Abstract
At least 7 million people in the United States have some form of visual impairment (i.e., blindness or low vision); among them, an estimated 1.62 million are aged under 40 years (Flaxman et al., 2021; WHO, 2019). Low vision can be defined functionally as chronic uncorrectable vision loss that impacts activities of daily living (Rubin et al., 2001). Prior studies have reported that individuals with low vision have a decreased health-related quality of life associated with difficulty performing daily tasks, social isolation, and emotional well-being (Lorenzini et al., 2019; Majerova, 2017; Stevelink et al., 2015).
Head-mounted display (HMD) and image processing systems represent a major innovation in vision-enhancing assistive technologies for individuals with low vision (Culham et al., 2004; Wolffsohn & Peterson, 2003). This technology has been available for more than 25 years, and it was initially created and developed for military use (Massof et al., 1994). Today, HMDs are extremely sophisticated electronic devices, but they include three main components: (1) a scene camera, (2) microelectronics to process and control the video, and (3) micro-displays that are placed in front of the eyes. The basis of operation is that the camera streams video, which is processed and redisplayed to the user.
HMDs have undergone significant transformations in recent years and are now viable consumer electronics (Harrison et al., 2010; Lorenzini et al., 2019). Current commercially available HMDs have a wide range of display resolutions from 640 × 480 to 1,210 × 920 (Deemer et al., 2018). Advanced color micro-display and better contrast enhancement technology can perform contrast-stretching, edge enhancement, and minimize color distortions (Moshtael et al., 2015). Recent advancements in virtual and augmented reality, along with progress in accuracy and precision of microelectronic technology, have not only improved the estimate of vital linear and angular motion to compensate for image motion magnification but also have led to the introduction of superior HMD devices (Angelopoulos et al., 2019; Deemer et al., 2018). Artifactual magnification, which is another important factor in low vision enhancement systems, has also been addressed by various commercially available HMD devices (Deemer et al., 2019). The use of see-through displays allows the user to maintain use of their natural vision while incoming information is overlaid using partially transparent displays.
Modern HMD devices may assist individuals with low vision who benefit from features like magnification and contrast enhancement, but also those with peripheral field loss who may benefit from field expansion (Ehrlich et al., 2017; Lorenzini et al., 2019; Trese et al., 2016). HMDs have therefore been implemented in various forms to improve the user's independence, functionality, and quality of life (Htike et al., 2020; Mojon-Azzi et al., 2008). Although HMDs have the potential to assist those with low vision, the importance of various HMD features must be further investigated to address the specific needs of individuals with low vision and to improve functionality for a more diverse range of users with low vision. Although commercially available HMDs exist for individuals with low vision, they are not well utilized on a regular basis (Gallos et al., 2018). Prior studies have found that factors, such as lack of headache and device-related quality of health, were associated with less device abandonment (Lorenzini et al., 2019). Bridging this gap between features and users’ experiences may allow for HMD design considerations to best match the daily needs and expectations of people with low vision. There is a notable gap in understanding how the range of functional impairments can best be addressed with these devices. Formulating a better understanding of the HMD features that assist individuals with low vision and allow for integration into everyday life is therefore critical. The purpose of this research was to link objective measures of vision-related quality of life (VRQOL) and visual function with experiential information among participants with mild to severe visual impairment using a mixed-methods approach (Jeganathan et al., 2019).
Methods
Design
We employed a convergent mixed-methods design (Fetters, 2020) to utilize both objective measures of visual loss and quality of life with simultaneously collected experiential qualitative data. The hallmark of a convergent mixed-methods design is the collection and analysis of both types of data: quantitative data (e.g., surveys) and qualitative data (e.g., interviews), at roughly the same time (Creswell & Clark, 2017). Qualitative research is well suited for understanding questions about the manner in which individuals perceive and use technology (Zolyomi et al., 2017). Through interviews with potential users, the experiences of adults with visual impairment can be explored, and their first impressions of HMDs can be coupled with quantitative clinical and survey data to augment an understanding of potential user preferences. This study was approved by the University of Michigan Institutional Review Board (IRB) and Western Michigan University ceded to the University of Michigan’s IRB. The research adhered to the tenets of the Declaration of Helsinki.
