About

Phillip Yuhas is an Assistant Professor at The Ohio State University College of Optometry, where he has completed his third year in this role. He is a clinician-scientist with a focus on the effects of repeated traumatic brain injuries on the retina and ocular biomechanics in conditions such as glaucoma and keratoconus. As a licensed optometrist, his clinical interests include the diagnosis and management of ocular disease, especially glaucoma, fitting specialty lenses, and understanding the consequences of blue light exposure on the eye and visual system. Yuhas contributes to the college by teaching OPTOM 7740, Management of Glaucoma, and staffing the Advanced Ocular Care clinic. He actively participates in the Faculty Advisory Committee, proctors student practical examinations, and leads the SocialEyes student group. He also serves on the VITA Advisory Committee for the university. Beyond Ohio State, he is involved with the National Board of Examiners in Optometry Refractive Error Committee and Council, and works with groups advancing the missions of the American Academy of Optometry and the American Optometric Association. His research includes studies on biomechanical waveform parameters in different cohorts, retinal nerve fiber layer degradation following traumatic brain injuries, and various aspects of pupillary responses to light stimuli, contributing to the understanding of ocular responses to traumatic brain injury and light exposure.

Research topics

• Ophthalmology
• Medicine
• Optics
• Physics
• Anatomy
• Geometry
• Materials science
• Psychology
• Optometry
• Mathematics
• Mathematical analysis
• Internal medicine

Selected publications

• AUBADE-syn: a novel deep learning ensemble method for glaucoma detection using synthetic fundus images on imbalanced datasets

  npj Digital Medicine · 2026-05-14

  articleOpen access

  Glaucoma is the leading cause of irreversible blindness worldwide. Early detection is essential to preserve vision. Deep learning approaches have shown promise in automating glaucoma detection. However, significant class imbalance in medical datasets often impairs classifier performance. To address this challenge, we propose AUBADE-syn, a deep learning ensemble framework that integrates synthetic image generation with structured class-balancing strategies. Our approach leverages optic nerve head-centered regions and a classifier-free guided diffusion model to generate realistic glaucomatous images, enriching the minority class and improving model generalization on highly imbalanced datasets. We benchmarked the AUBADE-syn algorithm against widely used methods for addressing class imbalance, including weighted loss functions, focal loss, Balanced-MixUp, ProCo, and FlexDA. On EyePACS, a large-scale public dataset with a 1:30 class imbalance ratio, AUBADE-syn achieved an area under the receiver operating characteristic curve of 0.992, outperforming all comparison methods. We also validated its performance across ten independent public datasets and fine-tuned the model on three additional public datasets, achieving top-tier or competitive results relative to previously published methods. Overall, these results demonstrate that AUBADE-syn consistently improves both discrimination and calibration for glaucoma detection in highly imbalanced settings, highlighting the effectiveness of domain-aware synthetic augmentation and structured ensemble learning for imbalanced medical imaging tasks.

  PublisherDOI

• Adaptive Temporal Mixture of Experts for Predicting Stiffness Metrics From the Ocular Response Analyzer and Identifying Keratoconus

  American Journal of Ophthalmology · 2026-02-27

  articleOpen access
  PublisherOA PDFDOI

• Waveform Score Influences the Outcome Metrics of the Ocular Response Analyzer in Patients with Keratoconus and in Healthy Controls

