of the visual field testing algorithm are taken into account.

Methods: Exhaustive computer-based analysis of all possible response sequences was used to derive 95% confidence intervals (CI) for end-point sensitivity measures using the Full Threshold algorithm under a variety of patient variability models. The models consisted of all combinations of frequency-of-seeing curve slopes following a Cumulative Gaussian curve with standard deviations of 1 through 5 dB and false positive and negative rates of 0%, 3%, 15% and 30%. For each model, the starting point of the Full Threshold algorithm (FT) was also varied to be ±4 dB away from the patient’s assumed “true” sensitivity.

Results:  The table shows 95% CIs for end-points of FT starting at the value in the leftmost column for a patient with a true sensitivity of 20 dB under selected slopes (σ) and false response rates.

Conclusions: Currently the limits used for determining whether a point has progressed or not using GCP is based solely on population values related to age-corrected baseline defect. This study demonstrates that individual patient variability and test procedure specific factors dramatically affect the 95% CIs of sensitivity measures, and should be considered when using GCP.

BAYES' THEOREM APPLIED TO PERIMETRIC PROGRESSION DETECTION IN GLAUCOMA

NM Jansonius. Department of Ophthalmology, University Hospital Groningen,  Groningen, The Netherlands.

Purpose: To calculate the posterior probability (positive predictive value) of progression for various established perimetric progression detection algorithms and for clinical assessment of series of visual fields.

Methods: Prior probability of progression was estimated from various published studies to be 5%, 10%, and 20% per year for well controlled, poorly controlled, and uncontrolled glaucoma respectively. Specificity of several established algorithms for progression detection (AGIS, CIGTS, GCP, PLRA) was estimated from literature to range from 0.80 to 0.99. Specificity of clinical assessment of series of visual fields was calculated to be 0.80 for 4 fields (2 baseline fields, 1 follow-up field after 1 year, and 1 confirmation of the suspected progression), 0.95 for 6 fields, and 0.99 for 10 fields. Calculations were performed for three different sensitivity values: 0.50, 0.80, and 1.00.   

Results: Positive predictive value ranged from 12% to 96%. Positive predictive value was 64% for a prior probability of 10%, a sensitivity of 0.80, and a specificity of 0.95.

Conclusions: Realistic series of visual fields that are apparently progressive do have a positive predictive value of typically 50%, i.e. half of them is actually stable. In case of a high prior probability (uncontrolled glaucoma), 4 fields may be acceptable. If the suspicion is low, on the contrary, then even the generally accepted number of 6 visual fields is hardly sufficient

A NEW APPROACH TO DETECTING CHANGE IN SERIES OF RETINAL IMAGES ACQUIRED FROM SCANNING LASER TOMOGRAPHY

DP Crabb¹, AJ Patterson¹, DF Garway-Heath².

The Nottingham Trent University, Nottingham, UK¹, Moorfields Eye Hospital, London, UK³.

Purpose: To describe and apply a collection of new statistical technique for detecting topographic changes in the optic disc and peripapillary retina measured with the Heidelberg Retinal Tomograph (HRT)

Methods: Quantitative techniques, collectively referred to as Statistic Image Mapping, are widely used to measure activity and change in fMRI (functional magnetic resonance imaging) and PET (positron emission tomography) images of the brain. These techniques are adapted and applied to HRT images. In particular, we generate a test statistic at each pixel in series of aligned, pre-processed images, formed from linear regression of the topographic height against time of follow up. The standard error for this univariate statistic is corrected by a linear combination of the error at neighbouring pixels. Further, we compare the observed test statistic at each pixel against a distribution of all possible statistics by creating a permutation distribution of the images. This non-parametric approach relies wholly on the patient’s own data and allows for meaningful results even if the data is sparse in time (short follow-up). Furthermore, use of the maximum test statistic permutation distribution solves the problem of multiple testing of pixels across the whole image, and thresholding discrete areas of changing pixels takes account of the spatial correlation that exists in the images. The techniques were developed from those used in fMRI brain activity experiments and implemented via purpose written software in C.

Results: The results from the analyses are summarised by an image superimposed on the HRT topography which, on a pixel by pixel basis, indicates areas that are changing beyond what would be expected by chance given the patient’s follow up data alone. The permutation methods are computationally intensive, but a typical analysis takes approximately 2 minutes on a Pentium IV 2GHz processor.

Conclusion: Statistic Image Mapping appears to be an appropriate analysis for detecting change in digital images acquired by scanning laser tomography.

CRITERIA FOR CHANGE WITH THE HEIDELBERG RETINA TOMOGRAPH IN HEALTHY SUBJECTS AND PATIENTS WITH GLAUCOMA.

BC Chauhan, PH Artes, Ophthalmol Vis Sci, Dalhousie University, Halifax, CANADA

Purpose: To investigate criteria for change with the Heidelberg Retina Tomograph in healthy subjects and patients with glaucoma.

Methods: Ninety-five patients with glaucoma and 60 healthy controls were followed with 6-monthly Heidelberg Retina Tomograph 1 imaging for a mean of 6.3 years. Optic disk change was evaluated using the probability maps of the Heidelberg Eye Explorer (v 3.0.4.6) which show superpixels with statistically significant surface height change from baseline (p<0.05) in 3 consecutive mean topographies. The criteria for change were based on the size of the largest cluster of red (progressing) and green (improving) superpixels within the contour line, expressed as a percentage of disk area. Kaplan-Meier survival curves were derived with 3 criteria (2%, 5%, 10%), corresponding to cluster sizes of approximately 20, 50, and 100 superpixels within a typical disk area of 2mm², respectively. 

Results: Clusters of both red and green superpixels occurred more often in glaucoma patients than in healthy controls (p<0.01 with all criteria). In the controls, rates of progression and improvement were similar with any of the three criteria (p>0.1, log-rank test). In contrast, the progression rates in the glaucoma group were, on average, twice as large as the rates of improvement. After 6 yrs of follow-up, the progression rate in glaucoma patients varied from 67% (largest cluster of red superpixels >2% of disk area) to 25% (largest cluster >10% of disk area); a criterion of 7.5% (largest cluster >75 superpixels in an average disk) gave an improvement rate just below 10% and a progression rate of 32% (95% CI; 24%, 40%). 

Conclusions: The separation between survival curves for progression and improvement observed in the glaucoma patients, but not the healthy controls, confirms the validity of the change probability approach for analysing longitudinal HRT data. While clinicians looking for change should take into account the image quality, the quality of alignment as well as the amount of absolute height change, the criterion of [“largest cluster >7.5% of disk size”] appears to offer reasonable specificity for following glaucoma patients with the present implementation of the change probability software.

