Comparison of Glaucoma Detection Performance of Binocular Perimetry Screening Program Using imo Perimetry With Frequency Doubling Technology

Abstract

Purpose: We aimed to preliminarily compare the glaucoma detection accuracy of a head-mounted binocular visual perimeter “imo” screening program (ISP) with that of frequency doubling technology (FDT).

Methods: This multicenter, diagnostic accuracy study based on prospectively collected data included 76 non-glaucoma (including pre-perimetric glaucoma) eyes and 92 glaucomatous eyes from patients visiting two hospitals. Patients underwent ISP and FDT (C-20-1 screening program) on the same day. Diagnostic efficacy was evaluated using receiver operating characteristic curves and areas under the curve (AUCs). In addition, we compared the ISP and FDT testing times.

Results: AUC values for ISP versus FDT were as follows: (1) mild-stage glaucoma (mean deviation [MD] > –6 dB), 0.82 (95% confidence interval [CI], 0.75–0.88) versus 0.76 (95% CI, 0.68–0.83); moderate-stage glaucoma (–6 dB ≥ MD ≥ –12 dB), 0.98 (95% CI, 0.95–1.00) versus 0.96 (95% CI, 0.93–1.00); and advanced-stage glaucoma (–12 dB > MD), 1.00 (95% CI, 1.00–1.00) versus 0.99 (95% CI, 0.98–1.00). In addition, mild-stage glaucoma was classified into two stages (MD > –3 D) and (–3 D ≥ MD > –6 D). AUC values were 0.81 (95% CI, 0.73–0.88) versus 0.76 (95% CI, 0.68–0.84) for MD > –3 D and 0.86 (95% CI, 0.77–0.94) versus 0.73 (95% CI, 0.61–0.86) for –3 D ≥ MD > –6 D. The testing time for the ISP was significantly shorter than that of FDT for all glaucoma stages (P < 0.001).

Conclusions: The ISP demonstrates non-inferiority in detecting glaucoma and has a shorter testing time compared with FDT. These findings provide evidence for applied further studies on large-scale population-based glaucoma screening.

Translational Relevance: Our study provides a non-inferior and quicker glaucoma screening than existing tools.

Introduction

Glaucoma is a leading cause of blindness worldwide. It affects approximately 76 million persons, a number that is expected to rise to 112 million by 2040.^1–3 The estimated prevalence of glaucoma is reportedly approximately 5% in people ≥40 years of age.^4,5 However, >50% of glaucoma patients are undiagnosed, often because these individuals are asymptomatic.^6–9 Therefore, early detection through screening is believed to help prevent blindness due to glaucoma.^10–13 Although traditional approaches to glaucoma screening include history taking, tonometry, and evaluating the retina and optic nerve head using fundus photography, their efficacy to detect glaucoma remains controversial.¹⁴ Concerning glaucoma diagnosis, the presence or absence of visual field defects (VFDs) in regions consistent with observed structural damage is a crucial factor in detecting glaucoma. For this purpose, visual field testing is performed to detect VFDs, but standard automated perimetry (SAP) takes approximately 10 minutes. Frequency doubling technology (FDT) has a short testing time and has been utilized for glaucoma screening in Japan, and we reported on its effectiveness in mass screening in a previous study.^5,15 However, FDT differs from SAP in its measurement principle.¹⁶ 

The “imo” (CREWT Medical Systems, Tokyo, Japan), a portable head-mounted perimeter, was recently developed.^17,18 The imo can measure the visual field using the same measurement principle as that used with SAP while the patient keeps both eyes open, and there is no need for a dark room. The imo has reached a level of accuracy comparable to that of the Humphrey Field Analyzer (HFA).^18–22 We previously established a screening visual field testing program using the imo screening program (ISP) and successfully showed that it can detect glaucoma with accuracy comparable to that of the HFA with shorter testing times, regardless of the glaucoma stage.²² However, we have not yet compared it with conventionally used visual field screeners such as FDT perimetry. We hypothesized that the ISP may have better ability to detect glaucoma than FDT. Therefore, this study sought to evaluate the detection accuracy and testing time of the ISP in comparison with those of FDT across various stages of glaucoma. This was a hospital-based study for future application to mass screening. 

