Introduction
Visual display units (VDUs) are essential devices in visual electrophysiology for presenting structured visual stimuli. Typically VDUs generate patterned or multifocal stimuli for clinical visual evoked potential (VEP), pattern electoretinogram (PERG) or multifocal electroretinogram (mfERG) recordings, for which there are international standards [
1‐
3]. There are prescribed technical requirements of such VDUs to ensure that they have sufficient properties of luminance, contrast, colour, alongside temporal characteristics. These precise measurements ensure that the recorded physiological potentials are predictable and reproducible.
There is currently a widespread deficit of adequate commercially available VDUs. Many widely used stimulators are either obsolete, or those available are unsuitable for visual electrophysiology testing. For example, many centres use cathode ray tube (CRT) stimulators despite these being obsolete and parts no longer manufactured, with older models requiring frequent calibration. At the authors institution, plasma display panels (PDP) are used, but are similarly rarely produced and obsolete. Modern VDUs such as liquid crystal display (LCD) screens or organic light emitting diode (OLED) displays can be largely unsuitable for electrophysiology testing. LCD displays unfortunately succumb to a detrimental transient luminance artefact with each pattern element shift to on- or off-states, which in some circumstances can be minimised using low contrast or in-built luminance adjustments, but often are not adequate for testing and risk being non-compliant with ISCEV standards [
4‐
6]. OLED displays are potential solutions to this issue, however many suffer a detrimental input lag jitter due to resampling of the incoming trigger so risk desynchronisation of the recorded response.
Digital light processing (DLP) laser projectors were first developed for the defence industry before being widely used within digital cinema [
7]. Developments in technology now mean that these devices are commercially available for personal use and can be used within an ultra-short throw ratio so do not require the large projection distances originally needed for large field sizes. DLP laser projectors involve projection of a light source, the laser, onto a digital micromirror device (DMD). The DMD is comprised of thousands of tiny micromirrors which can be individually controlled into on- or off-states at a rapid rate [
7]. Each mirror on the DMD represents a pixel, whereby each mirror reflects the light onto a light absorber or toward a projection lens. The light is typically passed through a high speed colour wheel to achieve a vast range of chromaticity followed by optical correction for the subsequent projector screen. The resultant screen can have very high resolution, luminance, temporal refresh rate and appreciable field size, making it a candidate to replace obsolete VDUs in visual electrophysiology.
The purpose of this study was to assess DLP laser projectors for their suitability for pattern visual electrophysiology tests, both through photometric and physiological perspectives.
Discussion
This study aimed to examine DLP laser projectors as potential VDUs for routine pattern use in visual electrophysiology tests. Our findings confirm that the Viewsonic DLP laser projector tested in this study is very suitable for these purposes, providing high luminance, high contrast and fast temporal profiles required of visual stimulators. Importantly the patterns are produced with temporally identical and balanced luminance on (rise) and off (fall) timings. Furthermore, we demonstrate that physiological responses recorded from the tested device is similar to those from existing, established VDUs at our centre. The tested DLP laser projector produced responses of comparable peak-times to existing validated systems, though response amplitudes were larger from the DLP device. The confirmation that some DLP laser projectors are suitable for electrophysiology testing is particularly important at the time of writing, given the increasing difficulty in sourcing suitable reliable VDUs and decreasing availability of remaining obsolete devices.
To our knowledge, the reported literature evaluating DLP technology in this setting is very limited with only one study assessing its use for chromatic VEPs [
11]. Alternative solutions such as LCD VDUs have been well documented, but highlight their significant limitations in terms of transient luminance artefacts and input lag [
4‐
6,
12‐
14]. Whilst CRT and PDP technology is robust and suitable for use, these devices are now obsolete and modern solutions are required. Although some OLED displays show useful properties as VDUs [
13,
15], it is the authors experience that a proportion of these devices exhibit an input lag jitter, similar to that observed with the HiSense device in this study and are therefore unsuitable for use. The main DLP device assessed in this study appears to be a robust, fast and capable stimulator for visual electrophysiology testing.