A Glaucoma Change Probability (GCP) Analysis Procedure for Frequency Doubling Technology (FDT) Perimetry

Chris A. Johnson¹, Paul G.D. Spry² and Balwantray Chauhan³

1 Devers Eye Institute, Portland, Oregon, USA 2 Bristol Eye Hospital, Bristol, England

3 Dalhousie University, Halifax, Nova Scotia, Canada

Purpose:  Frequency Doubling Technology (FDT) perimetry has been reported to be sensitive for detecting glaucomatous, retinal and neuro-ophthalmologic visual field loss. However, the ability of FDT perimetry to determine progression has not been adequately determined to date.  The purpose of this investigation was to evaluate the clinical efficacy of a Glaucoma Change Probability (GCP) analysis procedure for FDT perimetry to determine glaucomatous visual field progression.  Methods:  The Glaucoma Change Probability (GCP) and Glaucoma Progression Analysis (GPA) procedures are evaluations that permit examination of glaucomatous visual field progression for standard automated perimetry using the Humphrey Field Analyzer.  To produce a similar visual field change probability procedure for FDT, we performed repeated testing in 64 patients with glaucomatous visual field loss and 47 normal control subjects.  Standard automated perimetry and FDT perimetry (C-20 threshold test) were performed five times over a 4 week period of time.  Empirical 5^th and 95^th percentiles were derived for each mean FDT sensitivity value to establish visual field locations on followup testing that were significantly better (+), worse (-) or within (o) the limits of variability in comparison to the baseline FDT visual field.  The change probability analysis procedure is highly similar to the original methods employed by the Humphrey Field Analyzer for standard automated perimetry.  Results:   Our findings indicate that it is possible to use this FDT GCP analysis procedure to monitor the visual field status of glaucoma patients and glaucoma suspects over extended time periods.  Examples of glaucomatous visual field progression and stability will be presented for the FDT GCP analysis procedure.  Conclusions:  The FDT GCP analysis procedure may be a useful clinical tool for the determination of glaucomatous visual field progression for the FDT perimeter.  Further work at multiple centers will be needed to establish the overall clinical performance of this type of FDT visual field evaluation over time.

SPATIAL SUMMATION ESTIMATION IN THE CENTRAL VISUAL FIELD

Marta González-Hernández, Manuel González de la Rosa, Alicia Pareja Ríos, Fátima Mesa Lugo F.

University of La Laguna. Canary Islands. Spain

Introduction: Classic kinetic perimetry based the equivalences between luminance (L) and stimulus area (A) in the equation LxA^k=constant. Goldmann equivalences assume a value of k=0.83. Fankhauser estimated a difference between sizes Goldmann I and III of 3-4dB in the fovea (which is equivalent to k=0.291) and 12dB at 50º of eccentricity (k=0.998). Gramer estimated a change between sizes I and III and between III and V of 6-10dB (equivalent to a value of k=0.499-0.832). We have evaluated the k value in 66 positions of the central visual field.

Material and Methods: A USB photometer was designed for automatically controlling the luminance scale of a video screen. 10 eyes of 10 healthy subjects (mean age 38.5 years, s.d.= 16.6) were examined for luminous thresholds in 66 locations of the central visual field (horizontal= -30º to +30º) x (vertical= +24º to -24º) with the TOP strategy, using stimuli of five different sizes (Goldmann 4, 3.5, 3, 2.5 and 1.9). The k value was calculated estimating the average of threshold equivalences for the five sizes.

Results: In relation to the threshold value obtained for size 4, sensitivity decreased 1.5, 2.9, 5.4 and 7.8 dB for the following sizes respectively. Therefore, a mean value of k=0.607 was calculated for the whole visual field. The value of k increased in a lineal manner from the centre towards the periphery of the visual field, with a slope of 0.01 per degree  (r=0.98, p<0.01). It was slightly higher in the inferior hemi-fields (k=0.657) than in the superior hemi-fields (k=0.574).

Conclusions: Spatial summation in the central visual field has specific values for every position, with slight variations depending on the eccentricity of the stimulus. 

PSYCHOPHYSICAL ASSESSMENT OF CONTRAST GAIN ABNORMALITIES IN MAGNOCELLULAR AND PARVOCELLULAR PATHWAYS IN GLAUCOMA

AM McKendrick¹, DR Badcock¹, WH Morgan². School of Psychology, University of Western Australia, Perth, Australia¹, McCusker Glaucoma Centre, Lions Eye Institute, Perth, Australia².

Purpose: It is established that contrast sensitivity loss occurs in glaucoma. Contrast sensitivity is only one aspect of contrast processing. Another aspect is contrast gain control which enables adjustment of neuronal sensitivity to enable optimal response to the prevailing contrast conditions. This study compared the contribution of abnormalities of contrast sensitivity and contrast gain control to contrast processing abnormalities in early glaucoma.

Methods: Seventeen patients with primary open angle glaucoma and 17 approximately age-matched controls participated. Subjects were assessed foveally and mid-peripherally (12.5°). Glaucoma subjects were tested in a mid-peripheral region of normal visual field (neighbouring locations required to be within the normal 95% confidence limit on the Total Deviation plot of their most recent SITA/Full Threshold Humphrey Field Analyzer assessment). Control subjects were tested in matched locations. Contrast discrimination was assessed using the steady-pedestal (Magnocellular [M] pathway) and pulsed-pedestal (Parvocellular [P] pathway) stimuli of Pokorny and Smith (1997) for seven pedestal luminances between 15 and 75 cd/m², presented on a background of 30 cd/m².

Results: Glaucoma group thresholds were significantly elevated compared to controls foveally and peripherally on both the pulsed-pedestal (P) and steady-pedestal (M) tasks (p<0.01). Effect size statistics revealed a similar magnitude of deficits on M and P tasks. Foveal deficits were of a magnitude that could be explained by reduced contrast sensitivity, however the peripheral deficits were greater than predicted by reduced contrast sensitivity alone, consistent with an abnormality of contrast gain control.

Conclusions: Foveal and mid-peripheral dysfunction of M and P pathway contrast processing was identified in people with glaucoma, in areas of normal visual field performance. These findings are supportive of non-pathway-selective neural adaptation abnormalities in early glaucoma. This study suggests that contrast gain assessment may prove useful in detecting early glaucomatous functional loss.

Pokorny & Smith (1997) J Opt Soc Am A. 14: 2477-86.

THE INFLUENCE OF LIGHT SCATTERING ON FDT2 MEASUREMENTS

MJ Fredette, D Budenz, DR Anderson. Bascom Palmer Eye Institute, University of Miami, Miami, Florida, USA.

