1
|
Richardson A, Kundu A, Henao R, Lee T, Scott BL, Grewal DS, Fekrat S. Multimodal Retinal Imaging Classification for Parkinson's Disease Using a Convolutional Neural Network. Transl Vis Sci Technol 2024; 13:23. [PMID: 39136960 PMCID: PMC11323992 DOI: 10.1167/tvst.13.8.23] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 02/04/2024] [Accepted: 06/23/2024] [Indexed: 08/16/2024] Open
Abstract
Purpose Changes in retinal structure and microvasculature are connected to parallel changes in the brain. Two recent studies described machine learning algorithms trained on retinal images and quantitative data that identified Alzheimer's dementia and mild cognitive impairment with high accuracy. Prior studies also demonstrated retinal differences in individuals with PD. Herein, we developed a convolutional neural network (CNN) to classify multimodal retinal imaging from either a Parkinson's disease (PD) or control group. Methods We trained a CNN to receive retinal image inputs of optical coherence tomography (OCT) ganglion cell-inner plexiform layer (GC-IPL) thickness color maps, OCT angiography 6 × 6-mm en face macular images of the superficial capillary plexus, and ultra-widefield (UWF) fundus color and autofluorescence photographs to classify the retinal imaging as PD or control. The model consists of a shared pretrained VGG19 feature extractor and image-specific feature transformations which converge to a single output. Model results were assessed using receiver operating characteristic (ROC) curves and bootstrapped 95% confidence intervals for area under the ROC curve (AUC) values. Results In total, 371 eyes of 249 control subjects and 75 eyes of 52 PD subjects were used for training, validation, and testing. Our best CNN variant achieved an AUC of 0.918. UWF color photographs were the most effective imaging input, and GC-IPL thickness maps were the least contributory. Conclusions Using retinal images, our pilot CNN was able to identify individuals with PD and serves as a proof of concept to spur the collection of larger imaging datasets needed for clinical-grade algorithms. Translational Relevance Developing machine learning models for automated detection of Parkinson's disease from retinal imaging could lead to earlier and more widespread diagnoses.
Collapse
Affiliation(s)
- Alexander Richardson
- Duke Eye Center, Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
- Department of Computer Science, Duke University, Durham, NC, USA
| | - Anita Kundu
- Duke Eye Center, Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
| | - Ricardo Henao
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
- Department of Computer Science, Duke University, Durham, NC, USA
| | - Terry Lee
- Duke Eye Center, Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
| | - Burton L. Scott
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
- Department of Neurology, Duke University School of Medicine, Durham, NC, USA
| | - Dilraj S. Grewal
- Duke Eye Center, Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
| | - Sharon Fekrat
- Duke Eye Center, Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Duke University School of Medicine, Durham, NC, USA
- Department of Neurology, Duke University School of Medicine, Durham, NC, USA
| |
Collapse
|
2
|
Robbins CB, Rathinavelu J, Ma JP, Soundararajan S, Stinnett SS, Liu AJ, Johnson KG, Grewal DS, Fekrat S. Peripapillary Optical Coherence Tomography Angiography in Alzheimer's Disease, Mild Cognitive Impairment, and Normal Cognition. Ophthalmic Surg Lasers Imaging Retina 2024; 55:78-84. [PMID: 38346150 DOI: 10.3928/23258160-20240107-01] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/15/2024]
Abstract
OBJECTIVE This study aimed to identify peripapillary microvascular changes in Alzheimer's disease (AD) and mild cognitive impairment (MCI). PATIENTS AND METHODS In this prospective study, 66 eyes of 36 subjects with AD, 119 eyes of 63 with MCI, and 513 eyes of 265 controls with normal cognition were enrolled. Peripapillary capillary perfusion density (CPD), capillary flux index (CFI), and retinal nerve fiber layer (RNFL) thickness were determined. RESULTS Average CPD differed significantly between all three groups (P = 0.001), being significantly greater in AD vs controls (0.446 ± 0.015 vs 0.439 ± 0.017, P = 0.001) and MCI vs controls (0.443 ± 0.020 vs 0.439 ± 0.017, P = 0.007) but not AD vs MCI (P = 0.69). CFI and average RNFL thickness did not significantly differ among groups (all P > 0.05). CONCLUSION Peripapillary CPD is increased in eyes with AD or MCI compared to controls despite similar RNFL thickness. [Ophthalmic Surg Lasers Imaging Retina 2024;55:78-84.].
