Presentation: Monday Oct 26, 2009, 3:21 PM, Room NORTH 133

Background

Perimeters have poor repeatability and reliability. In an attempt to ameliorated these problems we developed a rapid, objective, noncontact multifocal perimeter. Data is not collected during fixation losses or blinks. Here we improve the method.

Precis

This study seeks to improve a new FDA approved, non-contact, objective perimeter: the TrueField Analyzer, which simultaneously assesses 44 visual field regions of each eye in 4 to 6 minutes recording time; response amplitude (sensitivity), response delay, measurement error, and data on afferent and efferent pupil defects, are provided at every field location; a 20% improvement in reliability and excellent diagnostic accuracy was obtained.

Abstract

Authors

T. Maddess., M. Kolic, R.W. Essex, A.C. James
ARC Centre of Excellence in Vision Science, Dept Ophthalmology The Canberra Hospital, Australian National University, Canberra, Australia.

Purpose

To compare 8 TrueField Analyzer methods.

Methods

Two blocks of trials contained 41 normal and 47 glaucoma subjects, and 40 normals and 39 patients. We compared 4 or 6 min stimulus durations and 2 methods of balancing luminances in an attempt to produce higher median reliability across the field as measured by t-statics for each field region. The dichoptic stimuli extended to 30 deg eccentricity. Maximum luminance and stimulus presentation rate were also varied. All subjects were also examined with Matrix and HFA perimetry, and Stratus OCT.

Results

The 44 region/eye, 1/s stimuli, gave areas under ROC plots of: moderate + severe fields of 0.86 ± 0.04, severe 0.98 ± 0.01. Median t-statistics improved by 20% to 3.28 ± 0.45.

Conclusion

High diagnostic accuracy and median t-stats were obtained from this improved binocular method.

Downloads

Check back here after the meeting to download the full paper.

Authors

Bell, A., James, A.C., Kolic, M., Essex, R.W. & Maddess, T. Centre of Excellence in Vision Science, Australian National University, Canberra, Australia.

Purpose

Using multifocal pupillographic perimetry, we examined differences in the visual fields of 23 subjects with early Type 2 Diabetes (T2D) and 23 age- and sex-matched controls.

Methods

Independent stimuli were delivered to 44 regions of each eye while pupil responses were recorded with infrared cameras. The stimuli were presented in 8 segments of 30 s, and both eyes of each subject were tested twice. The direct and consensual responses provided 88 responses/eye. We then examined the diagnostic power of the method by applying receiver operator analysis to the peak regional contraction amplitudes, time to peaks, and linear combinations of those.

Results

Dichoptic multifocal pupillography provided response amplitudes with a median z-score of 2.63 ± 0.26 SE. The diagnostic performance (expressed as areas under ROC plots) for the 8 subjects (32 fields) having T2D for at least 10 years was 0.87 ± 0.06 (mean ± SE) for response amplitude deviations from normative data, rising to 0.95 ± 0.04 when between-eye symmetry was considered. Mean pupil size did not have diagnostic power. Comparison of direct and consensual response fields indicated the observed localized field defects were afferent.

Conclusion

Reasonable diagnostic power was obtained, especially for the 16 eyes that had had T2D for more than 10 years, inferring that even in the near absence of visible diabetic retinopathy, some retinal damage had been sustained. This result, if confirmed in a wider group, suggests the method may be clinically useful in screening for early damage to the retina in T2D diabetes.

Downloads

Download the full paper here

Authors

Ted Maddess, Suzanne M Bedford, Xin-Lin Goh and Andrew C James.
ARC Centre of Excellence in Vision Science, Australian National University, Canberra, Australia.

Purpose

This preliminary study investigated a means of concurrently assessing the visual field defects of both eyes by recording pupillary responses to multifocal stimuli.

Methods

Twenty normal subjects and 26 primary open angle glaucoma patients, age and sex matched, were examined by slit-lamp, Humphrey Field Analyser II achromatic 24-2 perimetry and fundus photography. The patients had moderate to severe fields in at least one eye. Two stereoscopically arranged displays presented an array of 24 stimulus regions per eye extending from fixation to 30 degrees eccentricity. Pupil responses were recorded by video cameras under infrared illumination. Four stimulus conditions were tested: each stimulus region containing either a single or a 2 x 2 array of patches, presented either steadily for 133 ms or flickered at 15 Hz for 266 ms. Mean presentation rate was 1/s/region. The 4-min duration stimuli were presented in 8 segments of 30 s. Segments did not need to be repeated unless more than 15% of a segment record was lost as a result of blinks or fixation losses.