Purpose: To investigate the impact of light scattering, an optical property of certain type of cataracts, on the new frequency-doubling technology perimetry (FDT2: Humphrey Matrix with Welch Allyn FDT).

Method: 10 normal subjects (10 eyes) underwent 5 FDT2 measurements with 5 randomly ordered ground-glass diffusers including one clear glass. The effective optical densities of the diffusers were 0.06 (clear glass), 0.36, 0.78, 1.22, 1.67 log unit. The glare effect produced by these diffusers, measured by contrast sensitivity with a glare source (Miller-Nadler glare test), range from  perception of the 5% contrast target to non-perception of the 75% contrast target. FDT2 MD, PSD and threshold values (dB) at 0 degree, 4.3 deg, 8.6 deg, 12.9 deg, 17.2 deg and 21.5 deg nasally along the 180 degree supero-central meridian were analyzed.

Results: The mean decrease in FDT2 MD were 6 dB, 15 dB , 23 dB and 28 dB respectively with 0.36, 0.78, 1.22 and 1.67 diffusers (statistically significant with p<0.01 for each diffuser compared to the clear glass). On PSD, no statistically significant change were recorded with each diffuser (p>0.05). Threshold value decrease was similar at every eccentricity (0 to 21.5 degree). Using the same diffusers, standard automated perimetry with Humphrey Visual Field were previously shown to decrease the mean threshold value by 3.3 dB, 8.1 dB, 12.6 dB and 18.3 dB respectively for the same diffusers.

Conclusion: Even minimal light scattering, such as might be caused by a mild  cataract, may influence the threshold values measured by frequency-doubling technology; just as was previously shown for standard automated perimetric measurements. The difference in the magnitude of the impact may be explained by the different definition and conversion factor used to transform perceived contrast/light to a decibel scale.  Studies on eyes with cataracts that have scattering properties are needed.  Might abnormal optical properties of the ocular media help explain cases when decibel values of SAP and FDT2 Matrix fail to match?

Analysis of Local Damage to the Optic Nerve in Glaucoma with the Multifocal Electroretinogram Compared to the Multifocal Visual Evoked Potential Technique

R Blanco, R.Stamper & EE Sutter
Smith Kettlewell Eye Institute. San Francisco.California.
Dep. Ophthalmology. University of California San Francisco.
United States

 Purpose: 1. To evaluate the electrophysiologic function in glaucoma by using a new protocol of the multifocal electroretinogram (mf-ERG) that emphasizes response contributions from ganglion cell fibers (optic nerve head component (ONHC)).  2. To compare glaucomatous losses in the ONHC with those estimated through inter-ocular comparison of the multifocal visual evoked potentials (mf-VEP).

Methods: mfERGs and mfVEPs of 26 individuals with glaucoma and 26 normal subjects were recorded and analysed with the VERIS 5.1 multifocal recording system. Stimulation: The special, ganglion cell response enhancing protocol consisted of multifocal flash stimuli interleaved with two global flashes presented 13.3 ms and 40 ms after each multifocal frame. The intensity of both multifocal and global flashes was 2.7 cd•s/m^2. The stimulus array consisted of 103 scaled hexagons. The recording time was 9 minutes per each eye. Pupils were dilated. The mfVEP stimulus consisted of a 60-sector dartboard grid, with each sector containing a contrast reversing check pattern. The mean stimulus luminance was 200 cd/m^2 viewed through a natural pupil. All multifocal stimulus arrays subtended ca. 45 degrees. The net recording time was 14 minutes per eye. Analysis: The effect induced by the focal flashes on the second one of the following global flashes contains the most prominent ONHC and was thus used for the evaluation of the mfERG data. Inter-ocular differences in focal VEP amplitude ratios were evaluated against those due to the noise contamination in each record.

Results:  In advanced glaucoma the ONHC was mostly extinct. In early glaucoma areas with a visibly reduced ONHC generally matched but exceeded areas with local sensitivity changes seen in visual field by standard automatic achromatic perimetry. This suggests a far more undetected advanced damage to the ganglion cells in glaucoma. While large local inter-ocular differences were easily detected with the mfVEP technique, homonomous bilateral damage could not easily be distinguished from local signal reduction due to the convoluted cortical anatomy.

Conclusions: Our data suggest that in glaucoma, where the presentation is commonly bilateral, the ONHC protocol of the mfERG provides a better topographic evaluation than the inter-ocular mfVEP analysis. While a larger scale evaluation is needed, the study suggests that the ONHC analysis may outperform standard achromatic perimetry in sensitivity and reproducibility.

Supported by NIH grant EY06861, FIS grant 02/0926 and The Smith-Kettlewell Eye Research Foundation

Effect of Retinal Thickness on Central visual fields in diabetic retinopathy

AS Neubauer¹, C Chryssafis¹, C Hirneiss¹, MJ Thiel¹, MW Ulbig¹, A Kampik¹.

Dept. of Ophthalmology, Ludwig-Maximilians University, Muenchen, Germany¹.

Purpose: While central retinal thickness is known to correlate well with visual acuity, little is known about the effect of posterior pole thickening on the central visual field. This study therefore investigates the correlation of central visual field and retinal thickness in diabetic patients. 

Methods: On 39 eyes from 39 patients with systemic diabetes besides a complete clinical examination a 10-2 HFA perimetry and objective measurements of retinal thickness were performed by optical coherence tomography (OCT) and retinal thickness analyzer (RTA). Twenty-six patients had  previously received focal laser therapy, while the remaining 13 did never receive any laser treatment.

Results: A good correlation of visual acuity and central retinal thickness was found with both instruments, OCT and RTA, which correlated highly with each other (r=0.82, p<0.000). Visual acuity also correlated strongly with the mean defect (MD) on visual fields (VF; r=0.50, p=0.001) but not with the pattern standard deviation (PSD). Subgroup analysis was performed on three groups: 1) clinically no diabetic retinopathy, 2) clinically significant macular edema and 3) macular edema after focal laser treatment. Although group 1 had a tendency towards a lower MD and PSD on ANOVA no significant difference was found between the groups. Retinal thickness in group 1 without retinopathy was significantly lower than the other groups. The eyes after laser treatment (group 3) showed a tendency towards lower retinal thickness and better visual acuity than group 2 but the same MD and PSD on VF. A good topographic correlation of VF defects and retinal thickening could be shown.

Conclusions: Central macular retinal thickness is vital for visual acuity. Thickened retina corresponds topographic to scotomas in visual fields. Focal laser treatment seems not to cause significant field defects.

