Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography
IntroductionGlaucoma, optic neuritis (ON), and non-arteritic anterior ischemic optic neuropathy (NAION) produce distinct patterns of retinal ganglion cell (RGC) damage. We propose a booster Variational Autoencoder (bVAE) to capture spatial variations in RGC loss and generate latent space (LS) montag...
Saved in:
| Main Authors: | , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2025-02-01
|
| Series: | Frontiers in Ophthalmology |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fopht.2024.1497848/full |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1832548254657544192 |
|---|---|
| author | Jui-Kai Wang Jui-Kai Wang Brett A. Johnson Zhi Chen Zhi Chen Honghai Zhang Honghai Zhang David Szanto Brian Woods Brian Woods Michael Wall Young H. Kwon Young H. Kwon Edward F. Linton Edward F. Linton Andrew Pouw Mark J. Kupersmith Mark J. Kupersmith Mark J. Kupersmith Mona K. Garvin Mona K. Garvin Mona K. Garvin Mona K. Garvin Randy H. Kardon Randy H. Kardon |
| author_facet | Jui-Kai Wang Jui-Kai Wang Brett A. Johnson Zhi Chen Zhi Chen Honghai Zhang Honghai Zhang David Szanto Brian Woods Brian Woods Michael Wall Young H. Kwon Young H. Kwon Edward F. Linton Edward F. Linton Andrew Pouw Mark J. Kupersmith Mark J. Kupersmith Mark J. Kupersmith Mona K. Garvin Mona K. Garvin Mona K. Garvin Mona K. Garvin Randy H. Kardon Randy H. Kardon |
| author_sort | Jui-Kai Wang |
| collection | DOAJ |
| description | IntroductionGlaucoma, optic neuritis (ON), and non-arteritic anterior ischemic optic neuropathy (NAION) produce distinct patterns of retinal ganglion cell (RGC) damage. We propose a booster Variational Autoencoder (bVAE) to capture spatial variations in RGC loss and generate latent space (LS) montage maps that visualize different degrees and spatial patterns of optic nerve bundle injury. Furthermore, the bVAE model is capable of tracking the spatial pattern of RGC thinning over time and classifying the underlying cause.MethodsThe bVAE model consists of an encoder, a display decoder, and a booster decoder. The encoder decomposes input ganglion cell layer (GCL) thickness maps into two display latent variables (dLVs) and eight booster latent variables (bLVs). The dLVs capture primary spatial patterns of RGC thinning, while the display decoder reconstructs the GCL map and creates the LS montage map. The bLVs add finer spatial details, improving reconstruction accuracy. XGBoost was used to analyze the dLVs and bLVs, estimating normal/abnormal GCL thinning and classifying diseases (glaucoma, ON, and NAION). A total of 10,701 OCT macular scans from 822 subjects were included in this study.ResultsIncorporating bLVs improved reconstruction accuracy, with the image-based root-mean-square error (RMSE) between input and reconstructed GCL thickness maps decreasing from 5.55 ± 2.29 µm (two dLVs only) to 4.02 ± 1.61 µm (two dLVs and eight bLVs). However, the image-based structural similarity index (SSIM) remained similar (0.91 ± 0.04), indicating that just two dLVs effectively capture the main GCL spatial patterns. For classification, the XGBoost model achieved an AUC of 0.98 for identifying abnormal spatial patterns of GCL thinning over time using the dLVs. Disease classification yielded AUCs of 0.95 for glaucoma, 0.84 for ON, and 0.93 for NAION, with bLVs further increasing the AUCs to 0.96 for glaucoma, 0.93 for ON, and 0.99 for NAION.ConclusionThis study presents a novel approach to visualizing and quantifying GCL thinning patterns in optic neuropathies using the bVAE model. The combination of dLVs and bLVs enhances the model’s ability to capture key spatial features and predict disease progression. Future work will focus on integrating additional image modalities to further refine the model’s diagnostic capabilities. |
| format | Article |
| id | doaj-art-3a90b2bedafc4242903109293caedec0 |
| institution | Kabale University |
| issn | 2674-0826 |
| language | English |
| publishDate | 2025-02-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Ophthalmology |
| spelling | doaj-art-3a90b2bedafc4242903109293caedec02025-02-03T06:33:39ZengFrontiers Media S.A.Frontiers in Ophthalmology2674-08262025-02-01410.3389/fopht.2024.14978481497848Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomographyJui-Kai Wang0Jui-Kai Wang1Brett A. Johnson2Zhi Chen3Zhi Chen4Honghai Zhang5Honghai Zhang6David Szanto7Brian Woods8Brian Woods9Michael Wall10Young H. Kwon11Young H. Kwon12Edward F. Linton13Edward F. Linton14Andrew Pouw15Mark J. Kupersmith16Mark J. Kupersmith17Mark J. Kupersmith18Mona K. Garvin19Mona K. Garvin20Mona K. Garvin21Mona K. Garvin22Randy H. Kardon23Randy H. Kardon24Center for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesDepartment of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United StatesIowa Institute for Biomedical Imaging, University of Iowa, Iowa City, IA, United StatesDepartment of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United StatesIowa Institute for Biomedical Imaging, University of Iowa, Iowa City, IA, United StatesDepartment of Neurology, Icahn School of Medicine at Mount Sinai, New York, NY, United StatesDepartment of Ophthalmology, University Hospital Galway, Galway, IrelandDepartment of Physics, School of Natural Sciences, University of Galway, Galway, IrelandDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesCenter for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesCenter for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesDepartment of Neurology, Icahn School of Medicine at Mount Sinai, New York, NY, United StatesDepartment of Ophthalmology, Icahn School of Medicine at Mount Sinai, New York, NY, United StatesDepartment of Neurosurgery, Icahn School of Medicine at Mount Sinai, New York, NY, United StatesCenter for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesDepartment of Electrical and Computer Engineering, University of Iowa, Iowa City, IA, United StatesIowa Institute for Biomedical Imaging, University of Iowa, Iowa City, IA, United StatesCenter for the Prevention and Treatment of Visual Loss, Iowa City VA Health