Setting, Study Population, and Recruitment
The W.K. Kellogg Eye Center (University of Michigan, Ann Arbor, MI) and Western Michigan University (Kalamazoo, MI) served as study settings. The study population focus was on individuals with varying degrees of low visual acuity and ocular conditions. A purposeful sampling strategy identified both participants with primary peripheral vision loss and those early in the progression of their visual impairment who might begin using HMDs. Inclusion criteria included self-reported visual impairment and a diagnosis of age-related macular degeneration (AMD), diabetic retinopathy, glaucoma, or retinitis pigmentosa that was confirmed by a review of medical records (when available). These chronic eye diseases were selected as representative conditions seen commonly in a low vision practice (Owsley, 2009). Participants were excluded if they had been diagnosed with cognitive impairment, needed an interpreter, or had a physical disability that precluded mobility. Enrolled participants provided written informed consent prior to the commencement of the study.
Study Intervention
The participants compared three commercially available HMD platforms that were representative samples of available technology: the eSight Model 3, NuEyes Pro, and Epson Moverio BT-200 (Figures 1–3). All HMDs had front-facing cameras to capture video and micro-displays positioned in front of the eye with microelectronics for filtering the video prior to display. The NuEyes Pro were goggles with a wireless controller, while the eSight and Epson both had a wire connecting goggles to a controller. The eSight and NuEyes were specifically designed for people with low vision, while the Epson was a general HMD. A study team member taught participants to use the three different types of HMD; they were asked to use each HMD to read text up close (approximately 14 inches away) and to view images across the room (approximately 20 feet away). Each participant was given the same training, which ranged from 30 to 60 min depending on their technical proficiency, as determined by their ability to complete specific tasks (e.g., magnification).

eSight Model 3 (eSight Corp., Toronto, Canada).

NuEyes Pro (NuEyes, Newport Beach, California).

Epson Moverio BT-200 (Seiko Epson Corporation, Nagano, Japan).
Quantitative Data Collection and Analysis
Visual acuity and visual field data were confirmed with data abstracted from the most recent eye examination in the medical record (when available). During the study visit, participants completed the Impact of Vision Impairment (IVI) questionnaire. The 28-item, validated questionnaire provides 3–4 response categories for each item that range from “not at all” to “can't because of eyesight” (Goldstein et al., 2018; Lamoureux et al., 2006). The scores for the three IVI domains (reading, mobility, and emotional well-being) were then transformed using published Rasch item calibrations (Goldstein et al., 2018). This approach placed IVI domain scores on a continuous scale, which permitted parametric analyses.
Qualitative Data Collection and Analysis
A study team member conducted semistructured interviews formatted to focus on one HMD at a time, while participants answered a series of questions regarding each HMD. Data collection continued until thematic saturation was achieved (Fetters, 2020). The interviews were audio-recorded, transcribed verbatim, and coded using an inductive thematic approach where team members created a coding scheme based on emerging themes for a total of 40 codes. Transcripts were coded independently by two coders (73.5% average inter-coder agreement). For example, the code “usability” was used when participants discussed how easy or difficult it was to use different aspects of the HMD, as well as how easy or difficult it would be to use the device in everyday life. MAXQDA 2018 (VERBI Software, Berlin, Germany) was used to conduct the coding and subsequent analysis.
Mixed Data Analysis
The mixed-methods data analysis of the quantitative survey data and qualitative data interview data was guided by a seven-step approach and joint displays were created to analyze the mixed data findings (Fetters, 2020; Guetterman et al., 2015). In the joint display, we assessed HMD preference with reference to IVI well-being scores (rather than reading or mobility) due to the wide range of scores in this IVI domain. Using Rasch-adjusted scores for the well-being domain of the IVI, we separated participants into low, medium, and high well-being tertiles. Qualitative responses from participants in each IVI well-being tertile were reviewed and representative responses were extracted to populate the joint display table. Iterative versions of joint displays were developed to create the final joint displays to represent the mixed findings and optimize the presentation for the reader (final versions of joint displays are presented). Meta-inferences, namely, overall interpretations, were derived through consideration of the meaning of the qualitative and quantitative data together.