AUTOMATED STATIC AND KINETIC VISUAL FIELD TESTING IN CHILDREN

BK Wabbels, S Wilscher

University of Regensburg, Dept. of Paediatric Ophthalmology, Strabismology and Ophthalmogenetics (Director: Prof. Dr. B. Lorenz), Franz-Josef-Strauss Allee 11, D-93053 Regensburg, Germany

Purpose: Visual field testing in children is always a challenge. Testing is hampered by fatigue effects, rapid boredom, lack of comprehension and easy distraction. The few studies concerning automated static visual field testing in children mainly agree that interindividual variability is high. We now tested feasibility and outcome of visual field testing in a standard clinical setting, comparing fast threshold strategy with the new CLIP (continuous light increment perimetry) strategy and using automated kinetic perimetry.

Methods: We examined 28 children aged 5-14 years at the Twinfield-perimeter. Included were healthy children, children with unilateral pathologies (normal eye tested) and children with strabismus. Automated kinetic perimetry was performed according to the Goldmann-standard with a test velocity of 2°/s with stimuli III/₄, I/₄, I/₂ and I/₁. Static perimetry was done with fast threshold strategy and CLIP-strategy in randomised order. One eye per subject was examined, each test was performed twice.

Results: Interindividual variability was high, even at the same age. Not all children were able to complete the entire procedure. Starting from age 7, reliable results were obtained in many children, starting from age 12, in most cases adult testing strategies were possible with good reproducibility for static and kinetic testing. CLIP seemed easier to perform than fast threshold, and found higher mean sensitivities. There was no significant difference between the children with strabismus and the other children.

Conclusion: Formal visual field testing in children is time consuming, but can be successfully performed in many children. Test performance was more dependent on the child’s maturity and ability to concentrate than on age. If visual field testing is planned, desired accuracy and feasibility should be balanced. It is likely that children with visual disturbances might experience greater difficulty in undergoing such examinations than otherwise healthy children. Automated kinetic perimetry could be used to reliably monitor children under Vigabatrin therapy.

Central visual field in patients affected by diffuse macular edema, after intraocular injection of triamcinolone acetonide.

Morescalchi F., Gandolfo F., Musig A., Rovida F., Turano R., and Gandolfo E.

University Eye Clinic of  Brescia (Italy).

Purpose: To investigate the effects on visual performance of intravitreal triamcinolone acetonide for the treatment of refractory diabetic macular edema.

Methods: Prospective case series study including eyes with significant diffuse or cystoid diabetic macular edema unresponsive to previous sessions of grid or focal laser treatment. In all patients included in the study group, 20 mg of triamcinolone acetonide were injected into the vitreous cavity. The visual and the anatomic responses were observed as well as complication related to the procedure. Visual acuity (using the ETDRS chart), reading speed, contrast sensitivity, visual field test (program 10-2, HFA II) and fluoroangiographic examination were assessed.

Results: Thirty eyes of 27 patients completed 6 or more months of follow-up. The mean ETDRS visual acuity improved from 20/80 to 20/40 at the 6^th month follow-up visit. A dramatic improvement of contrast sensitivity and reading speed was registered in most of patients. This result was more consistent than the visual acuity improvement. The final reading speed value (mean: 98 words/minutes) was significantly better than the preoperative one (mean: 59 words/minutes). The central visual field improved in 22 patients. The central 4° showed the most significant improvement of mean sensitivity. There was a complete resolution of diabetic macular edema, documented by fluoroangiography in all eyes within 30 days after injection and in 24 eyes at the 6^th month follow-up. All patients were treated with topical ocular antihypertensive therapy; 2 patients experienced IOP elevation of 28 mmHg in spite of the maximal topical antiglaucomatous therapy. After ALT, the IOP levels returned to normality in both cases.

Conclusions: Intravitreal triamcinolone is a promising therapeutic method that improves the quality of vision of patients affected by diabetic macular edema that fails to respond to conventional laser treatments. No vitreo-retinal complications (retinal detachment, vitreous haemorrhage or endophthalmitis) were experienced in the current study group. The major ocular side effect was IOP elevation but, in this series, the IOP of all patients was normal without any therapy after 6 months.

The improvement in reading speed and contrast sensitivity was correlated to the improvement of the central retinal sensitivity assessed with standard automatic perimetry. Further studies are necessary to assess the long-term efficacy of this therapy.

Predicting reading speed from central visual field results.

Morescalchi F., Gandolfo F., Rovida F., Turano R, Redini A., and Gandolfo E.

University Eye Clinic of Brescia (Italy).

Purpose. To investigate the correlation between reading speed and mean sensitivity of different areas of the central visual field in patients with relative central scotomas caused by diffuse macular edema.

Methods: 90 visual fields were obtained in patients with varying degree of diffuse macular edema, using the Humphrey 10-2 sita standard procedure. All patients underwent at least two previous perimetric examinations and those with a history of poor fixation were excluded from the study. The average sensitivity of nine areas of the central visual field were examined: central 2°, 4°, 6°, 8°, 10° and four quadrants (upper-right and left, lower-right and left) of the central 4°.

All patients underwent a complete evaluation of macular functions including high and low contrast distance visual acuity, near reading acuity and oral reading speed. Differences in correlation between each functional parameter and oral reading speed were evaluated.

Results: The central 2° and 4° (r = 0.78 and r = 0.79, P = 0.000) and the 4° (r = 0.82, P = 0.000) upper-right quadrant provided better prediction of oral reading speed than the other parameters including MD (r = 0.62, P = 0.001) and distance visual acuity (r = -0.51, P = 0.013). Patients with a scotoma in the upper-right quadrant were significantly slower in reading performance. This finding was confirmed by the positive correlation between reading speed and PSD in this area (r = 0.43, P = 0.01). A significant correlation was also found between MS and reading speed with low contrast visual acuity.

Conclusion: The capacity to read has a major impact on the quality of life. Patients with relative scotomas in their central 10° often have visual performance difficulties far exceeding what would be expected from reduced visual acuity alone. On the other hand, patients with the same low visual acuity may complain different disability in reading, task performance and daily activities. Standard automatic perimetry may help to quantify their visual disability. The mean sensitivity of the central 4° and, in particular, of the upper-right quadrant mean sensitivity may provide good estimates of functional visual performance and related quality of life.

AUTOMATED STATIC PERIMETRY IN EYES WITH CENTRAL SEROUS CHORIORETINOPATHY

H Iijima. University of Yamanashi, Yamanashi, Japan

Purpose: To study the central visual field abnormality seen in eyes with central serous chorioretinopathy (CSC), which has rarely been evaluated quantitatively using automated static perimetry in a clinical study. The author claims that the severity of the disease should be assessed by the mean deviation (MD) of the central 10-degree automated static perimetry, which is independent of the visual acuity.