Methods

Study Design

This was a diagnostic accuracy study based on prospectively collected data that included patients without glaucoma and patients with primary open-angle glaucoma (POAG) and normal tension glaucoma (NTG) from the Jikei University School of Medicine and Atsugi City Hospital from January 2018 to September 2019.²² We included patients who visited our departments for treating or diagnosing glaucoma and who displayed signs of glaucoma, such as glaucomatous disc, high intraocular pressure (IOP), and retinal changes, including a large cup ratio, nerve fiber layer defect, and disc hemorrhage. All patients underwent a comprehensive ophthalmologic examination by a glaucoma specialist (EN, SO, KY, or TN), including assessments of best-corrected visual acuity, IOP via Goldmann applanation tonometry, slit-lamp biomicroscopy, gonioscopy, central corneal thickness via corneal pachymetry, dilated ophthalmoscopy, fundus photography, visual field via automated perimetry, and peripapillary retinal nerve fiber layer (RNFL) via optical coherence tomography (Carl Zeiss Meditec, Dublin, CA). Glaucoma diagnoses were made by glaucoma specialists (EN, SO, KY, or TN), as previously described,^4,10,23 based on the presence of glaucomatous optic neuropathy and VFDs consistent with optic changes. Glaucomatous optic neuropathy was diagnosed based on the following criteria: (1) rim-to-disc ratio at the superior portion (11–1 o'clock positions) or inferior portion (5–7 o'clock positions) was ≤0.1; (2) difference in the vertical cup-to-disc ratio between eyes was ≥0.2; or (3) an RNFL defect was found.^10,23 Visual field sensitivity was measured using the 30-2 or 24-2 Swedish Interactive Thresholding Algorithm (SITA) standard program with the HFA (Carl Zeiss Meditec). The outermost 22 test points in the 30-2 SITA test pattern were excluded. According to the Anderson–Patella criteria, a glaucomatous VFD was diagnosed based on the following criteria: (1) glaucoma hemifield test results were outside normal limits; (2) pattern deviation probability plots in the upper or lower hemifield showed a cluster of three or more non-edge contiguous points with a sensitivity of <5%, of which at least one point had a probability of <1%; or (3) a pattern standard deviation outside the 95% normal confidence interval limits was noted. Glaucoma was diagnosed when both glaucomatous optic neuropathy and VFD were present. In addition to diagnosing glaucoma, NTG and POAG were classified by IOP. NTG was defined by IOP levels <21 mm Hg on multiple occasions during the untreated period, whereas POAG was defined by IOP levels >21 mm Hg. 

The inclusion criteria of the present study were as follows: (1) SAP testing using the HFA within 3 months before ISP and FDT measurement; and (2) visual field testing reliability criteria set to fixation loss < 20%, false-positive (FP) response rate <15%, and false-negative (FN) response rate < 33%. Patients with systemic or neurological diseases that may impact the visual field, ischemic optic nerve disease, retinal disease, eye trauma, and history of eye surgery (excluding uncomplicated cataract surgery) were excluded. In addition, eyes with a decimal best-corrected visual acuity of <0.7 were excluded.²⁴ After informed consent was obtained, ISP and FDT tests were randomly performed on the same day with intervals between tests and within a standardized period of 3 months following the HFA test to ensure consistency to and mitigate time-related bias. The database used in the current study has not been made public, but it is available upon reasonable request from readers. 

imo Perimetry

The imo head-mounted perimeter has previously been described in terms of its basic structure, features, and functions.^17,18,22 The imo utilizes a full high-definition transmissive liquid crystal display to present test targets, and it employs separate high-intensity light-emitting diode backlights for the right and left eyes. Patients are examined with both of their eyes open. During the examination, patients are instructed to fixate their vision on a central target and are asked to respond by pressing a handheld button whenever a test target is detected. Stimuli are presented to both eyes randomly, making it difficult for the patient to determine which eye is being examined. 