Results
The study sample included 21 participants (eight had retinitis pigmentosa, five had diabetic retinopathy, four had AMD, and four had glaucoma) with a mean age of 58.2 years (range: 30–83 years), of whom 12 were male (57.1%). Participants had a median Snellen acuity of 20/40 (interquartile range: 20/20–hand movement). We defined a visual acuity of 20/30 as “blind” in a participant with retinitis pigmentosa with a central scotoma. Among those with retinitis pigmentosa (n = 8) who had undergone Goldman visual field testing, the mean maximum horizontal extent of the III4e visual field was 11.5 degrees (range: 5–37). Among participants with glaucoma (n = 4) who underwent Humphrey visual field testing, the average mean deviation was −18.35 decibels (range: −27.77 to −9.72). Notably, only three participants had prior experience with HMDs.
Figure 4 shows the participants’ age and HMD preference. Those with AMD tended to be older than those with glaucoma, retinitis pigmentosa, or diabetic retinopathy. Among the three HMDs, an equal number of participants (n = 9) had an overall preference for eSight and NuEyes versus Epson (n = 3). Participants with retinitis pigmentosa tended to prefer the NuEyes (75%), while those with AMD preferred the eSight (75%). The qualitative findings as a function of HMD preferences are summarized in Figure 5. Participants more often cited the usability of the HMD as an important factor than they did magnification, image clarity, or even visual improvement.

Participants’ Ages and Overall HMD Preferences Stratified by Diagnosis.
Tables 1–3 provide joint displays that represent the quantitative Rasch-adjusted survey data from the IVI well-being domain and qualitative interview findings. Participants with lower IVI well-being scores (with central or mixed vision loss) tended to prefer the eSight for the clarity, usability, and vision improvement the device provided (Table 1). Those with central and mixed vision loss appreciated being able to zoom in so as to view items at a distance, as well as having visible controls that made the device easy to adjust to their needs. For instance, one 71-year-old female participant with AMD noted that she preferred eSight because it was easy to see the controls. Only a single participant with a low IVI well-being score preferred NuEyes, and this was an individual with severe peripheral field loss due to retinitis pigmentosa (Table 1).
Reasons for Preferences Among Participants With Lower IVI Well-Being Scores.
Note. IVI = Impact of Vision Impairment questionnaire; Dx = diagnosis; BCVA = best-corrected visual acuity.
Reasons for Preference Among Participants With Moderate IVI Well-Being Scores.
Note. IVI = Impact of Vision Impairment questionnaire; Dx = diagnosis; BCVA = best-corrected visual acuity.
Reasons for Preference Among Participants With Higher IVI Well-Being Scores.
Note. IVI = Impact of Vision Impairment questionnaire; Dx = diagnosis; BCVA = best-corrected visual acuity.
Those in the middle tertile of self-reported well-being (with peripheral or mixed vision loss) primarily preferred NuEyes for its appearance, wireless design, and magnification enhancement (Table 2). These participants favored the sleek style, comfort, and clarity of vision of the NuEyes. The magnification of the device was “enough” for their level of comfort. However, neither of the two participants with AMD in this IVI well-being tertile preferred NuEyes.
Additionally, participants who scored highest in well-being (who most often had peripheral vision loss) were primarily concerned with usability, with no clear preference for one HMD over another (Table 3). Individuals with low IVI well-being scores seemed to have a greater need for an HMD device that helps in vision improvement, while individuals with moderate IVI well-being scores focused more on comfort and size. The participants with the highest IVI scores indicated usability as the most important factor when comparing the three devices.
Discussion
These findings suggest that device features and functions vary in clinically meaningful ways based on vision parameters. Practitioners should be aware of how patient and device variations influence preferences when they assist patient's choosing a wearable system, and for optimizing their training. We identified key human factor considerations associated with the use of HMD based on the application of a mixed-methods study design, consisting of chart reviews, quantitative surveys, and qualitative interviews. We found that both vision-related well-being and type of vision loss (central, peripheral, or mixed) influenced preference for specific HMD features. As HMD technology continues to evolve, our findings may inform future HMD design and modifications to meet the needs of those with various visual impairments.