Methods: We reviewed the results of Humphrey perimetry, central 10-2 program taken for 121 eyes of 118 patients (22 female and 96 male) with CSC seen in our hospital between 1993 and 2003. A total of 132 episodes of serous retinal detachment (98 initial and 34 recurrent) was studied. In order to study the effects of active serous retinal detachment on the central visual field, eyes with obvious RPE atrophy evidenced by the angiographic window defects were excluded.

Results: Central visual field abnormality in eyes with CSC varied from no defects to severe and large central scotoma. Mean deviation (MD) ranged from –21.1 dB to 2.6 dB with the mean of –3.8 dB. Thirty seven of 132 perimetric results (28%) showed MD lower than –5 dB. The best corrected visual acuity ranged between 0.07 and 1.5 with the median of 0.7. The correlation between MD and logarithm of minimum angle of resolution (logMAR) was not significant (r = -0.116, P = .186) implying that the severity of central field defects could not be predicted from the visual acuity.

Conclusion: Eyes with CSC show various degrees of central visual field loss.  Poor correlation between MD and corrected visual acuity implies that many eyes with CSC showing near normal visual acuity may suffer from severe central visual field abnormality. Automated static perimetry measuring the central 10-degree visual field in an eye with CSC provides additional information for assessing visual disability that could not be predicted by visual acuity testing
 

QUANTIFICATION OF METAMORPHOPSIA

IN PATIENTS WITH MACULAR HOLE USING M-CHARTS^TM

E. Arimura, C. Matsumoto, S. Okuyama, S. Takada, S. Hashimoto, Y. Shimomura

Department of Ophthalmology, Kinki University School of Medicine, Osaka-Sayama, Japan

Purpose: In case of macular hole, metamorphosia is one of the most important symptoms as well as the visual loss and the central scotoma. We investigated the relationship between the degree of the metamorphopsia and the morphological changes of macular hole.

Subjects and Methods: Using M-CHARTS™ developed by us, we quantified the metamorphopsia scores of 35 eyes of 35 patients with idiopathic macular hole. We also evaluated the corrected visual acuity and the central 10° differential light sensitivity using Octopus 101 program M2.  The size of the macular hole and the fluid cuff were measured using scanning laser ophthalmoscope (SLO) and OCT3 images. In 22 patients, we also evaluated improvement of the metamorphopsia scores after vitrectomy.

Results: The metamorphopsia of macular hole patients were characterized by the straight lines  bending toward the central scotoma. There was a significant correlation between the metamorphopsia score and the fluid cuff size. After vitrectomy, the visual acuity improved in 14 patients and the metamorphopsia sores improved in 16 patients.

Conclusion: M-CHARTS™ is a simple and useful method for the quantification and the follow-up of metamorphopsia in patients with macular hole.

TRANSITORY FUNCTIONAL DEFECTS SHOWED BY TENDENCY ORIENTED PERIMETRY (TOP) IN PATIENTS WITH MILD BRAIN TRAUMA

E Sorli ¹, M Gonzalez de la Rosa², A Fons³, JM Gonzalez-Darder⁴

1.- Ophthalmology Department. General Hospital of Castellon, Spain

2.- Ophthalmology Department, University of La Laguna, Tenerife, Spain

3.- Ophthalmology Department, Clinic Hospital, University of Valencia, Spain

4.- Neurosurgery Department, General Hospital of Castellon, Spain

Purpose: To analyze the results of the TOP perimetric strategy in patients with mild brain trauma. 

Methods: Both eyes of 36 patients, without previous ocular, vascular or neurological pathology; affected of mild brain trauma, (GCS among 13-15, without pathological findings in Computer Tomography examination, with loss of consciousness for less than 30 minutes and episode of amnesia for less than 24 hours), were examined using the TOP strategy and the Octopus 1-2-3 three times (twice a week after the traumatism and once a month later). The obtained data was compared with a control group of 36 subjects examined twice.

Results: The MD and LV square root (sLV) were significantly higher in the first two examinations (MD = 3.1 and 4.0dB, sLV = 2.8 and 2.9dB) than in the third one (MD = 1.0dB, sLV = 1.9dB) (p<0.01). There were not differences between the MD and sLV of the third exam of the pathological group and the two examinations of the control group (MD = 0.6 and 0.7dB, sLV = 1.9 and 1.8dB) (p>0.05), indicating a complete recovery. The test-retest threshold fluctuation of the pathological group (3.2 +/- 2.9dB) showed a highly significant difference with regard to the control group (1.2 +/- 1.2dB) (p<0.01). 28.9% of the examined points in the first two examinations of the pathological group showed relative scotomas. Among them, 52.1% were reproduced in the same position in both examinations and 48.2% coincided in homonimus points of both eyes.

Conclusions: It seems that a transitory neurological defect occurs in these patients; it can be evidenced as a diffuse deterioration of the retinal sensibility, with increased threshold fluctuation. The analysis of the focal visual field loss disposition indicates that there is a neurological suffering with some certain topographical preferences, which is specific in each case.

REHABILITATION OF THE VISUAL FIELD IN CEREBRAL LESIONS

F Dannheim¹, D Verlohr². Dept. of Ophthalmology, General Hospital Hamburg-Harburg, GER¹, Centre for Neurological Therapy, Jesteburg, GER².

Purpose: 20 - 30% of patients in rehabilitation centres for neurological disorders suffer from visual field deficits, most often as homonymous defects. Training programs for restitution of function by repeated stimulation of the defective field under steady fixation are of limited value. A diagnostic program is presented for the evaluation and monitoring of field defects during compensational training of search saccades.

Method: A computer program was developed projecting a scenic screen on which a central fixation icon is superimposed alternating with a search icon in 11 positions within the 45° field in random order. The time required to detect the search icon is recorded for each presentation. 2 consecutive series allow an estimation of short term fluctuation or short term training effects.

Results: In a pilot study, this program has been applied to a group of 10 patients suffering mainly from hemianopic field defects of different origin. A follow up after training of search saccades was possible in some of them. The time lag for the search in the affected hemifield was reduced. This effect corresponded to the normalization of a computer assisted line division test.

Conclusion: A diagnostic test for the evaluation of practical performance in the visual environment allows to monitor training effects of compensational visual rehabilitation in cerebral lesions.

kin-train – a computer-based

interactive teaching and scoring tool

FOR Practicing kinetic perimetry

U Schiefer¹, K Nowomiejska^1,2, E Krapp¹, J Pätzold¹.

¹University Eye Hospital, Tuebingen, Germany, ²1^st Eye Hospital, Lublin, Poland

Purpose: To create an interactive computer-based teaching software for kinetic perimetry with an implemented evaluation system for scoring examination technique.