imo Screening Program

The ISP test points have previously been described in detail.²² Briefly, 20 points were selected from the 24-2 SITA according to the frequency of occurrence of glaucomatous VFDs, and a further eight points were selected from the 10-2 SITA, making a total of 28 measurement points (see Supplementary Fig. S1). Each test point was judged as “seen” or “not seen” by stimulating it with an intensity corresponding to the 95th percentile of normal age-related sensitivity. If a reaction was observed, it was judged as “seen,” and the test point was randomly moved to another point. After the first round of examination with a single stimulus, the ISP presented a second stimulus with the same intensity to each test point where the patient missed the response to the first exposed stimulus. If the response to the second stimulus was confirmed, the ISP was judged as “seen” for the corresponding test point. If examinees did not respond to the second stimulus, the test point was judged as “not seen.” If one or more points were judged as “not seen,” the patient was assigned to the glaucoma group. 

FDT Perimetry

This study used the FDT C-20-1 screening program to evaluate 17 visual field locations.¹⁰ Participants were directed to fixate on a central black dot on the screen throughout the test and press the response button upon observing flickering black and white vertical bars on the screen. FDT perimetry tests that revealed the presence of at least one abnormal spot were defined as abnormal results. 

Statistical Analysis

To evaluate the ability of the ISP and FDT to distinguish between glaucomatous and normal eyes based on sensitivity, with 1-specificity being calculated for each cutoff point generated by the number of “seen” test points derived from the ISP or FDT, receiver operating characteristic (ROC) curves were plotted. Additionally, ROC curves were plotted to determine the testing condition for the maximum number of stimuli (MNS) and number of abnormal points (NAP) for the ISP to determine the glaucoma group. MNS refers to the maximum number of stimulus presentations at each measurement point, and NAP signifies the number of measurement points judged as abnormal (the total number of locations tested minus the number of seen points). The diagnostic capability of the ISP and FDT for the severity of VFDs examined by the HFA was determined using the area under the ROC curve (AUC).²⁵ Each point on the ROC curve represents the sensitivity and specificity plotted at a cutoff value, which is the number of undetected abnormal points among the measurement points of ISP or FDT. Moreover, Youden's index²⁶ was used to evaluate the cutoff point, and the maximum value of the index was used as a criterion for selecting the optimum cutoff point. The ISP and FDT testing times for one eye were compared using a paired t-test. P < 0.05 was considered statistically significant. Statistics were performed using JMP 15.2.0 software (SAS Institute, Cary, NC). 

