Evaluation of an Augmented Reality-Based Visual Aid for People with Peripheral Visual Field Loss
Round 1
Reviewer 1 Report
Comments and Suggestions for AuthorsThis study evaluates the clinical effectiveness of "Retiplus", an AR-based visual rehabilitation device for patients with retinitis pigmentosa. While the research aims to contribute to practical visual rehabilitation methods by investigating visual field expansion and improvements in daily activities in a clinical setting, the paper requires significant improvements in its technical explanations and experimental methodology.
- The mechanism of AR-based vision enhancement is not sufficiently explained. In particular, the paper needs to clarify how the "5x minification display mode" works and its role in expanding the visual field: when displaying minified wide-angle camera images, misalignment between the real world and the AR images is inevitable.
- The rationale for choosing an OST display remains unclear. Given the unavoidable misalignment between real world and AR images in the minification mode, a VST display that can occlude real-world light might be more effective. The paper should include a thorough discussion of this design choice.
- To understand the effects of AR-based vision enhancement, the paper must include actual AR images. For example, a comparison of AR display images captured by a camera positioned at the optical system with corresponding real-world images would better demonstrate the vision enhancement effects.
- The paper lacks detailed descriptions necessary for experimental reproducibility, including specific procedures for training sessions. For example, although different AR display modes would be used for low light support and visual field expansion, these distinctions are not specified.
- The calibration method for aligning the OST-HMD with the world camera, which is essential for proper operation, is not described.
In conclusion, although this study provides clinically important findings, it needs substantial improvements in its methodological descriptions and systematic analysis of results. I recommend resubmission after major revision.
Author Response
Reviewer 1
We appreciate the comments which have helped us improve the manuscript.
This study evaluates the clinical effectiveness of "Retiplus", an AR-based visual rehabilitation device for patients with retinitis pigmentosa. While the research aims to contribute to practical visual rehabilitation methods by investigating visual field expansion and improvements in daily activities in a clinical setting, the paper requires significant improvements in its technical explanations and experimental methodology.
Comment 1. The mechanism of AR-based vision enhancement is not sufficiently explained. In particular, the paper needs to clarify how the "5x minification display mode" works and its role in expanding the visual field: when displaying minified wide-angle camera images, misalignment between the real world and the AR images is inevitable.
Response 1. Thank you for your valuable comment. We agree that the explanation of the 5x minification display mode required further clarification. In the revised manuscript, we have expanded the description to explain that this mode displays a wide-angle camera image reduced to one-fifth of its original size, allowing users to perceive a broader scene within their remaining central visual field (Section 2.2, Page 4, Lines 160-169). The AR vision enhancement system is implemented through an optical see-through (OST) head-mounted display (Epson Moverio BT-350). The device combines the user's residual natural vision with digital information projected onto transparent lenses. In our approach, a front-facing wide-angle camera captures real-time video of the environment. In the 5x minification display mode, the captured wide-angle video feed is digitally reduced (minified) by a factor of five and overlaid onto the transparent display of the HMD. This minified image is shown within the user's intact central field of vision, effectively compressing peripheral scene information into a smaller area that can be perceived centrally. For patients with peripheral visual field loss—such as those with retinitis pigmentosa—this mode compensates for their restricted FOV by bringing peripheral elements of the scene into the central visual area without requiring head or eye movements to scan the environment. Thus, the user gains spatial awareness of obstacles, people, or hazards located outside their residual field.
Regarding misalignment, we acknowledge that geometric discrepancies between the real-world view and the AR overlay are unavoidable due to the offset between the camera's position and the user's eyes, as well as system processing delays. However, the goal of the minification mode is not precise spatial registration. Instead, it aims to provide complementary situational awareness by alerting the user to the presence of objects beyond their natural field, rather than to replace or perfectly align with their direct view. Users are instructed to interpret the minified image as an additional, generalized spatial cue rather than an exact overlay of reality. We have added this clarification to the Discussion section (Page 12, Lines 454-461).
Comment 2. The rationale for choosing an OST display remains unclear. Given the unavoidable misalignment between real world and AR images in the minification mode, a VST display that can occlude real-world light might be more effective. The paper should include a thorough discussion of this design choice.