Whilst we found that the assessed DLP laser projector is highly suitable for clinical testing, the individual model specification is evidently critical. We discovered a detrimental input lag and jitter in the second Hisence DLP laser projector device, which prevented any further appraisal of this device. It is likely that this jitter was caused by resampling the incoming signal within the projector system, which caused a frame shift or desynchronisation of the resultant signal output. All digital processing settings within the projector had been turned off for testing, but it is possible that some devices retain an inherent processing of incoming signals which makes them unsuitable for electrophysiology testing. The authors personal observations are that some OLED devices suffer a similar input lag jitter, but this is similarly model dependent. Of note, the Hisense device was advertised as an ‘entertainment’-based DLP projection system, whereas the Viewsonic device was advertised as an office/work-based projector. It is possible that entertainment-based DLP devices may process video signals to enhance performance, which evidently may preclude their use for visual electrophysiology. Based on these observations, we strongly recommend that anyone considering use of DLP projectors should assess individual model feasibility before clinical implementation.
We found response times to be very fast for the DLP device, with rise and fall times of 0.5–1 ms. This is comparable or faster than that observed for CRT and LCD stimulators, respectively [
16]. The manufacturer data for the DMD chips (Texas Instruments) suggest that response times may far exceed that recorded in this study (movement speeds up to 10,000 Hz), suggesting that these response times may reflect a simplification or limitation of the graphical output from the Espion E
3 system [
7]. During set up, a 60 Hz calibration file was generated, this could theoretically be increased to 100 Hz with this system but comes at the compromise of resolution due to the pixel clock rate. Overall, the DLP’s luminance and contrast ratios were widely sufficient for clinical testing and far exceeded the minimum standards required for PERG and PVEP testing [
2,
3]. Importantly, we found a significant input lag for the DLP device, taking 50 ms from trigger to stimulus change. Whilst significant, this was very stable and adjustments can be easily made for this input lag by adjusting time zero to coincide with the onset or half-point of reversal change, as indicated in clinical standards [
2,
3].
We found that warm-up time (i.e. time from a ‘cold start’ turn on to being fully operational) was immediate for the PDP device, but took around two minutes for the Viewsonic DLP device. This time is certainly an acceptable level for clinical circumstances, particularly since LCD and CRT VDUs can take or exceed a 60 min warm-up [
12,
17], after this LCDs are also sensitive to changes in ambient temperature [
12]. Furthermore, we found no delay in response time over this period, suggesting the device is fully operational within two minutes warm-up. This is in further contrast to LCD devices which have slow response time warm-up periods, with some devices taking up to one hour until reaching optimal response time [
18]. Certainly, based on the mechanical properties of stimulus presentation which is based on the DMD chip speed, we would not have expected any delay in response time over this period.
Luminance properties of the device were very advantageous, capable of maintaining high contrast at mean luminance around 300 cd/m
2. We found mild luminance variance across the projection screen, but this was within 91.8% of maximum so within acceptable recording standards [
2,
3]. Nevertheless, the luminance distribution appeared to follow a pattern whereby those closest to the DLP laser projector had higher luminance than those furthest away. This may be a feature of the ultra-short throw ratio used of this projector, creating highly oblique angles to the projection screen. It is suspected that DLP laser projectors with longer throw ratios may show less of this spatial variance in luminance.
We found that the spectral properties of the stimulus showed a large peak in the ‘blue’ wavelength with broader energy at longer wavelengths (Fig.
1D). There is no specific reference to the spectral properties of stimuli for clinical PERGs or PVEPs [
2,
3]. Existing visual stimulators have widely varying spectral properties, so the spectral profile observed for the DLP device is likely insignificant in this context, as the correlated colour temperature was very close to 6500K (6495K). Furthermore, the spectral properties of DLP laser projectors may vary per device depending on the composition of the colour wheel. Perhaps most curious of our observations was the perceived ‘rainbow effect’, which occurred with fast eye movements (supplementary Fig. 2). This is a type of stroboscopic artefact giving a spectral inhomogeneity of the pattern stimulus due to the colour wheel used, and is particularly marked for white checks. It is a result of the colour being rendered sequentially through the colour wheel causing temporal inhomogeneities, typically at 2–4 times the framerate. From our observations, this was only apparent with very rapid eye movements and perception varied according to observer. Nevertheless, the influence of this finding on patients with unstable fixation (i.e. children) or involuntary eye movements (i.e. nystagmus) is uncertain.