Methods: This software is based on the original user interface of the new semi-automated kinetic perimetry (SKP) feature of the OCTOPUS 101 perimeter (HAAG-STREIT, Koeniz, CH).

The trainer creates a (pseudo-) 3D individual “hill of vision” with a specific pathology. Response characteristics can be modified by altering the individual frequency of seeing curve, reaction time, fixation quality, ocular alignment, and pupil size of the virtual patient. The trainer may also enter relevant findings and images regarding patient´s history, current complaints etc. into a comprehensive electronic medical chart.

The trainee can individually select target characteristics (angular velocity, stimulus size and luminance). He also independently defines origin, end and thereby direction of each kinetic stimulus with the help of so-called vectors. As soon as the kinetic stimulus, moving along a given vector, is perceived by the virtual patient according to the previously defined conditions, this position is marked as “local kinetic threshold”. Quality of the perimetric examination can be quantitatively assessed by the ratio of intersection area and union area of the individual trainee´s result and the related trainer-defined original isopter. These ratio values and other parameters, such as detection of the blind spot, classification of the assessed scotoma, examination duration etc., determine a score. Weight of the parameters can be adjusted. The score allows for certification of the trainee, based on predefined standards.

Results: Meanwhile representative scotoma patterns (e.g. hemianopic field loss, nerve fiber related defects, concentric constrictions, central scotoma), together with related patients´ charts have been entered into the actual training software version, which has already been successfully used in several perimetric courses.

Conclusion: KIN-TRAIN  is a computer-based interactive learning tool, which allows for certifiable education in kinetic perimetry.

Fundus perimetry with the MP 1 in normals – comparison to conventional threshold perimetry

C Springer, K Rohrschneider. Department of Ophthalmology, University of Heidelberg, Heidelberg, Germany.

Purpose: The Micro Perimeter 1 permits an automated full threshold static fundus perimetry. For the evaluation of the perimetric results normal values are essential, but do not exist up to now. Aim of this study was to determine light sensitivity threshold values obtained with the MP1 in healthy volunteers and to correlate them with conventional automated static perimetry using the Octopus 101 Perimeter.

Methods: In 25 healthy eyes of 25 healthy volunteers static threshold perimetry was performed with the Octopus 101 (Haag-Streit AG, Switzerland), program CT1, and the MP1 Micro Perimeter (Nidek Inc., Italy) in random order. Light increment threshold sensitivity values were compared for 21 matching points in a rectangular test grid using similar examination settings (Goldmann III stimuli, presentation time 100 ms, white background illumination of 1.27 cd/m²). The Octopus 101 CT1 program tested 77 locations in a rectangular 6-degree grid in an area of 28 x 28 degrees while the MP1 tested 70 stimulus locations in a rectangular 3-degree grid covering an area of 27 x 18 degrees.

Results: For the 21 matching locations mean light sensitivity was 15.5 dB (range 13.0-17.1 dB) with the MP 1 and 30.3 dB (range 27.7-33.9 dB) with the Octopus. On the average the Octopus showed higher threshold values for all test locations than the MP1 did. The mean difference between both examinations was 14.8 ± 1.3 dB comprising all locations and 15 ± 1.2 dB excluding the test locations at the blind spot. The difference between the two devices varied from 12.3 to 17.5 dB showing nearly the same difference for each location except in the surrounding of the blind spot.

Conclusion: The results display that there exists a systemic difference between both devices of about 15 dB. The stimulus location did not influence this difference significantly. Light sensitivity values in microperimetry with the MP1 are comparable to the threshold values obtained with the Octopus 101 using a correction factor of about 15 dB.

Fundus perimetry in the long‑term follow-up of Stargardt’s disease

K Rohrschneider, C Springer. Department of Ophthalmology, University of Heidelberg, Heidelberg, Germany.

Purpose: To assess and evaluate functional changes in fundus perimetry during the long-term follow-up of patients with Stargardt’s disease and fundus flavimaculatus.

Methods: Fundus perimetry with the Scanning Laser Ophthalmoscope was performed in 35 eyes of 18 patients with stargardt’s macular dystrophy or fundus flavimaculatus over a mean follow-up period of 4 years (1 to 9 years). Static threshold perimetry with a 4-2-1 staircase strategy and Goldmann III stimuli was carried out with an average number of 60 stimuli. The depth and size of the scotomata as well as the stability and location of fixation (preferred retinal locus, PRL) during perimetry were analyzed for each examination. The results were compared intraindividually in relation to the first exam and correlated with best corrected visual acuity.

Results: Fundus perimetry lasted for 521 ± 194 sec. During the baseline perimetric exam 31 eyes (89%) showed an absolute central scotoma measuring 4.5 cm² in average. During follow-up this central scotoma increased by 3.2 cm² in average while there was no significant change in scotoma depth.

Regarding the area and stability of fixation the mean number of patients already used an extrafoveal retinal locus for fixation at the beginning of the follow-up, the mean deviation around the mean fixation point was 0.84 degrees. During follow-up stability of fixation did not change significantly while the PRL remained at the upper border of the scotoma and moved upwards according to the increasing scotoma size. Visual acuity remained stable over the follow-up (mean progression 0,06 lines) at an average of 20/100.

Conclusions: Fundus perimetry represents an effective device to evaluate functional changes in the follow-up of patients with Stargardt’s disease. Although visual acuity showed only minor changes, an enlargement of the central scotoma and a movement of the PRL were observed during fundus perimetry explaining the increasing problems during reading and visual performance over time.

influence of stimulus duration and fixation object in fundus perimetry

C Springer, K Rohrschneider. Department of Ophthalmology, University of Heidelberg, Heidelberg, Germany.

Purpose: Fundus perimetry is useful in the assessment of the macular function. Aim of this study was to determine and quantify the influence of different parameter settings on perimetric results in fundus-controlled perimetry obtained with the MP1 Microperimeter.

Methods: In 66 healthy eyes of 33 volunteers microperimetry was performed with the MP1 Micro Perimeter (Nidek Inc., Italy) using two different settings for stimulus duration and fixation object in random order. Stimulus presentation time was set to 100 ms in one eye and to 200 ms in the other eye. Either a black or a red cross served as fixation object. A rectangular 3-degree grid with 70 stimulus locations covering an area of 27 x18 degrees and a 4-2-1-staircase strategy was used in both settings. For each test point location light sensitivity threshold values were analyzed and compared intraindividually between both eyes. Fixation stability was assessed by evaluating the deviation from the mean fixation point during microperimetry.