Response 2. Thank you for your insightful comment. We would like to clarify that the use of an Optical See-Through (OST) display in this study was determined by the commercial nature of the device (Epson Moverio BT-350), which is not modifiable in terms of display type or optical configuration. Our study focused on evaluating the potential benefits of an existing, commercially available AR solution for individuals with peripheral field loss, rather than designing a new system from the ground up.
We agree that a Video See-Through (VST) system could provide advantages in terms of alignment and the possibility of fully controlling the visual scene, including occluding real-world light. However, commercial VST solutions are currently less accessible, and OST systems like the one used in this study remain more practical for daily life due to their lower latency, lighter weight, and the ability to preserve the user's direct view of the environment through transparent lenses.
To address your suggestion, we have added a discussion in the revised manuscript acknowledging the limitations of the OST approach, the potential advantages of VST systems, and the relevance of future studies exploring alternative display technologies that might overcome the misalignment challenges associated with OST designs (Discussion section, Page 11, Lines 406-410 and Page 12, Lines 454-461, 466-468).
Comment 3. To understand the effects of AR-based vision enhancement, the paper must include actual AR images. For example, a comparison of AR display images captured by a camera positioned at the optical system with corresponding real-world images would better demonstrate the vision enhancement effects.
Response 3. Thank you for this suggestion. We agree that including actual AR display images would help illustrate the system's visual effects. Accordingly, we have added a representative image captured through the OST-HMD's optical path, showing the minification display mode. This image is now included in a new figure (Section 2.2., Figure 1.b) alongside corresponding real-world scene images for direct comparison. We believe this addition significantly improves the readers’ understanding of the system's visual output.
Comment 4. The paper lacks detailed descriptions necessary for experimental reproducibility, including specific procedures for training sessions. For example, although different AR display modes would be used for low light support and visual field expansion, these distinctions are not specified.
Response 4. Thank you for this observation. We have revised the Methods section to include a more detailed description of the experimental protocol, including the specific procedures used during participant training sessions (Section 2.3, Page 5, Lines 174-198). We have clarified the distinctions between the different AR display modes—such as low-light enhancement and visual field expansion—and how participants were instructed to use and switch between them. These additions aim to ensure full reproducibility of the study (Section 2.2, Pages 4-5, Lines 144-149, 154-171).
Comment 5. The calibration method for aligning the OST-HMD with the world camera, which is essential for proper operation, is not described.
Response 5. Thank you for your comment. Regarding the calibration between the OST-HMD and the world camera, we would like to clarify that the system used in this study is a commercial device designed for people with low vision. This device comes pre-configured from the manufacturer to ensure proper alignment between the virtual overlay and the real-world scene captured by the camera. Therefore, no additional calibration was required on our part, as the system was used according to the manufacturer’s specifications.
Although no manual calibration is performed, the manufacturer has designed the system with presetting based on user models, ensuring optimal alignment in most cases. This design choice minimizes the need for individual adjustments, while maintaining consistent performance among different users.
To clarify this point, we have added the following sentence to the manuscript (Section 2.2, Page 4, Lines 130-132):
“This device is a commercial, pre-configured system that does not require manual calibration by the user or the researchers to ensure correct alignment between the world camera and the projection on the OST-HMD”.
Reviewer 2 Report
Comments and Suggestions for AuthorsThis is a well-rounded research into the effect of AR visual-aid on activities of daily living for people with peripheral vision loss. The authors conducted visual function tests and functional vision surveys with and without aid and present evidence that the aid is an overall improvement on ADL.
However, the methodology has some minor limitations:
- Why was full-field perimetry not used for measurement of visual field?
- Five weekly sessions constitute significant time commitment from the subjects. Were all AR training performed in consecutive weeks and were any subjects excluded for being unable to fulfill the training?
- The training protocol needs to be mentioned. Without that knowledge, readers have no indication of the necessity of five one-hour sessions. Would fewer sessions still have similar results?