We suspect that the rainbow effect would have little significant influence on the PERG or PVEP, as the colour wheel frequency was measured here to be 120 Hz (twice the 60 Hz framerate) which is far faster than the time-locked presented visual stimuli, temporal resolution of visual contrast systems [
19] and is around the temporal resolution limit of cone photocurrents [
20]. Furthermore these artefacts are not constant inhomogeneities, instead are rapidly changing temporally, so any resultant physiological differences would likely average to noise levels. Nevertheless, this may be dependent on device used, as some newer or more expensive DLP devices may use × 4 or higher colour wheel frequencies relative to framerate, which would minimise this effect. Early DLP projection devices used a 1 × colour wheel making the rainbow effect markedly evident for most observers, but are now rarely used. Furthermore, technical developments in this area are continuing and are likely to further minimise this effect, such as development of the 3-chip DLP (comprising of three DMD chips for red, green and blue lasers) or three laser DLPs (removing the need for the colour wheel). Considering these points, this effect is considered to have a negligible influence on the PERG or PVEP.
In the physiological experiments, we did not find any significant differences in peak-time between P50 and P100 components between devices, suggesting the tested DLP device performed similarly to our existing systems in this respect. The largest discrepancy in our results were larger responses for PERGs and PVEPs and earlier peak-times for small check width PVEPs from the DLP laser projector than the existing PDP VDU, despite spatial and photometric matching. This is an interesting finding, which suggests that the DLP device may perform better than existing PDP devices. The explanation for this difference in view of photometric and spatial matching may, we suspect, result from originate from two possible mechanisms. Firstly response times from the DLP device are faster than PDP VDUs, or secondly the improved resolution of DLP laser projection systems which affects different or enhanced physiological properties.
Response times observed from the DLP stimulator assessed in this study were fast, in the order of 0.5–1 ms. It is possible, that this faster response time of DLP relative to existing PDP VDUs may allow better temporal synchronisation of the physiological substrate of interest. An abrupt response change may theoretically cause more simultaneous activation of retinal and neural cells which may therefore improve response amplitude as observed in this study. It has been demonstrated that response times do not significantly alter the PVEP below 10 ms, although likely alter between 8 and 16 ms to affect the PVEP which may explain our findings [
21]. This is supported by the upper limit of frequency–response curves of the pattern VEP being 15-20 Hz [
22]. Therefore, faster rise times are a likely cause for the larger PERG and PVEP amplitudes observed in our study, which is likely advantageous for clinical testing but highlights a need for locally derived reference data for implementation of these new devices.
It is fairly well known that whilst CRT and PDP systems are suitable VDUs, they have relatively poor spatial resolution due to pixel size and therefore edge contrast can be low. Reducing edge contrast can have a direct effect on the PVEP amplitude, as the pattern stimulus waveform becomes more sinusoidal similar to a change in modulation transfer function [
23,
24]. Therefore, the relative higher resolution and sharpness of a DLP stimulus may therefore improve the respective retinal contrast, which would be particularly evident to small check widths as observed in PERG and PVEP data of our study. We suspect these changes may, at least in part, be responsible for the differences in amplitude between devices observed.
A very beneficial feature of DLP laser projectors is that they are capable of extra-large field stimulation within ultra-short throw ratios, meaning very large field sizes can be achieved without the need for large laboratory space. We calculate that the Viewsonic projector as used in this study, at a working distance of 125 cm, could present stimuli in visual fields of up to 90 degrees. Whilst large field sizes are particularly useful for paediatric practice, there comes a point whereby larger field size becomes detrimental to the PVEP P100 component. With increasingly large field sizes the paramacular PVEP components become more pronounced, and if large enough they can degrade the macular driven P100 component of interest [
10]. There may be some applications for large field sizes which are beneficial to avoid short viewing distances, such as for the mfERG or mfVEP, but for routine clinical PERG and PVEP testing, it seems that exceeding a 30 degree field may not yield any significant benefit, hence our aim to spatially match existing PDP system dimensions.
Lastly, whilst our study assessed the DLP device in front-projection mode, it is likely that in clinical circumstances a back-projection would be far more beneficial to avoid any potential interference of the projection beam by patients, staff or equipment.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.