Collapse
|
3
|
Joseph S, Robbins CB, Allen A, Haystead A, Hemesath A, Kundu A, Ma JP, Johnson KG, Agrawal R, Gunasan V, Stinnett SS, Grewal DS, Fekrat S. Differences in Retinal and Choroidal Microvasculature and Structure in Dementia With Lewy Bodies Compared With Normal Cognition. JOURNAL OF VITREORETINAL DISEASES 2024; 8:67-74. [PMID: 38223776 PMCID: PMC10786081 DOI: 10.1177/24741264231206607] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 01/16/2024]
Abstract
Purpose: To evaluate the retinal and choroidal microvasculature and structure in individuals with dementia with Lewy bodies (DLB) compared with controls with normal cognition using optical coherence tomography (OCT) and OCT angiography (OCTA). Methods: An institutional review board-approved cross-sectional comparison of patients with DLB and cognitively normal controls was performed. The Cirrus HD-OCT 5000 with AngioPlex (Carl Zeiss Meditec) was used to obtain OCT and OCTA images. Results: Thirty-four eyes of 18 patients with DLB and 85 eyes of 48 cognitively normal patients were analyzed. The average capillary perfusion density (CPD) was higher in the DLB group than in the control group (P = .005). The average capillary flux index (CFI) and ganglion cell inner-plexiform layer (GC-IPL) thickness were lower in the DLB group than in the control group (P = .016 and P = .040, respectively). Conclusions: Patients with DLB had an increased peripapillary CPD, decreased peripapillary CFI, and attenuated GC-IPL thickness compared with those with normal cognition.
Collapse
Affiliation(s)
- Suzanna Joseph
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | - Cason B. Robbins
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | - Ariana Allen
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | | | - Angela Hemesath
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | - Anita Kundu
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | | | - Kim G. Johnson
- Department of Neurology, Duke University School of Medicine, Durham, NC, USA
| | - Rupesh Agrawal
- National Healthcare Group Eye Institute, Tan Tock Seng Hospital, Singapore, Singapore
- Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore
| | - Vithiya Gunasan
- National Healthcare Group Eye Institute, Tan Tock Seng Hospital, Singapore, Singapore
- Lee Kong Chian School of Medicine, Nanyang Technological University, Singapore, Singapore
| | - Sandra S. Stinnett
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | - Dilraj S. Grewal
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
| | - Sharon Fekrat
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- iMIND Research Group, Durham, NC, USA
- Department of Neurology, Duke University School of Medicine, Durham, NC, USA
| |
Collapse
|
4
|
Nakamura M, Sakamoto M, Ueda K, Okuda M, Takano F, Yamada-Nakanishi Y. Detection of Relative Afferent Pupillary Defect and Its Correlation with Structural and Functional Asymmetry in Patients with Glaucoma Using Hitomiru, a Novel Hand-Held Pupillometer. J Clin Med 2023; 12:3936. [PMID: 37373631 DOI: 10.3390/jcm12123936] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 03/01/2023] [Revised: 06/03/2023] [Accepted: 06/07/2023] [Indexed: 06/29/2023] Open
Abstract
Patients with asymmetric glaucomatous optic neuropathy (GON) present a relative afferent pupillary defect (RAPD) in the eye with more advanced damage. Although useful, pupillometric RAPD quantification is not widely used as it is not portable. Whether asymmetry of the peripapillary capillary perfusion density (CPD) detected using optical coherence tomography angiography correlates with the severity of RAPD remains unknown. This study assessed RAPD in 81 patients with GON using Hitomiru, a novel hand-held infrared binocular pupillometer. The correlation and ability to detect clinical RAPD based on the swinging flash light test of two independent RAPD parameters (the maximum pupil constriction ratio and the constriction maintenance capacity ratio) were assessed. The coefficient of determination (R2) was calculated between each of the two RAPD parameters and asymmetry of the circumpapillary retinal nerve fiber layer thickness (cpRNFLT), ganglion cell layer/inner plexiform layer thickness (GCL/IPLT), and CPD. The two RAPD parameters showed a correlation coefficient of 0.86 and areas under the receiver operating characteristic (ROC) curve of 0.85-0.88, with R2 being 0.63-0.67 for the visual field, 0.35-0.45 for cpRNFLT, 0.45-0.49 for GCL/IPLT, and 0.53-0.59 for CPD asymmetry. Hitomiru has high discriminatory performance in detecting RAPD in patients with asymmetric GON. CPD asymmetry may better correlate with RAPD than cpRNFLT and GCL/IPLT asymmetry.