Results

The 48 stimuli produced 96 direct and consensual responses per subject. The single patch, nonflickered stimulus condition produced the best diagnostic performance, an area under the curve of 84%. The contraction amplitudes for that stimulus gave a median z-score of 3.2.

Conclusion

The method produced diagnostic accuracy approaching that of automated perimetry, but unlike perimetry provides standard errors for every point in each field as well as information on response delay and efferent defects. Only one pupil needs to function to measure both visual fields.

Downloads

Download the full paper here

Authors

T. Maddess(1), M. Kolic(1), RW Essex(2), A.C. James(1).
(1) ARC Centre of Excellence in Vision Science, Australian National University, Canberra, Australia.
(2) Dept. Ophthalmology, Australian National University, Canberra, Australia.

Purpose

To investigate the diagnostic power and repeatability of 8 variants of multifocal pupillographic perimetry in open angle glaucoma.

Design

Experimental design.

Participants

Eight stimulus protocols were examined in two blocks of experiments. Block 1 contained 40 normal and 39 glaucoma subjects; block two: 41 normal and 47 glaucoma subjects. Diagnosis was confirmed by examining all subjects with HFA achromatic, and Matrix 24-2 perimetry, Stratus OCT, slit lamp and tonometry. Informed written consent was obtained from all subjects under ANU ethics approval 238/04.

Methods

Independent multifocal stimuli were presented concurrently to both eyes with a dartboard layout, having 44 independent test regions/eye extending to 30 deg eccentricity. The recording duration for 5 protocols was 4 min., divided into 8 segments of 30 s each, and for the other 3 was 6 min. divided into 9 segments of 40 s. Stimuli in each protocol could differ in the presentation rate per stimulus region (0.25, 1, presentations/s), or luminosity (150, 180, 290 or 340 cd/m²). Background luminance was 10 cd/m². Since both pupils responded to stimuli from both eyes, 88 responses/eye were obtained giving 176 contraction amplitudes and 176 delays per protocol, with SE for all 352 measures. Retest was done within 4 weeks. Visual fields were classified by HFA mean defects: moderate: 6 to 12 dB, severe: >12 dB.

Main outcome measures

The relative diagnostic power of the 8 protocols was examined using areas under receiver operator plots (AUC). The signal qualities were quantified as the median t-static across regions and subjects for peak (relative) constriction amplitude. Test-rest quality was quantified by the width of the 25th to 75th and 5th to 95th percentiles on plots of visit 1 versus visit 2 defects.

Results

In Block 1 for severe fields the mean of the 20 regional amplitudes that most deviated from the normative data gave an AUC of 0.98 ± 0.01 (mean ± SE), and for combined moderate and severe fields 0.86 ± 0.04. The median t-stat for that protocol was 2.79 ± 0.29. That protocol had a mean presentation rate of 0.25/s and luminance of 150 cd/m2. These results were reproduced in Block 2 and a 6 min. version of the best protocol of Block 1 had a median t-stat of 3.26 ± 0.45, with a concomitant improvement in test-test variability.

Conclusions

This study indicates that multifocal pupil perimetry can yield acceptable diagnostic power, excellent median signal quality and test-retest variability comparable to the Matrix perimeter using a test duration equivalent to 3 min/eye. Data on efferent and afferent defects is obtained for all regions and data from blinks and fixation losses are automatically discarded. That protocol had a mean presentation rate of 0.25/region/s and luminance of 150 cd/m2. These results were reproduced in Block 2 and a 6 min. version of the best protocol of Block 1 had a median t-stat of 3.26 ± 0.45, with a concomitant improvement in test-test variability.

Downloads

Download the full paper here.

Presentation: Thursday, May 07, 2009, 8:30 AM -10:15 AM

Authors

A.C. James, X.-L. Goh, M. Kolic, R.W. Essex, T. Maddess. Centre of Excellence in Vision Science, Australian National University, Canberra, Australia.

Purpose

To investigate 4 variants of multifocal pupillographic perimetry using a prototype of the TrueField Analyser, which objectively assesses both visual fields concurrently.