- How the subjects use this aid is also very unclear. Do they alter enhancements per task, or is it consistent after the calibration performed by the low vision specialist.
- If the camera capture range is 23 degrees, how does it improve VF range in people with VF greater than 23 degrees? Does the 23 degree FOV of the camera have a non-overlapping region with respect to the eyes?
- No detail is provided about the filter, full-screen modes.
- I would recommend including tasks and questions that are dependent and indicative of central field proficiency in the survey. If the without aid variant does not show performance degradation in those tasks, that would be inconsistent with VA reduction. For example, in the table 4, item 8, the subjects were asked about the difficulty of seeing signs. However, their performance in the task actually improved, although with reduced VA, their performance should have worsened.
- In page 10, some limitations of the HoloLens system (such as lag, dynamic range, brightness etc.) were mentioned. However, no comparison was made between the specifications of the Epson HMD and the HoloLens.
- The authors should mentioned the remote controller usage pattern of participants during VA, VFA and ADL tasks. This would help the reader to understand the cognitive load necessary for ADL and would also indicate daily usage.
Overall, the study has great merit and I would recommend accept with revisions.
Comments for author File: Comments.pdf
Author Response
Reviewer 2
We appreciate the comments which have helped us improve the manuscript.
This is well-rounded research into the effect of AR visual-aid on activities of daily living for people with peripheral vision loss. The authors conducted visual function tests and functional vision surveys with and without aid and present evidence that the aid is an overall improvement on ADL.
However, the methodology has some minor limitations:
Comment 1. Why was full-field perimetry not used for measurement of visual field?
Response 1. We acknowledge the importance of full-field perimetry; however, due to the specific characteristics of the augmented reality (AR) aid, we opted for evaluate the binocular visual fields using a tangent screen at 1 m. In this way, it is possible to assess the visual field while the patients are wearing the augmented reality aid. We have clarified this point in Section 2.3. (Page 5, Lines 209-211) of the revised manuscript.
Comment 2. Five weekly sessions constitute significant time commitment from the subjects. Were all AR training performed in consecutive weeks and were any subjects excluded for being unable to fulfill the training?
Response 2. All AR training were conducted in consecutive weeks to ensure consistency in adaptation and skill acquisition. No subjects were excluded based on their inability to complete the training. We have added this information in Section 2.3. (Page 5, Lines 173-174, 199).
Comment 3. The training protocol needs to be mentioned. Without that knowledge, readers have no indication of the necessity of five one-hour sessions. Would fewer sessions still have similar results?
Response 3. We have now included a detailed description of the training protocol in Section 2.3 (Page 5, Lines 174-198). The five-session structure was determined based on pilot testing. While fewer sessions may yield some improvement, the five-session design ensured optimal adaptation and familiarization with the AR system.
Comment 4. How the subjects use this aid is also very unclear. Do they alter enhancements per task or is it consistent after the calibration performed by the low vision specialist.
Response 4. The enhancements remained consistent after the initial calibration by the low vision specialist. Participants did not manually adjust the enhancements for different tasks, as the calibration was designed to provide optimal assistance across various activities. Participants could only modify brightness, contrast, and zoom during the training sessions and the execution of ADL tasks. We measured visual acuity and visual field under the same conditions for all patients. We have clarified this aspect in Section 2.2 (Page 4, Lines 144-149, Page 5, Lines 169-171).
Comment 5. If the camera capture range is 23 degrees, how does it improve VF range in people with VF greater than 23 degrees? Does the 23-degree FOV of the camera have a non-overlapping region with respect to the eyes?
Response 5. The AR aid enhances visual awareness by dynamically projecting captured visual information within the user’s functional field of view. Even for users with VF greater than 23 degrees, the system provides an advantage by allowing the user to retain his habitual vision while benefiting from the 5x minification display mode. However, since Retiplus is an AR aid designed for patients with orientation and mobility issues, and in accordance with other works (Xu et al., 2024) we believe that Retiplus could have a greater effect on VF enlargement in patients with more severe VF damage (VF ≤ 10º).