Collapse
Affiliation(s)
- Makoto Nakamura
- Division of Ophthalmology, Department of Surgery, Graduate School of Medicine, Kobe University, 7-5-2 Kusunoki-cho, Kobe 650-0017, Japan
| | - Mari Sakamoto
- Division of Ophthalmology, Department of Surgery, Graduate School of Medicine, Kobe University, 7-5-2 Kusunoki-cho, Kobe 650-0017, Japan
| | - Kaori Ueda
- Division of Ophthalmology, Department of Surgery, Graduate School of Medicine, Kobe University, 7-5-2 Kusunoki-cho, Kobe 650-0017, Japan
| | - Mina Okuda
- Division of Ophthalmology, Department of Surgery, Graduate School of Medicine, Kobe University, 7-5-2 Kusunoki-cho, Kobe 650-0017, Japan
| | - Fumio Takano
- Division of Ophthalmology, Department of Surgery, Graduate School of Medicine, Kobe University, 7-5-2 Kusunoki-cho, Kobe 650-0017, Japan
| | - Yuko Yamada-Nakanishi
- Division of Ophthalmology, Department of Surgery, Graduate School of Medicine, Kobe University, 7-5-2 Kusunoki-cho, Kobe 650-0017, Japan
| |
Collapse
|
5
|
Akrobetu DY, Robbins CB, Ma JP, Soundararajan S, Quist MS, Stinnett SS, Moore KPL, Johnson KG, Liu AJ, Grewal DS, Fekrat S. Intrasession Repeatability of OCT Angiography Parameters in Neurodegenerative Disease. OPHTHALMOLOGY SCIENCE 2023; 3:100275. [PMID: 36950088 PMCID: PMC10025280 DOI: 10.1016/j.xops.2023.100275] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Subscribe] [Scholar Register] [Received: 12/14/2022] [Revised: 01/09/2023] [Accepted: 01/20/2023] [Indexed: 02/02/2023]
Abstract
Purpose To assess the intrasession repeatability of macular OCT angiography (OCTA) parameters in Alzheimer's disease (AD), mild cognitive impairment (MCI), Parkinson's disease (PD), and normal cognition (NC). Design Cross sectional study. Subjects Patients with a clinical diagnosis of AD, PD, MCI, or NC were imaged. Images with poor quality and of those with diabetes mellitus, glaucoma, or vitreoretinal disease were excluded from analysis. Methods Intervention or Testing All participants were imaged using the Zeiss Cirrus HD-5000 with AngioPlex (Carl Zeiss Meditec, Software Version 11.0.0.29946) and repeat OCTA images were obtained for both eyes. Perfusion density (PFD), vessel density (VD), and Foveal avascular zone (FAZ) area were measured from 3 × 3 mm and 6 × 6 mm OCTA images centered on the fovea using an ETDRS grid overlay. Main Outcome Measures Intraclass correlation coefficients were used to quantify repeatability of PFD, VD, and FAZ area measurements obtained from imaging. Results 3 × 3 mm scans of 22 AD, 40 MCI, 21 PD, and 26 NC participants and 6 × 6 mm scans of 29 AD, 44 MCI, 29 PD, and 30 NC participants were analyzed. Repeatability values ranged from 0.64 (0.49-0.82) for 6 × 6 mm PFD in AD participants to 0.87 (0.67-0.92) for 3 × 3 mm PFD in AD participants. No significant differences were observed in repeatability between NC participants and those with neurodegenerative disease. Conclusions Overall, similar OCTA repeatability was observed between NC participants and those with neurodegeneration. Regardless of diagnostic group, macular OCTA metrics demonstrated moderate to good repeatability. Financial Disclosures The authors have no proprietary or commercial interest in any materials discussed in this article.