Methods

We tested 35 normal and 44 glaucoma subjects. All eyes were examined with HFA achromatic, SWAP and Matrix 24-2 perimetry, Stratus OCT. Visual fields were classified by HFA mean defects: moderate: 6 to 12 dB, severe: >12 dB. Glaucoma subjects had a moderate to severe visual field in at least one eye. All subjects gave informed written consent. Multifocal stimuli of 4 min. duration, divided into 8 segments of 30s, were presented dichoptically with 24 or 44 regions/eye extending to 30o eccentricity. Individual 30s segments were only repeated if more than 15% of the data was lost due to blinks or fixation losses, both automatically monitored in real time. Mean presentation intervals (MPI) per region were 0.25, 1, or 4s. Most subjects were tested twice with each of the 4 stimuli. The background luminance was 10cd/m2 , and the maximum luminance of all stimuli was the same at 290cd/m2. The stimuli were thus not balanced for visual field sensitivity as in our more recent experiments.

Results

44 region, 1 MPI stimuli gave the best diagnostic efficiency, providing area under ROC plots for severe fields of 0.92 ± 0.03 (mean ± SE) and for combined moderate and severe fields of 0.80 ± 0.04.

Conclusion

Good diagnostic accuracy for test duration of 2 min/eye was obtained from this new binocular method. The new method resolves many of the problems of subjective perimetry.

Downloads

Download the full paper here

Presentation: Thursday, May 07, 2009, 8:30 AM -10:15 AM

Authors

T.L. Maddess(1A), M. Kolic(1A), R.W. Essex(1B), A.C. James(1A).
(A) ARC CoE Vision Science, CVS, (B) Dept of Ophthalmology, CVS, (1) Australian National University, Canberra, Australia.

Purpose

To investigate 4 variants of multifocal pupillographic perimetry in glaucoma to further explore the effects of balancing the stimulus luminances across the visual field to match the sensitivity of the pupillary field.

Methods

Following a smaller scale experiment (Kolic et al. ARVO 2009 submitted) we tested 40 normal and 39 glaucoma subjects to further explore the effects of luminance balancing. All eyes were examined with HFA achromatic, SWAP and Matrix 24-2 perimetry and Stratus OCT. Visual fields were classified by HFA mean defects: moderate: 6 to 12 dB, severe: >12 dB. Glaucoma subjects had a moderate or severe visual field in at least one eye. All subjects gave informed written consent. Multifocal stimuli having 44 test regions/eye, extending to 30 deg eccentricity, were presented concurrently to both eyes using a prototype of the TrueField Analyser. Recording duration was 4 minutes, divided into 8 segments of 30 s. Pupil diameter was monitored under infrared illumination. The 4 stimulus protocols examined differed in terms of mean presentation intervals (MPI) of 1 or 4 s per region, and balancing strategy. The balancing strategies assumed stimulus/response functions of the form R=Sz where z was 0.5 or 0.66. The peak test luminances were 150, 290 or 340 cd/m2. The backgrounds were 10 cd/m2. Almost all subjects were tested twice with the 4 stimulus variants. Diagnostic performance was assessed by areas under ROC curves (AUCs) for the N-worst response amplitudes.

Results

For all visual field severities the best AUCs were produced by a stimulus having MPI=4 s, luminance 150 cd/m2 and z=0.5. For severe fields the mean of the 20 regional amplitudes that most deviated from the normative data gave an AUC of 0.978 ± 0.012 (mean ± SE), and for combined moderate and severe fields 0.862 ± 0.038.

Conclusion

Careful balancing of the stimuli to cater to differences in field sensitivity, minimized the effects of response saturation, as characterised by the exponent z, and improved the diagnostic efficiency of pupillographic multifocal perimetry. The pupillographic method eliminates several problems associated with conventional subjective perimetry.

Downloads

Download the full paper here

Presentation: Thursday, May 07, 2009, 8:30 AM -10:15 AM

Authors

M. Kolic(1), T. Maddess(1), R.W. Essex(2), A.C. James(1).
(1) ARC CoE in Vision Science, CVS, Australian National University, Canberra, Australia; (2) Department of Ophthalmology, Canberra Hospital, Canberra, Australia.

Purpose

To investigate 4 variants of multifocal pupillographic perimetry in glaucoma to test the effects of balancing luminances across the field to match the sensitivity of the pupillary field.