  Current Eye Research · 2025-04-10 · 2 citations

  article1st authorCorresponding

  PURPOSE: To test whether the intraocular-pressure (IOP) and biomechanical outcome metrics from the Ocular Response Analyzer (ORA) differ between the measurement with the highest waveform score and the average of four measurements of any waveform score in participants with keratoconus and in controls. METHODS: Patients with diagnosed keratoconus and healthy controls were recruited prospectively. Four measurements were made using a third-generation ORA. Goldmann-correlated IOP (IOPg), corneal-compensated IOP (IOPcc), corneal hysteresis (CH), corneal resistance factor (CRF), waveform score, and six waveform parameters (p1area, p2area, w1, w2, h1, and h2) were considered as outcome metrics. In the left eye, outcomes from the measurement with the highest waveform score were compared against averaged outcomes from four measurements of any waveform score using either paired t-tests or Wilcoxon signed-rank tests. Receiver operating characteristic (ROC) curves tested ability of both data-selection approaches to differentiate the cohorts. RESULTS: = 47), waveform score, p1area, p2area, h1, and h2 were all significantly greater for the best-waveform-score measurement than for the average-waveform-score measurement. W2 was significantly less for the best-waveform-score measurement than for the average-waveform-score measurement. The area under the ROC curve was high for both data-selection approaches. CONCLUSION: In general, the best measurement had higher and narrower waveform peaks than the averaged measurement, which suggests better alignment between the device and the eye in the former than in the latter. Thus, making multiple measurements and then analyzing the one with the single highest quality may be preferred to analyzing the average of the group.

  PublisherDOI

• Intra- and Inter-Observer Agreement of a Portable A-Scan Ultrasound Biometer in Sitting and Supine Positions and Validity Against an Optical Biometer

  Clinical Optometry · 2025-09-01

  articleOpen accessSenior author

  Purpose: The purposes of this study were to test the validity of a portable ultrasound biometer (6000 Scanmate-A, DGH Technology) against an optical biometer and to establish its repeatability and reproducibility on patients in sitting and supine positions. Patients and Methods: Healthy adults (N = 51) were recruited prospectively. At two study visits, Observer 1 made five measurements of AL and ACD in the right eye using an optical biometer (IOLMaster, Zeiss). Then two sets of three measurements of AL and ACD were made using the Scanmate-A, one while participants were sitting and another while they were supine. Observer 2 made the same measurements at one visit on a subset of participants (n = 42). Averages of the measurements were prepared for analysis (significance threshold set to α = 0.01) according to Bland and Altman, repeated-measures analyses of variance, linear correlations, and intraclass correlation coefficients (ICC). Results: AL and ACD were significantly shorter when measured by the Scanmate-A (median [interquartile range] = 23.4 [22.5, 24.7] mm and mean ± standard deviation = 3.22 ± 0.45 mm, respectively) than when measured by IOLMaster (23.8 [23.1, 25.2] mm and 3.51 ± 0.33 mm, respectively). For both body positions, there were no differences in ACD and AL between visit 1 and visit 2. Correlations were strong between the visits, but the limits of agreement varied. Similarly, there were no significant differences between Observer 1 and Observer 2. Correlations and ICC were strong-to-moderate between the observers, and the limits of agreement varied. Conclusion: Consistent with reports on other a-scan devices, the Scanmate-A measured shorter AL and shallower ACD than the IOLMaster. Although, mean or median AL and ACD did not differ across study visits, observers, and body positions, large limits of agreement on Bland-Altman analyses must be accounted for by users of the Scanmate-A.

  PublisherOA PDFDOI

• Corneal Cross-Linking Induces Increased Corneal Refractive Index, Which Generates Measurement Error of Central Corneal Thickness

  Ophthalmology Science · 2025-08-05 · 1 citations

  articleOpen accessSenior author

  Purpose: ) on measured central corneal thickness (CCT) by Scheimpflug tomography and anterior-segment OCT. Design: Theoretical analysis with prospective cohort study validation. Participants: Twenty-four eyes from 23 participants met the criteria for inclusion in data analysis. Methods: < 0.05. Device-specific equations for CCT and n were solved iteratively at follow-up for those subjects with minimal difference in CCT at baseline to account for measurement error. Main Outcome Measures: Central corneal thickness and n Results: < 0.001). Conclusions: to determine CCT. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

  PublisherDOI

• Intraocular pressure and ocular biomechanical parameters vary between generations of the ocular response analyzer in healthy and ectatic eyes