On the other hand, since the device utilizes a semi-transparent display, the patient's natural field of vision beyond the screen remains unaffected. This allows them to perceive their surroundings and objects below by shifting their gaze. Additionally, because the display overlays information rather than obstructing the view, there is no complete occlusion of the environment, ensuring that the patient can still use their residual peripheral vision effectively. We have added a paragraph clarifying this aspect in Section 2.2 (Page 3, Lines 125-130).
- Xu, D.; Yu, M.; Zheng, C.; Ji, S.; Dai, J. The effects of an electronic head-mounted display in vision rehabilitation for patients with tunnel vision. Int. Ophthalmol. 2024, 44(1), 109.
Comment 6. No detail is provided about the filter, full-screen modes.
Response 6. The filter used is a category 3 neutral density filter designed to ensure that the patient perceives only the augmented reality image. This setup encourages a bioptic viewing strategy, where the patient consciously shifts their gaze to access their residual vision. In the final training session, the patient lifts the filter to perceive both the augmented reality and their natural vision simultaneously. However, this advanced mode was not the focus of the present study. Regarding the full-screen mode, it allows the augmented reality information to occupy the entire display. In this study, all patients used the device with the filter in place and the full-screen mode activated.
According to the reviewer’s comment, we have added detailed descriptions of the filter and full-screen modes in Section 2.2 (Pages 4-5, Lines 155-171), specifying their purpose, implementation, and impact on user experience.
Comment 7. I would recommend including tasks and questions that are dependent and indicative of central field proficiency in the survey. If the without aid variant does not show performance degradation in those tasks, that would be inconsistent with VA reduction. For example, in the table 4, item 8, the subjects were asked about the difficulty of seeing signs. However, their performance in the task actually improved, although with reduced VA, their performance should have worsened.
Response 7: Retiplus is designed as a minification aid for patients with peripheral field loss, not as a magnification tool for those with central field loss. For this reason, the survey primarily focuses on items related to orientation and mobility rather than central field-dependent tasks. When performing near-vision tasks, participants either do not use the device or rely on the bioptic mode.
Regarding the improvement in sign perception, this can be explained by the device’s adjustable brightness, contrast and zoom settings. These allow for a controlled reduction in minification, preventing a significant loss of visual acuity. As a result, participants may experience enhanced perception of signs despite the expected VA reduction. Additionally, it is possible that the subjective perception of difficulty reported by patients does not always correlate with objective visual acuity measurements.
For future studies, we will consider incorporating the reviewer’s suggestion by adding questions specifically designed to assess central field proficiency. Furthermore, we have clarified this aspect in the results section (Page 7, Lines 279-283).
Comment 8. In page 10, some limitations of the HoloLens system (such as lag, dynamic range, brightness etc.) were mentioned. However, no comparison was made between the specifications of the Epson HMD and the HoloLens.
Response 8: Thank you for your valuable feedback. We have now included a comparison between the Epson HMD and the HoloLens in section Discussion (Page 11, Lines 406-412), focusing on key technical specifications and limitations.
Comment 9. The authors should mention the remote controller usage pattern of participants during VA, VFA and ADL tasks. This would help the reader to understand the cognitive load necessary for ADL and would also indicate daily usage.
Response 9: During the training sessions, participants were able to familiarize themselves with the use of the remote controller to adjust zoom, contrast, and brightness. During the execution of ADL tasks, they could also modify these parameters as needed. The use of the remote controller is intuitive and comparable to using a TV remote to change the volume, so it did not represent a significant cognitive load. No participants reported difficulties in handling it, suggesting that this type of interaction did not interfere with task execution. This ease of use suggests that the system could be easily adopted in daily life without adding an extra cognitive burden.
According to the reviewer’s comment, we have included a detailed explanation of the remote controller usage in Section 2.2 (Page 4, Lines 144-149). This addition provides insights into cognitive load considerations and potential implications for daily use of the AR aid.
Round 2
Reviewer 1 Report
Comments and Suggestions for AuthorsI appreciate the authors' careful and detailed revisions.
They have adequately addressed the points raised in the previous manuscript, significantly enhancing the quality of the paper.
Therefore, I recommend this paper for acceptance.