Previous studies have shown that eye characteristics, such as position and size of the pupil, field of view, contrast and spectral sensitivity, resolving power of the fovea, depth of vision, and color perceptions, determine the features and effectiveness of a specific HMD to assist individuals with low vision (Ehrlich et al., 2017; Lorenzini et al., 2019). Lorenzini et al. (2019) found that those who did not experience a headache with the use of eSight were more likely to be sustained users. Furthermore, higher scores on the Psychological Impact of Assistive Devices Scale (PIADS) and the Quebec User Evaluation of Satisfaction with Assistive Technology (QUEST) scale predicted continued use of eSight. In that study, demographics, ocular or general health, and low vision rehabilitation experience were not predictive of sustained HMD use.
In the current study, preferences for specific HMD features and devices were related to the type of vision loss and VRQOL. Older participants with AMD and central visual loss generally preferred eSight. Ease of use and clarity were the primary reasons cited for this preference. eSight's manufacturer states that the device allows the user to adjust and optimize color, contrast, focus, brightness, and magnification, and this may allow the user to see finer details with less eye strain (eSight—Electronic Eyewear for the Visually Impaired; Wittich et al., 2018). Loss of central and peripheral vision affecting hand coordination has been previously described and may affect grasp, reach, place times, and increased path lengths when performing tasks (Baker et al., 2017). Moreover, eSight users had significantly improved visual ability, including visual–motor coordination (Wittich et al., 2018). Hence, the preference for eSight may be related to less eye strain and better usability in older participants with central vision loss compared to the NuEyes and Epson. Similar findings (enhanced peripheral vision and improved visual motor coordination) were replicated in participants who had glaucoma when they used Novel Virtual Reality Digital Spectacles (Sayed, Abdel-Mottaleb, et al., 2020; Sayed, Kashem, et al., 2020).
The IVI well-being items ask about vision loss causing embarrassment, frustration, isolation, sadness, worry about the future, and feeling such as being a burden, among others. Participants with lower scores on the IVI well-being domain in our study preferred eSight, citing clarity, ease of use, and vision improvement as the primary reasons. Conversely, those with moderate IVI well-being scores tended to prefer the NuEyes due to its appearance, wireless design, and magnification properties. Finally, those with the highest IVI well-being domain were most likely to have peripheral vision loss and no clear preference for a specific HMD, though they did have a clear preference for a device with good usability features such as controls and screen.
Study participants cited a desire for HMD that looked “normal” and would not draw attention in public. NuEyes scored better in this regard, whereas the eSight was rated lower on appearance because it did not look like a regular pair of glasses. Individuals who are more impacted by their vision loss might be more willing to accept a device that is heavy, bulky, or has a less desirable appearance. On the other hand, those who feel their vision loss has had less of an effect on their quality of life may be more likely to consider appearance and style (bulkiness was a reason that many users cited as reasons they did not like the eSight), and instead prefer a device that is lightweight and aesthetically pleasing. It is important to note that the hardware platform for NuEyes has changed since this study began, and the preference for NuEyes was based on the appearance of many participants in this study.
In prior studies, users’ likelihood of embracing or rejecting technology depended on the perceived usefulness and ease of using such technology according to the technology acceptance model (Djamasbi et al., 2006; Zolyomi et al., 2017). Moreover, overall technological acceptance and the intention to use technology diminish with increasing age (Hauk et al., 2018; Vaportzis et al., 2017). In our study, especially among older individuals with low self-reported well-being scores, perceived ease of use or usability and perceived usefulness may have further influenced their HMD preference decisions. However, future generations of people with low vision may be more accepting of technology due to their lifelong use of mobile devices. We should therefore not assume that the trends we found in this group of participants extrapolate to future potential users with low vision.