Methods

We tested 21 normal and 21 glaucoma subjects. Glaucoma subjects had moderate to severe visual fields in at least one eye. All subjects were examined with HFA achromatic, SWAP and Matrix 24-2 perimetry, Stratus OCT. Visual fields were classified according to their HFA mean defects: mild <=6 dB, moderate 6 to 12 dB, severe > 12 dB. Informed written consent was obtained from all subjects. Multifocal stimuli having 44 test regions/eye, extending to 30 deg eccentricity, were presented concurrently to both eyes using a prototype of the TrueField Analyser. Recording duration was 4 minutes, divided into 8 segments of 30 s. Individual 30 s segments were repeated if more than 15% of the data was lost due to blinks of fixation losses, both automatically monitored in real time. Pupil diameter was monitored under infrared illumination. Four stimulus protocols were examined which differed in terms of mean presentation intervals (MPI) of 1 or 4 s per region. One each of the 1 and 4 MPI stimuli had their luminances balanced so that responsive field regions received less light than less responsive regions. The mean test luminances were 210 cd/m2, and the background was at 10 cd/m2. Measures of field loss included the N-worst amplitudes, response delays, or pair-wise linear combinations of those.

Results

Diagnostic performance was assessed by areas under ROC curves (AUCs). For all visual field severities the best AUCs were produced by the balanced 1 s MPI stimulus protocol. For severe fields the mean of the 3 regional amplitudes that most deviated from the normative data gave an AUC of 1.0 ± 0.0 (mean ± SE), the corresponding AUCs for moderate and mild fields were 0.82 ± 0.11 and 0.764 ± 0.06.

Conclusion

In this preliminary study balancing seemed to improve diagnostic accuracy. The pupillographic method eliminates several problems associated with subjective testing as employed in conventional perimetry.

Downloads

Download the full paper here

Presentation: Tuesday, May 05, 2009, 1:45 PM – 2:00 PM , Room 315

Authors

C.F. Carle(1A), T.L. Maddess(1A), M. Kolic(1A), R.W. Essex(1B,2), A.C. James(1A). (A) ARC Centre of Excellence in Vision Science, CVS – Centre for Visual Sciences, (B) ANU Medical School, (1) Australian National University, Canberra, Australia; (2) Dept of Ophthalmology, The Canberra Hospital, Canberra, Australia.

Purpose

To investigate topographic variation in contraction amplitudes of direct and consensual pupil responses to multifocal stimuli.

Methods

Pupillary contraction amplitudes were analysed from five studies undertaken over 15 months in which 120 normal subjects (mean age 54.0±13.7 years) were tested with differing subsets of 26 stimulus protocols. All stimuli were dichoptically presented and responses of both pupils were recorded concurrently. The multifocal stimulus arrays subtended ±30° of visual field and varied in: the number of stimulus regions (24, 40, 44 or 60), mean regional presentation interval (0.25, 0.5, 1, 4 or 16 s), and pulse time-course (33 to 150 ms flickered or steady). Luminance of the test-regions was 290 cd/m2 on a 10 cd/m2 background. For each protocol, the ratios between direct and consensual responses were calculated for each region and linear regression performed.

Results

Independent of regional differences in sensitivity, direct responses within temporal hemifields were significantly larger than consensual in all stimulus protocols. Across the 26 protocols these differences ranged in magnitude between 9.2% (b = 0.38 dB, t (1496) = 2.46, p <.02) and 34.0% (b = 1.27 dB, t (1528) = 4.44, p <.00001). Whilst direct/consensual ratios differed between temporal and nasal regions they were reasonably uniform within each hemifield. This distribution did not directly relate to the pattern of regional sensitivity observed in response amplitudes which is believed to be due to variation in photoreceptor and ganglion cell density. The magnitude of temporal differences for each protocol was somewhat correlated with the mean size of pupil contractions for the particular protocols but more strongly with the mean baseline pupil diameter (r2 = 0.4268).

Conclusion

Stimulating the nasal retina produces larger direct than consensual responses in a pattern that does not correlate with that of afferent retinal sensitivity. The combination of uneven distribution of signals in efferent pupillary pathways accompanied by efferent response saturation is a possible cause.

Downloads

Not available yet – check back here after the meeting.