Results

We included 184 eyes of 92 patients in this study. Among them, 168 eyes met the inclusion criteria and were divided into the normal (76 eyes) and glaucoma (92 eyes) groups. Based on the mean deviation (MD) value obtained from the HFA test, the glaucoma group was further classified into three grades: mild (MD > –6 dB), moderate (–6 dB ≥ MD ≥ –12 dB), and advanced (MD < –12 dB). The normal group had an MD of 0.56 ± 1.59 dB; the glaucoma group included 67 eyes in the mild stage (MD, –1.29 ± 2.13 dB), 14 in the moderate stage (MD, –8.57 ± 1.32 dB), and 11 in the advanced stage (MD, –16.56 ± 4.21 dB). Table 1 presents the characteristics of the study participants. Our previous screening program was found to be somewhat less accurate in detecting mild-stage glaucoma.²² Therefore, we first examined the most accurate conditions for the ISP in terms of the MNS and NAP using ROC curves, AUCs, and Youden's index (see Supplementary Fig. S2). In the ISP, the MNS was set to two times, and glaucoma was diagnosed if more than one abnormal point was detected. The highest Youden's index values for the glaucoma stages were 0.72 for mild stage, 0.88 for moderate stage, and 0.88 for advanced stage (see Supplementary Table S1). Therefore, we chose to compare the ISP with FDT under this condition. The detecting abilities of the ISP and FDT were evaluated using ROC curves (Fig. 1). The AUC values for the ISP and FDT were 0.82 and 0.76 in the mild stage, 0.98 and 0.96 in the moderate stage, and 1.00 and 0.99 in the advanced stage, respectively (Table 2). The sensitivities of the ISP and FDT were 0.84 and 0.66 in the mild stage, 1.00 and 1.00 in the moderate stage, and 1.00 and 1.00 in the advanced stage, respectively. In patients with mild-stage glaucoma, the sensitivity of the ISP was significantly higher than that of FDT. Specificities of the ISP and FDT were 0.88 and 0.82, respectively (Table 2). The ISP testing time was 35 to 87 seconds (normal, 44 ± 9 seconds; mild glaucoma, 53 ± 11 seconds; moderate glaucoma, 64 ± 11 seconds; and advanced stage glaucoma, 79 ± 8 seconds). The FDT testing time was 44 to 161 seconds (normal, 57 ± 13 seconds; mild stage, 81 ± 24 seconds; moderate stage, 107 ± 32 seconds; advanced stage, 140 ± 21 seconds). The ISP testing times at all glaucoma stages were significantly shorter than those of FDT (Table 3). In addition, patients with mild-stage glaucoma were classified into two stages (MD > –3 D) and (–3 D ≥ MD > –6 D), and we evaluated the ability of the ISP and FDT to detect these stages of glaucoma using ROC curves (Fig. 2). The AUC values for the ISP and FDT were 0.81 and 0.76 (MD > –3 D) and 0.86 and 0.73 (–3 D ≥ MD > –6 D), respectively (Table 4). The sensitivities of ISP and FDT were 0.83 and 0.66 (MD > –3 D) and 0.86 and 0.64 (–3 D ≥ MD > –6 D), respectively. At both stages, the sensitivity of the ISP was significantly higher than that of FDT (Table 4). 

Table 1.

 

Patient Characteristics

 

Figure 1.

 

ROC curves for ISP and FDT. Shown is a comparison of “invisible” points in the ISP and FDT using ROC curves, with the existence or absence of mild (blue), moderate (orange), and advanced (green) HFA-derived VFDs as the gold standard.

Table 2.

 

AUC, Youden Index, Detection Power, and Testing Time for ISP and FDT

 

Table 3.

 

Visual Field Testing Time of ISP and FDT

 

Figure 2.

 

ROC curves for ISP and FDT in patients with MD > –3 D or –3 D ≥ MD > –6 D glaucoma. Shown is a comparison of “invisible” points in the ISP and FDT using ROC curves in the MD > –3 D (violet) and –3 ≥ MD > –6 D (red) groups.

Table 4.

 

AUC, Youden Index, and Detection Power for ISP and FDT in Patients With Mild-Stage Glaucoma

 

Discussion

This is the first study in which we evaluated the diagnostic capability and duration of test execution of the ISP compared with that of FDT, spanning multiple stages of glaucoma. Our findings suggest that the demonstrated performance of the ISP was not inferior to that of FDT in detecting all glaucoma stages and could achieve such evaluations in a shorter time. 

We initially found that the MNS = 2 and NAP = 1 condition yielded the highest detection accuracy for early glaucoma, with decreased testing times observed with the ISP. Many population- or community-based studies have reported relatively high FP and FN response rates with visual field screening for glaucoma.^5,22,27–29 In the ISP, increasing the MNS and NAP is expected to increase FN response rates, whereas decreasing the MNS and NAP is expected to increase FP response rates. In this study, decreasing the MNS from 3 to 2 and maintaining the NAP at 1 was primarily effective in reducing FN response rates, which can be understood as an improvement in the AUC value. 

The most substantial finding of this study is that the ISP in the aforementioned condition demonstrated a high AUC, positive predictive value, and statistically significantly higher sensitivity values for mild-stage glaucoma and exhibited comparable AUC and sensitivity values to FDT for moderate- and advanced-stage glaucoma. FDT has been firmly established as a glaucoma screening tool, and its efficacy has been reported.^5,10 However, its low sensitivity has also been previously reported,^27,29,30 and our current results suggest the potential for the ISP to serve as a non-inferior screening visual field measurement program compared with FDT. 

