Effect of 50-Hz Filters on Pattern Electroretinogram
Article information
Abstract
Purpose
Pattern electroretinogram (PERG) is used to evaluate the function of retinal ganglion cells. However, the amplitude of PERG is quite small, making the examination challenging to perform. Waveform noise may be minimized by applying various filters. We aimed to investigate the effect of 50-Hz filters on PERG test results.
Methods
This is the retrospective observational study. PERG tests were performed using the RETI-scan system according to the International Society for Clinical Electrophysiology of Vision guidelines. Three types of 50-Hz filters (soft, middle, and hard) were applied. The differences in parameters (N35 peak time, P50 peak time, N95 peak time, P50 amplitude, N95 amplitude, and N95 to P50 ratio) were analyzed. Based on the provided normal range, the changes from normal to abnormal range or vice versa were investigated.
Results
A total of 24 waveforms were analyzed. After filtering, the P50 and N95 amplitudes showed a significant reduction of 8% to 15% (P50 amplitude: 5.1 ± 2.7 μV without filter, 4.6 ± 2.3 μV with 50-Hz soft filter, 4.3 ± 2.1 μV with 50-Hz middle filter, 4.3 ± 2.1 μV with 50-Hz hard filter; N95 amplitude: 7.2 ± 4.2 μV without filter, 6.6 ± 3.8 μV with 50-Hz soft filter, 6.3 ± 3.6 μV with 50-Hz middle filter, 6.1 ± 3.6 μV with 50-Hz hard filter). This pattern was more prominent in normal subjects. All latencies except the N35 peak time exhibited no differences between the tests. The N95 to P50 ratio decreased after 50-Hz middle and hard filtering. Considering the normative data, switching between normal and abnormal results was rare.
Conclusions
Although peak time was not significantly affected, amplitude was significantly reduced after using 50-Hz filters. Thus, 50-Hz filters can smoothen the waveform. Nevertheless, caution must be exercised while taking readings.
Pattern electroretinogram (PERG) is an electrophysiologic test for evaluation of central retinal function, especially that of the retinal ganglion cells. Peak time and amplitude of P50 and N95, the N95 to P50 ratio, is an important index for interpretation of functional status [1,2]. Electrophysiologic tests are sensitive and often do not yield the desired waveform owing to noise. The amplitude of PERG, which is relatively small (2–3 μV), makes this technique more challenging to perform than other such tests [3]. In cases where the waveform is not typical, the investigator must manually determine the peak of wave, but in cases where there is a lot of noise, the measurement is very subjective, which can lead to errors.
There are various methods that can be considered to reduce noise and make it easier to analyze and evaluate waveforms, including applying filters to refine the waveforms into ones that are easier to analyze. However, filters used in electrophysiology tests may affect the results. Bandpass filters used for electroretinogram tests significantly suppress the amplitude of b wave and oscillatory potentials [4,5]. Moreover, powerline interference occasionally has critical effects on biopotential experiments [6], and notch filtering is often used to reduce 50- or 60-Hz powerline noise [7]. A notch filter is the inverse of a bandpass filter and attenuates only a narrow band of frequencies. Although 50- and 60-Hz notch filters affect multifocal electroretinogram test results, the clinical effect is not significant [8]. To date, no study has examined the effect of 50-Hz filters on the PERG test. Therefore, in this study, we evaluated these effects.
Materials and Methods
Ethics statement
This study was approved by the Institutional Review Board of Chung-Ang University Gwangmyeong Hospital (No. 2303-066-020). Informed consent was waived due to the retrospective nature of the study. All procedures were conducted in accordance with the 1964 Declaration of Helsinki and its later amendments.
Study design and setting
This retrospective analysis evaluated 24 PERG waveforms from medical charts collected between March 3 and December 31, 2022.
The PERG tests were performed according to the International Society for Clinical Electrophysiology of Vision (ISCEV) guidelines [9] using the RETI-port/scan 21 system (Roland Consult). Transient PERG was obtained using an LCD (liquid-crystal display) monitor. The stimulus for the transient PERG was checkerboard pattern (check size, 0.8), luminance 220 cd/m2 and more than 90% of contrast. RETI-port/scan 21 system supplied various digital filters. To evaluate filter effects on the pattern electroretinogram, 50-Hz soft, middle, and hard filters were applied. The amplitudes of N35, P50, and N95; the peak times of P50 and N95; and the N95 to P50 ratio were evaluated, and the amplitude, peak time, and ratio before and after filtering were compared (Fig. 1A–1D). The results were analyzed by applying each filter after obtaining the waveform.
Effect of 50-Hz filter on pattern electroretinogram. (A) Without filter. (B) With 50-Hz soft filter. (C) With 50-Hz middle filter. (D) With 50-Hz hard filter.
Statistical analysis
The data were analyzed using IBM SPSS ver. 26.0 (IBM Corp). The repeated measures analysis of variance was used to evaluate changes in the amplitude, peak time, and N95 to P50 ratio.
Results
A total of 14 patients (8 men and 6 women) were included in this study. Among them, four had Leber hereditary optic neuropathy, four with optic atrophy, and six with normal optic nerve function. The mean age was 33.2 years (range, 17–44 years).