Prior studies investigated the role of HMDs in augmenting performance on a variety of tasks among individuals with low vision (Culham et al., 2004; Lorenzini et al., 2019; Wittich et al., 2018). Wittich et al. (2018) noted that the greatest improvement was seen in eSight users with reading, followed by tasks attributed to visual stimuli (e.g., contrast sensitivity) and visual motor activities (e.g., face perception task). That study reported no significant change related to mobility. However, during qualitative data collection, participants noted difficulties in certain mobility scenarios when using HMDs. In our study, no participants thought that currently available HMDs would be useful for orientation and mobility tasks, and many cited safety concerns. Prior studies using HMDs have shown that navigation is a learned skill that improves over time (Angin & Bhargava, 2011; Hicks et al., 2013). Hicks et al. (2013) found that as participants used HMDs regularly, their navigation skills improved, and they were able to complete a walking path significantly more efficiently. With greater experience, participants with low vision using HMDs were also better able to see and react to items in the peripheral and central vision areas as efficiently as normally sighted participants who used the devices. Sensory conflict in first-time HMD users, which results in poorer first experiences and, in some cases, motion sickness, could be another reason as to why all the participants said they would not consider using HMDs for navigation. Finally, the failure of HMDs to achieve quality spatial fidelity is another reason why some individuals with visual impairments may be reticent to consider these devices for navigation (Kinateder et al., 2018; Srivastava et al., 2019). People with visual impairments may have decreased spatial perception; however, currently available HMDs have not adequately addressed this challenge (Kunz et al., 2018; Marston et al., 2006). Therefore, novel HMDs may be needed to address issues of motion sickness, spatial perception, and be more responsive to the orientation and mobility needs of low vision users.
The results of this study may better inform the future design of HMD-based assistive technology. Since this cohort of participants expressed widely varying opinions on what is important, we can expect the general population of potential users to also have diverse needs and preferences. Since HMD systems are programmable, there is an opportunity to optimally tailor each system to meet the visual needs and usability concerns of a specific user. Ultimately, there is a strong potential for HMD technology to provide considerable benefit to individuals with low vision and various forms of visual impairment (Brock & Kristensson, 2013; Kunz et al., 2018). Future HMD-based assistive technology devices may incorporate advanced features that improve usability, such as optical character recognition (e.g., text-to-speech and speech-output) and face and object recognition through artificial intelligence (Hanson et al., 2020; Liu et al., 2018). NuEyes includes text-to-speech as a feature. Prior research has identified the limitations of available HMDs and areas for improvement in the future (Deemer et al., 2019; Gallos et al., 2018; Twa, 2018). Consistent with the development of mobile health applications (Alwashmi et al., 2019), using a mixed-methods approach proved essential for characterizing user preferences and their relationship with VRQOL, visual acuity, and ocular diagnosis. This study provides insights that may shape future device development.
Limitations
There were several limitations to this study. First, participants were asked to operate and evaluate the HMDs shortly after being taught to use them. Particularly among participants who were unfamiliar with technological devices, “discomfort” with technology may have affected HMD and usability preferences; these preferences could change over time as users become more familiar and comfortable with HMD use. Eye fatigue could have also resulted with a longer duration of device use (Wang et al., 2019). Moreover, not all eye diseases that result in visual impairment were represented in this study. Caution should be extended in extending to individuals with other conditions or types of vision loss (e.g., contrast sensitivity, stereoacuity, or dark adaptation impairment). Nevertheless, the convergent mixed-methods study design provided the opportunity to study granular and nuanced insights of study participants, while also determining how these insights varied based on key objective data like ocular diagnosis and visual acuity, as well as quantitatively assessed VRQOL.
Conclusions
Technology for HMD low vision enhancement systems will continue to evolve in the future and people will benefit from commercially available devices that will be more powerful, aesthetically acceptable, and user-centered. This study illustrates that usability, clarity, magnification, and vision improvement are primary factors that are important to individuals with low vision when considering HMDs. Furthermore, the user preferences of visually impaired adults are varied, and customization and user-centered design features may facilitate wider adoption of HMD technologies by people with low vision even as technology continues to advance. Additional research and targeted device development are therefore required to optimally meet the specific needs and preferences of people with low vision.

Reasons Participants Stated for Their HMD Preference.
Footnotes
Authors’ Note
SEM (nonrelevant to the content of this research) and JDW (nonrelevant to the content of this research).
Acknowledgments
The authors would like to thank the Michigan Institute for Clinical & Health Research grant support (CTSA: UL1TR002240) for the use of REDCap software relating to this project. The authors also thank Dr Ecosse Lamoureux, Dr Eva Fenwick, and the Singapore Eye Research Institute for providing the IVI Questionnaire for use in this study.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by an unrestricted grant from Research to Prevent Blindness to the Department of Ophthalmology and Visual Sciences at the University of Michigan. JRE is supported by a grant from the National Eye Institute (K23EY027848). Pilot grant support was provided to JRE from the Michigan Institute for Clinical & Health Research (CTSA: UL1TR002240).