  Frontiers in Medicine · 2025-07-24 · 1 citations

  articleOpen access1st authorCorresponding

  Introduction This study evaluated the agreement between a third-generation (G3) ocular response analyzer (ORA) and a first-generation (G1) ORA, and tested the ability of the keratoconus match index (KMI) to identify keratoconus. Methods Healthy participants ( n = 149 eyes) and participants with keratoconus ( n = 78 eyes) were enrolled for this study. Four measurements were taken bilaterally using the G1 and G3 ORA. Goldmann-correlated intraocular pressure (IOPg), corneal-compensated IOP (IOPcc), corneal hysteresis (CH), waveform score, KMI, and waveform parameters area under the first applanation peak (p1area), area under the second applanation peak (p2area), width of the first applanation peak (w1), width of the second applanation peak (w2), height of the first applanation peak (h1), and height of the second applanation peak (h2) were recorded from the measurement with the highest waveform score in the left eye. Paired t -tests or Wilcoxon signed-rank tests were used to assess agreement between the devices, and receiver-operating characteristic curves determined the ability of KMI to identify eyes with keratoconus. Results There was no difference in IOPcc or IOPg between the devices in both cohorts. CH was significantly greater for the G3 than for the G1 in healthy participants but not in keratoconus participants. For both cohorts, measurements of waveform score, KMI, p1area, p2area, w2, h1, and h2 were greater for the G3 than for the G1. Only w1 was smaller for the G3 than for the G1. There was no difference in the ability of KMI to differentiate ectatic from healthy eyes between the devices. Discussion Although the G1 and G3 can identify keratoconus using KMI, there is meaningful variation between them in IOP and biomechanical outcome parameters. Thus, clinicians and researchers should compare results between the devices with caution and should state which generation produced the data.

  PublisherOA PDFDOI

• Contact lenses and digital eye strain

  Clinical and Experimental Optometry · 2025-11-13

  article1st authorCorresponding

  Digital eye strain (DES) is a group of eye- and vision-related problems elicited by engagement with screens. Over half of screen users experience its symptoms, which include eyestrain, headaches, blurred vision, dry eyes, and neck and shoulder pain. Although contact lenses are a popular option for the correction of refractive error, their role in the aetiology and management of DES is still emerging. Recent review publications on DES only briefly discuss contact lenses. Thus, there is a need (1) to synthesise the current body of literature that has examined the relationship between contact lenses and DES and (2) to identify gaps in the literature that are barriers to employing contact lenses to manage DES. The purpose of this narrative review is to evaluate evidence of ways in which contact lenses may contribute to and may help relieve DES. Although there is little direct evidence to support or refute the use of contact lenses for the management of DES, advances in contact-lens technology to maintain the ocular surface and to correct vision at multiple distances in multiple gazes support the potential of future research. Moreover, there is little high-quality evidence that blue-light filters reduce the symptoms of DES. Nevertheless, eye care professionals should consider the aforementioned potential benefits of wearing contact lenses during computer tasks, versus possible lens-associated ocular-surface disruption and residual astigmatism, when discussing management options with patients. Future gains in knowledge of the interaction between DES and contact lenses will have significance to eye care professionals and to their patients.

  PublisherDOI

• Keratoconus cone location influences ocular biomechanical parameters measured by the Ocular Response Analyzer