Collapse
Key Words
- AD, Alzheimer's disease
- Alzheimer
- CI, confidence interval
- D, diopters
- FAZ, Foveal avascular zone
- ICC, intraclass correlation
- MCI, mild cognitive impairment
- MSE, mean square error
- Mild cognitive impairment
- NC, normal cognition
- OCTA
- OCTA, OCT angiography
- PD, Parkinson's disease
- PFD, Perfusion density
- Parkinson
- Repeatability
- SSI, strength signal index
- VD, vessel density
Collapse
Affiliation(s)
- Dennis Y Akrobetu
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Cason B Robbins
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Justin P Ma
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Srinath Soundararajan
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Michael S Quist
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Sandra S Stinnett
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Kathryn P L Moore
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Neurology, Duke University School of Medicine, Durham, North Carolina
| | - Kim G Johnson
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Neurology, Duke University School of Medicine, Durham, North Carolina
| | - Andy J Liu
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Neurology, Duke University School of Medicine, Durham, North Carolina
| | - Dilraj S Grewal
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
| | - Sharon Fekrat
- iMIND Research Group, Duke University School of Medicine, Durham, North Carolina
- Department of Ophthalmology, Duke University School of Medicine, Durham, North Carolina
- Department of Neurology, Duke University School of Medicine, Durham, North Carolina
| |
Collapse
|
6
|
Lee T, Rivera A, Brune M, Kundu A, Haystead A, Winslow L, Kundu R, Wisely CE, Robbins CB, Henao R, Grewal DS, Fekrat S. Convolutional Neural Network-Based Automated Quality Assessment of OCT and OCT Angiography Image Maps in Individuals With Neurodegenerative Disease. Transl Vis Sci Technol 2023; 12:30. [PMID: 37389540 PMCID: PMC10318591 DOI: 10.1167/tvst.12.6.30] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 04/01/2023] [Accepted: 06/04/2023] [Indexed: 07/01/2023] Open
Abstract
Purpose To train and test convolutional neural networks (CNNs) to automate quality assessment of optical coherence tomography (OCT) and OCT angiography (OCTA) images in patients with neurodegenerative disease. Methods Patients with neurodegenerative disease were enrolled in the Duke Eye Multimodal Imaging in Neurodegenerative Disease Study. Image inputs were ganglion cell-inner plexiform layer (GC-IPL) thickness maps and fovea-centered 6-mm × 6-mm OCTA scans of the superficial capillary plexus (SCP). Two trained graders manually labeled all images for quality (good versus poor). Interrater reliability (IRR) of manual quality assessment was calculated for a subset of each image type. Images were split into train, validation, and test sets in a 70%/15%/15% split. An AlexNet-based CNN was trained using these labels and evaluated with area under the receiver operating characteristic (AUC) and summaries of the confusion matrix. Results A total of 1465 GC-IPL thickness maps (1217 good and 248 poor quality) and 2689 OCTA scans of the SCP (1797 good and 892 poor quality) served as model inputs. The IRR of quality assessment agreement by two graders was 97% and 90% for the GC-IPL maps and OCTA scans, respectively. The AlexNet-based CNNs trained to assess quality of the GC-IPL images and OCTA scans achieved AUCs of 0.990 and 0.832, respectively. Conclusions CNNs can be trained to accurately differentiate good- from poor-quality GC-IPL thickness maps and OCTA scans of the macular SCP. Translational Relevance Since good-quality retinal images are critical for the accurate assessment of microvasculature and structure, incorporating an automated image quality sorter may obviate the need for manual image review.
Collapse
Affiliation(s)
- Terry Lee
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
| | - Alexandra Rivera
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Pratt School of Engineering, Duke University, Durham, NC, USA
| | - Matthew Brune
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Pratt School of Engineering, Duke University, Durham, NC, USA
| | - Anita Kundu
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
| | - Alice Haystead
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Pratt School of Engineering, Duke University, Durham, NC, USA
| | - Lauren Winslow
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Pratt School of Engineering, Duke University, Durham, NC, USA
| | - Raj Kundu
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Pratt School of Engineering, Duke University, Durham, NC, USA
| | - C. Ellis Wisely
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
| | - Cason B. Robbins
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
| | - Ricardo Henao
- Department of Electrical and Computer Engineering, Duke University, Durham, NC, USA
- Department of Biostatistics & Bioinformatics, Duke University, Durham, NC, USA
| | - Dilraj S. Grewal
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
| | - Sharon Fekrat
- iMIND Study Group, Duke University School of Medicine, Durham, NC, USA
- Department of Ophthalmology, Duke University School of Medicine, Durham, NC, USA
- Department of Neurology, Duke University School of Medicine, Durham, NC, USA
| |
Collapse
|