Presentation: Monday, May 04, 2009, 8:30 AM -10:15 AM

Authors

A. Bell, A.C. James, M. Kolic, T. Maddess. ARC Centre of Excellence in Vision Science, Australian National University, Canberra, Australia

Purpose

We sought to derive perimetric measures from the responses of pupils to novel spatial and temporal patterns of dichoptic multifocal visual stimuli; we then investigated whether the measures could distinguish 23 subjects in the early stages of type 2 diabetes from 23 normal subjects.

Methods

We used a prototype of the TrueField Analyzer to deliver a multifocal sequence of flashed stimuli to both eyes at the same time. This device uses a stereoscopic pair of LCD displays to deliver pseudorandomly modulated arrays of light stimuli to multiple regions of each retina while pupil responses are recorded with infrared cameras. The multifocal stimuli covered 44 regions per eye and induced variations in pupil diameter which were measured across 8 segments of 30 s. The method was largely immune to the effects of blinks and fixation losses. Applying receiver operator analysis, we then examined whether the pupil responses of the diabetic patients could be reliably discriminated from those of normal subjects. We examined the n-worst constriction amplitudes, time to peak, and linear combinations of those.

Results

Dichoptic multifocal pupillometry provided robust plots of pupil contraction versus post-stimulus time for each stimulus region. These region-by-region constrictions were reliable, giving median z-scores of 2 to 3. Responses of the normal and diabetic subjects were statistically different when region-by-region effects were considered (p<0.0005), but not when aggregated (p=0.07). The diagnostic performance (expressed as areas under ROC plots) for the 8 subjects who had been diagnosed with type 2 diabetes for at least 10 years was 0.89 ± 0.06 (mean ± SE), rising to 0.97 ± 0.03 when between-eye asymmetry was considered.

Conclusion

In a pilot study of 23 patients diagnosed with type 2 diabetes, dichoptic multifocal pupillography produced perimetric measures that were statistically different to those seen in 23 matched controls, especially for those who had had the disease for more than 10 years. This result, if confirmed in a wider group, suggests that the method may be clinically useful.

Downloads

Download the full paper here

Presentation: Sunday, May 03, 2009, 11:15 AM – 1:00 PM

Authors

F. Sabeti(1A), T.L. Maddess(1A), R.W. Essex(1B), A.C. James(1A)

(1) Australian National University, Canberra, Australia, (A) ARC Centre of Excellence in Vision Science, Centre for Visual Sciences, (B) College of Medicine, Biology and Environment, Department of Ophthalmology Canberra Hospital

Purpose

To investigate the sensitivity and specificity of 4 stimulus variations of multifocal pupillographic perimetry in unilateral exudative macular degeneration (MD).

Methods

Pupillary contraction amplitudes and time to peak contraction were analysed for 29 normal (mean age 70.9 ±6.0) and 20 unilateral exudative MD (mean age 78.0 ±5.3) subjects with 4 different stimulus protocols. Stimuli were presented dichoptically and pupil responses were measured concurrently. All protocols presented multifocal stimulus arrays subtending ±15° of visual field. A dart board layout having 24 or 44 independent test regions/eye with a mean presentation interval of 1 or 4 s/region and a presentation duration of 33 ms on each presentation was employed. Luminance of the stimulus regions was 250 cd/m2 and background 10 cd/m2. Test duration was 4 minutes separated into 8 segments of 30 second recording intervals. Cameras under infrared illumination monitored pupil responses. Data during blinks and fixation losses were excluded to a maximum of 15% of responses beyond which a segment was repeated.

Results

Stimuli presented in a 24 region layout with a 4 s/region presentation rate achieved the largest responses by a factor of 2.3 (b = 3.63 dB, t = 3.57, p <.00001); however this was not found to be most diagnostic, achieving an ROC area under the curve (AUC) of 83.31%. A linear discriminant model incorporating contraction amplitude and time to peak found the 44 region layout with 4 s/region presentation rate to be the most diagnostic achieving an AUC of 89.51%.

Conclusion

The clinical application of multifocal pupillography utilizing a 44 region stimulus with a slow presentation rate can produce ROC AUC of 89% in the diagnosis of unilateral exudative MD. Further investigation into the assessment of non-exudative MD through pupillography may facilitate early diagnosis and therapeutic intervention.

Downloads

Download the full paper here