Results: Light sensitivity threshold values between the two settings varied according to the chosen stimulus presentation time. For a stimulus duration of 200 ms the mean light sensitivity (18.0 ± 0.7 dB) was significantly higher than for the 100 ms stimulus duration (16.0 ± 0.6 dB). Fixation stability employing a red cross as fixation object was superior to the use of a black cross. While the eyes fixating a red cross showed a stable fixation in 93.7 %, only 78.8 % of the eyes looking at a black cross had a stable fixation.

Conclusion: Light sensitivity values in microperimetry with the MP1 depend highly on the choice of parameter settings. Reduction of the stimulus presentation time from 200 ms to 100 ms leads to a decrease of light sensitivity threshold values of about 2 dB. A red and well visible fixation object can augment the fixation stability in comparison to a black fixation object. The choice of parameter settings can influence microperimetric results and should therefore be carefully selected prior to examination.

AN EXPERIMENTAL AUTOMATIC PERIMETER THAT DISPLAYS THE FUNDUS IMAGE ON A liquid crystal display AND CAN detect visual LOss USING VERY SMALL TARGETS

T Sawada¹, Y Takihata², M Tajika², T Minamikawa², Y Nishida², K Kani². Toyosato Hospital, Shiga, JAPAN¹, Shiga University of Medical Science, Shiga, JAPAN²

Purpose: We have developed an experimental automatic perimeter that displays the fundus image on a monitor, for measuring the visual field disturbance in cases of nerve fiber bundle defect (NFBD), using very small targets.

Methods: The system of our perimeter was developed on Visual Basic.net. Two liquid crystal displays (LCD) were used. One was used as the campimeter and the other as the monitor for examination. The position of the macula and the center of the optic disc were marked after testing the blind spot. The examiner could decide measuring points while observing the fundus image on the LCD screen. The test target was white, 2.9 min in diameter, and 100 ms in duration.

Results: Ten cases were examined. Scotomata were detected in cases in which no changes could be detected with the conventional automatic perimeter.

Conclusions: These results suggested that using very small targets was useful in detecting visual losses in NFBD when the receptive fields were sparse.

AUTOMATED FLICKER PERIMETRY USING OCTOPUS 311

C. Matsumoto, S. Takada, S. Okuyama, E. Arimura, S. Hashimoto and Y. Shimomura

Department of Ophthalmology, Kinki University School of Medicine, Osaka-Sayama, Japan

Purpose: Using a new automated perimeter Octopus 311, we evaluated the clinical usefulness of the flicker perimetry in glaucoma and glaucoma suspect patients.

Subjects and Methods: Forty five eyes of 45 normal subjects, 62 eyes of 62 glaucoma patients, 28 eyes of 28 glaucoma suspect patients were examined by light-sense perimetry, flicker perimetry and frequency doubling perimetry. Flicker perimetry was performed using the Octopus 311 and its remote software package.  The suprathreshold 4-zone ‘probability’ strategy was used for classify the CFF probability level.  The classified levels were set at 5%, 1% of probability of normality and 5 Hz.  Frequency doubling perimetry was performed using C-24-5, C-24-1, N-30-5, N-30-1 screening program using the Matrix. Light-sense perimetry was performed using HFAII full threshold program 24-2. The OCT3 was used for evaluating the RNFL thickness in all glaucoma and glaucoma suspect patients. 

Results: In the early stage of glaucoma, the area under ROC curves were about 0.94 in both flicker and Matrix N-30-1, and 0.89 in C-24-1.  In the moderate and advanced stages of glaucoma, the areas under ROC curves were almost 1.0 in all tests. In glaucoma suspect and normal hemifields of glaucoma patients, abnormal CFF values were detected by flicker perimetry corresponding with the fundus changes by OCT3. The average test duration of flicker perimetry was about 3 minutes in normal eyes and about 5.5 minutes in glaucoma patients.

Conclusion: The 4-zone ‘probability’ strategy in Octopus 311 is a useful method for evaluating the flicker field in glaucoma and glaucoma suspect patients.

RAREBIT PERIMETRY: A NEW NON-CONVENTIONAL VISUAL FIELD TESTING METHOD FOR EARLY GLAUCOMATOUS FUNCTIONAL DAMAGE DETECTION.

Paolo Brusini, Maria Letizia Salvetat, Lucia Parisi

Department of Ophthalmology – S. Maria della Misericordia Hospital, Udine (Italy)

Purpose: To evaluate the use of Frisén’s Rarebit Perimetry (RBP), a new non-conventional visual field testing method, in detecting early glaucomatous visual field damage, and to compare it with standard automated perimetry (SAP)

Methods: 43 patients with ocular hypertension (OHT), 39 patients with early primary open-angle glaucoma (POAG) and 41 normal controls were considered. All patients underwent testing with both the Humphrey Field Analyzer (HFA) SITA 30-2 program, and with RBP. In the analysis of the results, the following items were taken into consideration: the HFA-MD and HFA-PSD; the RBP mean hit rate (MHR); and, the number and pattern of non-hit-rate areas.

Sensitivity (Se), specificity (Sp) and the area under the receiver operating characteristic curve (AROC) for RBP, were all calculated using different algorithms. 

Results: The RBP-MHR was 88.6 % ± 14.8 in the control group, 79.1 % ± 10.9 in the OHT group and 64.3 % ± 13.8 in POAG group (differences statistically significant). The Number and the magnitude of the RBP-non-hit-rate areas were significantly higher in the POAG group. The largest AROC (0.95), giving rise to a Se of 97.4% and a Sp of 92.7%, was obtained when an abnormal RBP test was defined as having at least one of the following conditions: 1) MHR <80%; 2) >15 areas having a non-hit-rate of >10%; 3) >2 areas with a non-hit-rate of >50%; or, 4) at least 1 area with a non-hit-rate of >70%. Abnormal RBP results were observed in 44.2%-65.1% of the patients in the OHT group.

Conclusions: The RBP is a quick, comfortable and inexpensive non-conventional perimetric test. It is easily available (only a PC device is required), and is quite effective in detecting early glaucomatous functional defects.

RAREBIT: A NOVEL PERIMETRIC STRATEGY FOR DETECTION OF SUBTLE OPTIC NERVE DYSFUNCTION.

SA Newman, SW Whitford.  University of Virginia Health Science Center, Charlottesville, Virginia, USA.

Purpose: Because of visual system redundancy, subtle visual field defects may not be obvious on standard automated perimetry (SAP). A recently developed program (by Dr. Lars Frisen) that uses tiny points of light, RAREbit, purportedly better detects subtle defects. We undertook a study of patients with known asymmetric optic nerve function, yet normal fields using a Humphrey 24-2 SITA-fast program (size III test object) to determine the ability of RAREbit perimetry to accurately detect subtle field defects missed by SAP.