A total of 24 waveforms were analyzed. The P50 and N95 amplitudes showed a significant reduction of 8% to 15% after filtering (P50 amplitude: 5.1 ± 2.7 μV without filter, 4.6 ± 2.3 μV with 50-Hz soft filter, 4.3 ± 2.1 μV with 50-Hz middle filter, 4.3 ± 2.1 μV with 50-Hz hard filter; N95 amplitude: 7.2 ± 4.2 μV without filter, 6.6 ± 3.8 μV with 50-Hz soft filter, 6.3 ± 3.6 μV with 50-Hz middle filter, 6.1 ± 3.6 μV with 50-Hz hard filter).
The amplitude showed a pattern of decreasing significantly as the strength of the filter application increased (Fig. 2A, 2B). All latencies, except the N35 peak time demonstrated no differences between the tests (N35 peak time: 26.10 milliseconds without filter, 25.84 milliseconds with 50-Hz soft filter, 25.14 milliseconds with 50-Hz middle filter, and 24.98 milliseconds with 50-Hz hard filter; p < 0.001). Analyzing the difference before and after the filter by dividing it into the normal group and the optic neuropathy group, amplitude showed a greater decrease in the normal group than in the optic neuropathies group (Fig. 3A–3C). In the optic neuropathy group, peak time was found to be somewhat prolonged in all groups except for the hard filter (Fig. 3D–3F).
Changes in (A) P50 amplitude and (B) N95 amplitude. Error bars represent 95% confidence intervals. Application of the 50-Hz filter significantly decreased P50 and N95 amplitudes.
Differences before and after filtering in normal and optic neuropathy group. Amplitude decrease with (A) 50-Hz soft filter, (B) 50-Hz middle filter, and (C) 50-Hz hard filter. Prolonged peak time with (D) 50-Hz soft filter, (E) 50-Hz middle filter, and (F) 50-Hz hard filter.
The N95 to P50 ratios while applying without filter, with 50-Hz soft, middle, and hard filters were 1.41 ± 0.35, 1.42 ± 0.34, 1.01 ± 0.05, and 0.76 ± 0.21, respectively. All differences in values obtained using the 50-Hz filter were statistically significant (p < 0.001), except for those using no filter and the 50-Hz soft filter (p = 0.897).
Considering normative data, switching between normal and abnormal results occurred rarely (P50 peak time changed in three cases [abnormal range to normal range in two cases when the 50-Hz hard filter was used]; N95 peak time changed in one case when the 50-Hz soft filter was used [normal to abnormal range] and one case when the 50-Hz hard filter was used [abnormal to normal range]).
Discussion
This study demonstrated that 50-Hz filters can affect PERG results, and considering the changes in values at different 50-Hz filter levels, the results must be carefully interpreted.
The 50-Hz soft, middle, and hard filters are selectively applied according to the height of the waveform around 50 to 60 Hz (depending on the size of the noise) because resistor-capacitor equipment is used around 50 to 60 Hz. In other words, 50-Hz soft filter is applied when the noise is small, middle filter is applied when the noise is normal, and hard filter is applied when the noise is large. However, filters can affect outcomes to varying degrees. The 50-Hz filters did not significantly affect peak time; however, the amplitude was reduced at each filtering level. The P50 and N95 amplitudes were reduced by approximately 15% after application of the 50-Hz hard filter. This pattern was more prominent in normal subjects. This is because the amplitude is larger in normal subjects than in optic neuropathy, so the rate of reduction may be greater. A previous study showed that 60-Hz notch filters decrease b-wave amplitude by 30% in the full field electroretinogram [5]. For this reason, the ISCEV standards prohibit the use of notch filters.
We found that the 50-Hz middle and hard filters caused greater reduction in the N95 to P50 ratio than did the 50-Hz soft filter, suggesting that the reduction in N95 amplitude is relatively more when the 50-Hz middle and hard filters are applied than when the 50-Hz soft filter is applied. When the filter was applied, the decrease in amplitude was more evident in the normal group. This is thought to be because the amplitude in the normal group is relatively larger than in the patient group, resulting in a greater decrease.
The normal range of each component was provided with the RETI-port/scan 21 system (P50 peak time, 46.0–61.0 milliseconds; N95 peak time, 81.0–108.0 milliseconds; P50 amplitude, 1.6–7.6 μV; N95 amplitude, 1.6–12.5 μV). The application of filters can induce the results to change from normal to outside of the normal range or from abnormal to normal range. Although the switching is rare in the study, these changes can affect interpretation; hence, comparison and analysis before and after filter application are necessary.
Filter types and parameters should be considered because these factors can modify the results [7,10]. Specifying whether 50-Hz filters are applied in each study is important for consistent investigation. However, a facility to check whether 50-Hz filters are applied to the report form in the RETI-port/scan system is not available. Hence, an update addressing this issue is required. In addition, since the results may be judged as false negative or false positive in diagnosis depending on the results, the results must be derived by comprehensively judging the results before and after applying the filter. Above all, when determining the progression of a disease, it is necessary to examine and analyze under the same conditions, so it is necessary for medical staff with a high level of understanding to conduct and interpret the examination.
There are some limitations in the present study. First, the number of subjects is not large. However, despite the number of patients, there is a significant difference that can be reflected in clinical practice. Secondly, several patient groups are mixed, which could affect the result. Finally, the results can be only applied to a specific electroretinogram machine. Therefore, further research on various diseases is needed in the future.
In conclusion, while 50-Hz filters can smoothen the waveform, the results must be interpreted carefully because they affect the wave amplitude and the N95 to P50 ratio. Therefore, a standard application procedure, clear statement of whether filters have been applied, and careful interpretation of results are necessary.
Notes
Conflicts of Interest
None.
Acknowledgements
None.
Funding
None.