  Eye and Vision · 2024-01-03 · 16 citations

  articleOpen access1st authorCorresponding

  BACKGROUND: Keratoconus is characterized by asymmetry in the biomechanical properties of the cornea, with focal weakness in the area of cone formation. We tested the hypothesis that centrally-measured biomechanical parameters differ between corneas with peripheral cones and corneas with central cones. METHODS: Fifty participants with keratoconus were prospectively recruited. The mean ± standard deviation age was 38 ± 13 years. Axial and tangential corneal topography were analyzed in both eyes, if eligible. Cones in the central 3 mm of the cornea were considered central, and cones outside the central 3 mm were considered peripheral. Each eye was then measured with the Ocular Response Analyzer (ORA) tonometer. T-tests compared differences in ORA-generated waveform parameters between cohorts. RESULTS: Seventy-eight eyes were analyzed. According to the axial topography maps, 37 eyes had central cones and 41 eyes had peripheral cones. According to the tangential topography maps, 53 eyes had central cones, and 25 eyes had peripheral cones. For the axial-topography algorithm, wave score (WS) was significantly higher in peripheral cones than central cones (inter-cohort difference = 1.27 ± 1.87). Peripheral cones had a significantly higher area of first peak, p1area (1047 ± 1346), area of second peak, p2area (1130 ± 1478), height of first peak, h1 (102 ± 147), and height of second peak, h2 (102 ± 127), than central cones. Corneal hysteresis (CH), width of the first peak, w1, and width of the second peak, w2, did not significantly differ between cohorts. There were similar results for the tangential-topography algorithm, with a significant difference between the cohorts for p1area (855 ± 1389), p2area (860 ± 1531), h1 (81.7 ± 151), and h2 (92.1 ± 131). CONCLUSIONS: Cone location affects the biomechanical response parameters measured under central loading of the cornea. The ORA delivers its air puff to the central cornea, so the fact that h1 and h2 and that p1area and p2area were smaller in the central cone cohort than in the peripheral cone cohort suggests that corneas with central cones are softer or more compliant centrally than corneas with peripheral cones, which is consistent with the location of the pathology. This result is evidence that corneal weakening in keratoconus is focal in nature and is consistent with localized disruption of lamellar orientation.

  PublisherOA PDFDOI

• Letter to the Editor: The Spectral Bands Task Force: A consensus on the representation of high‐energy visible radiation

  Optometry and Vision Science · 2024-04-01 · 4 citations

  letter
  PublisherDOI

• Henle fiber layer thickening and deficits in objective retinal function in participants with a history of multiple traumatic brain injuries

  Frontiers in Neurology · 2024-02-06 · 2 citations

  articleOpen accessSenior authorCorresponding

  Introduction This study tested whether multiple traumatic brain injuries (TBIs) alter the structure of the Henle fiber layer (HFL) and degrade cell-specific function in the retinas of human participants. Methods A cohort of case participants with multiple TBIs and a cohort of pair-matched control participants were prospectively recruited. Directional optical coherence tomography and scanning laser polarimetry measured HFL thickness and phase retardation, respectively. Full-field flash electroretinography (fERG) assessed retinal function under light-adapted (LA) 3.0, LA 30 Hz, dark-adapted (DA) 0.01, DA 3.0, and DA 10 conditions. Retinal imaging and fERG outcomes were averaged between both eyes, and paired t-tests or Wilcoxon signed-rank tests analyzed inter-cohort differences. Results Global HFL thickness was significantly ( p = 0.02) greater in cases (8.4 ± 0.9 pixels) than in controls (7.7 ± 1.1 pixels). There was no statistically significant difference ( p = 0.91) between the cohorts for global HFL phase retardation. For fERG, LA 3.0 a-wave amplitude was significantly reduced ( p = 0.02) in cases (23.5 ± 4.2 μV) compared to controls (29.0 ± 8.0 μV). There were no other statistically significant fERG outcomes between the cohorts. Discussion In summary, the HFL thickens after multiple TBIs, but phase retardation remains unaltered in the macula. Multiple TBIs may also impair retinal function, indicated by a reduction in a-wave amplitude. These results support the potential of the retina as a site to detect TBI-associated pathology.

  PublisherOA PDFDOI

Frequent coauthors

• Cynthia J. Roberts

  University of Manchester

  47 shared

• Andrew T. E. Hartwick

  SUNY College of Optometry

  13 shared

• Joshua Canavan

  The Ohio State University

  10 shared

• Catherine McDaniel

  The Ohio State University

  9 shared

• Ashraf M. Mahmoud

  The Ohio State University

  8 shared

• Patrick Shorter

  Optica

  8 shared

• Michael Earley

  7 shared

• Cora McHugh-Morrison

  The Ohio State University

  6 shared

Labs

• Karla Zadnik LabPI

Education

• PhD, College of Optometry

  The Ohio State University

  2019

• MS, Vision Science

  The Ohio State University

  2014

• OD, Optometry

  The Ohio State University

  2014

• BA

  University of Notre Dame

  2010

Awards & honors

• Alumni Awards (2021, 2022, 2023, 2024, 2025, 2015, 2019)
• Reunion Weekend Alumni Awards (2021, 2022, 2023, 2024, 2025,…