Methods: 7 patients (5 women, 2 men) with equal acuity but asymmetric optic nerve function (based on the presence of an afferent papillary defect) agreed to participate in the study. All 7 with normal (symmetric with the uninvolved eye) Humphrey 24-2 SITA-fast fields using a size III test object were retested using the 24-2 full-threshold program utilizing a size I test object.  This was followed by the RAREbit “rabbit” test utilizing the protocol outlined by Lars Frisen. The three results for each eye were then compared side by side for asymmetry that would corroborate the presence of  asymmetric optic nerve function.

Results: The RAREbit perimetry “rabbit” test invariably (7 of 7) detected the eye with the afferent pupil defect. This was corroborated by the size I FT Humphrey 24-2 in all but one case. The “rabbit” test took slightly more than half of the time required of the Humphrey 24-2 FT size I test (average “rabbit” time 6 minutes 34 seconds compared to average 24-2 FT size I time of 11 minutes 26 seconds.)

Conclusions: RAREbit perimetry can detect subtle field defects that are not seen on Humphrey 24-2 SF size III testing. The test is faster and easy to perform than the 24-2 I size.

Virtual Perimetry for glaucoma screening in the population

Sean Ianchulev MD MPH, Peter Pham MD, Vladimir Makarov PhD,  Brian Francis MD, Don Minckler MD

Purpose: Peristat is a virtual perimetry system that allows self-testing on any standard computer monitor via Internet connection.  Sensitivity and specificity of Peristat to detect visual field defects were compared to standard Humphrey Visual Field Analyzer data.

Design: Prospective, comparative observational case series

Participants: 58 eyes of 33 patients.

Main Outcome Measures/Testing:  Semi-quantitative analyses comparing Peristat and Humphrey visual field scores. The study evaluated patients with an established glaucoma diagnosis and glaucoma suspects who had undergone comprehensive ophthalmologic examinations including prior office perimetry evaluation (Humphrey Field Analyzer).  Inclusion criteria were: diagnosis or suspicion of glaucoma, best corrected visual acuity better than 20/200 and reliable performance on prior standardized office perimetry. Computer literacy was not required and over 40% of the patients tested were computer illiterate, with no previous computer or Internet experience. All of the glaucoma suspects had cup-to-disc ratios greater than 0.5. A total of 58 eyes [of 33 patients] were interrogated with the Humphrey and the Peristat systems – 10 eyes of 5 patients without documented glaucomatous field loss and 48 eyes with mild-to-severe scotomas by standard 24-2 office perimetry. Severe glaucomatous field damage was defined as MD>10; mild-moderate visual field defects were defined as MD <10; and normal visual fields had no deviation from age-matched controls as defined by the Humphrey system. A standard computer set-up was used with a 17” monitor, keyboard and mouse. The program was delivered through a remote connection with a server and the patients interacted unassisted, after a brief instruction on the browser-enabled program interface.

Results:  All patients completed the Peristat test without difficulty.  Testing time varied between 3 and 9 minutes, tending to be longer with more severe visual field defects. Test results were reviewed, masked, by 2 glaucoma specialists and 1 general ophthalmologist. Each quadrant for every eye tested was graded for visual field defects for both the Humphrey and the Peristat. The density of the scotomas was graded on a scale from 0-3 (O = none/artifact/nonspecific; 1 = minimal; 2 = moderate; 3 = severe scotomas). The Peristat demonstrated a high degree of correlation with the Humphrey instrument. Among the three reviewers, sensitivity ranged from 80-83%.  Similarly, test specificity was between 94% and 96% for all three reviewers. The inter-observer variability was negligible. In a second sub-analysis in which cases with mild defects were excluded, the Peristat’s efficacy was further optimized – sensitivity between 84%-86% and specificity between 94% and 97%. Patients performed the Peristat test with similar facility to their Humphrey test. Fixation losses and test reliability were comparable for both. 

Conclusion: Peristat is a reliable self-test which demonstrates high clinical utility for the detection of visual field defects at a fraction of the cost of standard office perimetry. In selected populations, the Peristat could be a valuable tool for cost-effective self-screening for visual field loss and detection of glaucoma.

TESTING A NEW APPROACH TO DETECTING CHANGE IN SERIES OF RETINAL IMAGES ACQUIRED FROM SCANNING LASER TOMOGRAPHY

AJ Patterson¹, DP Crabb¹, DF Garway-Heath².

The Nottingham Trent University, Nottingham, UK¹, Moorfields Eye Hospital, London, UK².

Purpose: To evaluate a new Statistic Image Mapping technique for detecting topographic changes in Heidelberg Retinal Tomograph (HRT) images in comparison with ‘Topographic Change Analysis’ (TCA super-pixel analysis; Chauhan et al 2001 Arch Ophthalmol 119:1492-99) available on the HRTII.

Methods: Fifty series of HRT images were generated using a novel computer simulation. For each series the same single HRT image was replicated 3 times to represent a baseline visit and then replicated 3 times again at 5 follow-up ‘visits’. Two types of noise were then added to these identical images; first, noise that would be attributed to movement during image acquisition was mimicked using a re-alignment algorithm acting in 3 translations and 3 rotations; next, Gaussian noise was added to each pixel by sampling from a Normal distribution with a mean set at the topographic height at that pixel. Another 50 series were generated in the same way but the topographic height of a small, discrete part of the image was fixed to deteriorate over the follow up period. We then replicated the TCA super-pixel analysis and applied the published criteria for change to the two sets of data. Next we then applied the new Statistic Image Mapping methods to the same series of images.

Results: With the 50 stable virtual ‘patients’, the TCA super-pixel method falsely flagged 12% as ‘changing’ whilst the new Statistic Image Mapping technique did not flag any as falsely ‘changing’. TCA super-pixel correctly identified 88% of the ‘changing’ patients but the new Statistic Image Mapping technique identified nearly all of the ‘changing’ patients (98% sensitivity).

Conclusion: This computer simulation experiment indicates that the Statistic Image Mapping techniques have better diagnostic precision in detecting change in series of HRT images as compared to a commercially available algorithm (TCA super-pixel). The new techniques are the subject of further computer experiments, and may prove to be clinically useful in detecting changes in the optic disc in glaucoma.

Clinical Evaluation of the Humphrey Matrix

Paul GD Spry, HM Hussein and John M Sparrow. Bristol Eye Hospital, Bristol, England.

Purpose:  The Humphrey Matrix is able to examine the central visual field using Frequency Doubling Technology with a 24-2 stimulus pattern and fast-thresholding a