Care System, Iowa City, IA, United StatesDepartment of Ophthalmology and Visual Sciences, University of Iowa, Iowa City, IA, United StatesIntroductionGlaucoma, optic neuritis (ON), and non-arteritic anterior ischemic optic neuropathy (NAION) produce distinct patterns of retinal ganglion cell (RGC) damage. We propose a booster Variational Autoencoder (bVAE) to capture spatial variations in RGC loss and generate latent space (LS) montage maps that visualize different degrees and spatial patterns of optic nerve bundle injury. Furthermore, the bVAE model is capable of tracking the spatial pattern of RGC thinning over time and classifying the underlying cause.MethodsThe bVAE model consists of an encoder, a display decoder, and a booster decoder. The encoder decomposes input ganglion cell layer (GCL) thickness maps into two display latent variables (dLVs) and eight booster latent variables (bLVs). The dLVs capture primary spatial patterns of RGC thinning, while the display decoder reconstructs the GCL map and creates the LS montage map. The bLVs add finer spatial details, improving reconstruction accuracy. XGBoost was used to analyze the dLVs and bLVs, estimating normal/abnormal GCL thinning and classifying diseases (glaucoma, ON, and NAION). A total of 10,701 OCT macular scans from 822 subjects were included in this study.ResultsIncorporating bLVs improved reconstruction accuracy, with the image-based root-mean-square error (RMSE) between input and reconstructed GCL thickness maps decreasing from 5.55 ± 2.29 µm (two dLVs only) to 4.02 ± 1.61 µm (two dLVs and eight bLVs). However, the image-based structural similarity index (SSIM) remained similar (0.91 ± 0.04), indicating that just two dLVs effectively capture the main GCL spatial patterns. For classification, the XGBoost model achieved an AUC of 0.98 for identifying abnormal spatial patterns of GCL thinning over time using the dLVs. Disease classification yielded AUCs of 0.95 for glaucoma, 0.84 for ON, and 0.93 for NAION, with bLVs further increasing the AUCs to 0.96 for glaucoma, 0.93 for ON, and 0.99 for NAION.ConclusionThis study presents a novel approach to visualizing and quantifying GCL thinning patterns in optic neuropathies using the bVAE model. The combination of dLVs and bLVs enhances the model’s ability to capture key spatial features and predict disease progression. Future work will focus on integrating additional image modalities to further refine the model’s diagnostic capabilities.https://www.frontiersin.org/articles/10.3389/fopht.2024.1497848/fullvariational autoencoder (VAE)glaucomaoptic neuritis (ON)non-arteritic anterior ischemic optic neuropathy (NAION)retinal ganglion cell (RGC) lossoptical coherence tomography (OCT) |
| spellingShingle | Jui-Kai Wang Jui-Kai Wang Brett A. Johnson Zhi Chen Zhi Chen Honghai Zhang Honghai Zhang David Szanto Brian Woods Brian Woods Michael Wall Young H. Kwon Young H. Kwon Edward F. Linton Edward F. Linton Andrew Pouw Mark J. Kupersmith Mark J. Kupersmith Mark J. Kupersmith Mona K. Garvin Mona K. Garvin Mona K. Garvin Mona K. Garvin Randy H. Kardon Randy H. Kardon Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography Frontiers in Ophthalmology variational autoencoder (VAE) glaucoma optic neuritis (ON) non-arteritic anterior ischemic optic neuropathy (NAION) retinal ganglion cell (RGC) loss optical coherence tomography (OCT) |
| title | Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography |
| title_full | Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography |
| title_fullStr | Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography |
| title_full_unstemmed | Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography |
| title_short | Quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography |
| title_sort | quantifying the spatial patterns of retinal ganglion cell loss and progression in optic neuropathy by applying a deep learning variational autoencoder approach to optical coherence tomography |
| topic | variational autoencoder (VAE) glaucoma optic neuritis (ON) non-arteritic anterior ischemic optic neuropathy (NAION) retinal ganglion cell (RGC) loss optical coherence tomography (OCT) |
| url | https://www.frontiersin.org/articles/10.3389/fopht.2024.1497848/full |
| work_keys_str_mv | AT juikaiwang quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT juikaiwang quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT brettajohnson quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT zhichen quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT zhichen quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT honghaizhang quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT honghaizhang quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT davidszanto quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT brianwoods quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT brianwoods quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT michaelwall quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT younghkwon quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT younghkwon quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT edwardflinton quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT edwardflinton quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT andrewpouw quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT markjkupersmith quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT markjkupersmith quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT markjkupersmith quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT monakgarvin quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT monakgarvin quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT monakgarvin quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT monakgarvin quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT randyhkardon quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography AT randyhkardon quantifyingthespatialpatternsofretinalganglioncelllossandprogressioninopticneuropathybyapplyingadeeplearningvariationalautoencoderapproachtoopticalcoherencetomography |