In our study, the ISP demonstrated a higher AUC value than that for FDT, not only in mild-stage glaucoma but also in MD > –3 D or –3 D ≥ MD > –6 D glaucoma, although the difference was not statistically significant. In our study, glaucoma was diagnosed based on the HFA results. The measurement points were not located on the nasal side where they are more likely to be abnormal in patients with mild-stage glaucoma with FDT (C-20-1), whereas the ISP measurement points included the nasal side and applied the same measurement principle as that used with SAP, which might have affected the results. Larger investigations are required to clarify this aspect in the future. 

In addition, the ISP could be conducted in a shorter amount of time than FDT in all stages of mild to advanced glaucoma. FDT can perform screening tests quickly and has been highly regarded for its utility in large-scale health check-ups in Japan.^5,15 Considering the ability of the ISP to detect glaucoma in all stages and the short testing time, these findings suggest that the ISP may be a more suitable screening program than other previously reported screening tools, such as FDT. However, it should be noted that the testing times reported in this study do not include the set-up time for each device. The set-up time for the ISP is relatively short, as it does not require occlusion of one eye or mechanical manipulation when switching the eye being tested; therefore, it is expected that the overall time advantage of the ISP would not be significantly affected. Future studies may wish to quantify and compare the set-up times for each device to provide a more comprehensive assessment of the total time required for testing. 

Our study had some limitations. First, most of the patients were normal or had mild-stage glaucoma, and only a small number of patients had moderate or advanced-stage glaucoma. Our results suggest that both the ISP and FDT, under conditions considered most suitable for screening, can detect at least moderate to advanced stages of glaucoma almost without fail, consistent with previous reports on FDT.^4,5,15 ISP and FDT were compared under the best screening conditions for each. The realistic goal of screening was to detect mild-stage glaucoma as accurately as possible without missing moderate- and advanced-stage glaucoma. Therefore, the low number of patients with moderate- to advanced-stage glaucoma compared with patients with mild-stage glaucoma likely did not affect our findings. Second, this study was conducted for glaucoma screening; hence, the eyes that did not meet the SAP criteria were considered normal. Thus, the normal group might have included patients with pre-perimetric glaucoma. Third, because the study was conducted at a tertiary-level hospital, the actual glaucoma detection power in health check-ups remains unclear. This study was a hospital-based investigation intended for future mass screening applications; therefore, a population-based study with a larger sample size is needed to assess the effectiveness of the ISP in health check-ups, such as our previously published study.¹⁵ Fourth, there was a 10-year mean age difference between the normal and glaucoma groups. Although not statistically significant, the accuracy of age correction may differ between ISP and FDT, which could have influenced the results. Future studies should compare age-matched groups. Finally, in this study, we adopted the C-20-1 program for FDT, which presents stimuli at a contrast expected to be seen by 99% of normal observers. However, ISP employed 95% limits of normal. This difference might have influenced the comparison of detection accuracy. We chose to use C-20-1 because we have previously reported its usefulness for mass screening.¹⁵ In this study, which compared ISP with C-20-1, no marked difference in detection accuracy was observed between ISP and FDT. However, future studies may have to compare ISP with C-20-5 (i.e., the FDT test that uses 95% limits of normal). 

In conclusion, based on its AUC, sensitivity, specificity, positive predictive value, and testing time, the ISP demonstrates non-inferiority in detecting glaucoma compared to FDT and has the potential to contribute to improved glaucoma screening through visual field testing. 

Acknowledgments

The authors thank Yuki Komagata and Asami Itoh for their clinical evaluation of the patients. 

Supported by a grant from the Japan Society for the Promotion of Science KAKENHI (JP20K18396 to SO). 

Disclosure: E. Nishijima, None; D. Hosaka, None; S. Ogawa, None; Y. Itoh, None; T. Noro, None; S. Okude, None; K. Sano, None; K. Yoshikawa, None; M. Tatemichi, None; T. Nakano, None