Dataset Information

Generating retinal flow maps from structural optical coherence tomography with artificial intelligence.

ABSTRACT: Despite advances in artificial intelligence (AI), its application in medical imaging has been burdened and limited by expert-generated labels. We used images from optical coherence tomography angiography (OCTA), a relatively new imaging modality that measures retinal blood flow, to train an AI algorithm to generate flow maps from standard optical coherence tomography (OCT) images, exceeding the ability and bypassing the need for expert labeling. Deep learning was able to infer flow from single structural OCT images with similar fidelity to OCTA and significantly better than expert clinicians (P < 0.00001). Our model allows generating flow maps from large volumes of previously collected OCT data in existing clinical trials and clinical practice. This finding demonstrates a novel application of AI to medical imaging, whereby subtle regularities between different modalities are used to image the same body part and AI is used to generate detailed inferences of tissue function from structure imaging.

SUBMITTER: Lee CS

PROVIDER: S-EPMC6450899 | biostudies-literature | 2019 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Generating retinal flow maps from structural optical coherence tomography with artificial intelligence.

Lee Cecilia S CS Tyring Ariel J AJ Wu Yue Y Xiao Sa S Rokem Ariel S AS DeRuyter Nicolaas P NP Zhang Qinqin Q Tufail Adnan A Wang Ruikang K RK Lee Aaron Y AY

Scientific reports 20190405 1

Despite advances in artificial intelligence (AI), its application in medical imaging has been burdened and limited by expert-generated labels. We used images from optical coherence tomography angiography (OCTA), a relatively new imaging modality that measures retinal blood flow, to train an AI algorithm to generate flow maps from standard optical coherence tomography (OCT) images, exceeding the ability and bypassing the need for expert labeling. Deep learning was able to infer flow from single s ...[more]

PMID: 30952891

Similar Datasets

Project description:PurposeTo predict visual outcomes and treatment needs in a treat & extend (T&E) regimen in neovascular age-related macular degeneration (nAMD) using a machine learning model based on quantitative optical coherence tomography (OCT) imaging biomarkers.Materials and methodsStudy eyes of 270 treatment-naïve subjects, randomized to receiving ranibizumab therapy in the T&E arm of a randomized clinical trial were considered. OCT volume scans were processed at baseline and at the first follow-up visit 4 weeks later. Automated image segmentation was performed, where intraretinal (IRF), subretinal (SRF) fluid, pigment epithelial detachment (PED), hyperreflective foci, and the photoreceptor layer were delineated using a convolutional neural network (CNN). A set of respective quantitative imaging biomarkers were computed across an Early Treatment Diabetic Retinopathy Study (ETDRS) grid to describe the retinal pathomorphology spatially and its change after the first injection. Lastly, using the computed set of OCT features and available clinical and demographic information, predictive models of outcomes and retreatment intervals were built using machine learning and their performance evaluated with a 10-fold cross-validation.ResultsData of 228 evaluable patients were included, as some had missing scans or were lost to follow-up. Of those patients, 55% reached and maintained long (8, 10, 12 weeks) and another 45% stayed at short (4, 6 weeks) treatment intervals. This provides further evidence for a high disease activity in a major proportion of patients. The model predicted the extendable treatment interval group with an AUROC of 0.71, and the visual outcome with an AUROC of up to 0.87 when utilizing both, clinical and imaging features. The volume of SRF and the volume of IRF, remaining at the first follow-up visit, were found to be the most important predictive markers for treatment intervals and visual outcomes, respectively, supporting the important role of quantitative fluid parameters on OCT.ConclusionThe proposed Artificial intelligence (AI) methodology was able to predict visual outcomes and retreatment intervals of a T&E regimen from a single injection. The result of this study is an urgently needed step toward AI-supported management of patients with active and progressive nAMD.

Project description:Purpose: To investigate whether hyperreflective foci (HRF) exhibit flow projection artifact on OCTA, and study the efficacy of commercial projection artifact removal software (PAR-OCTA, Optovue, Inc), and a custom projection resolved OCTA (PR-OCTA) in distinguishing artifacts from true flow in retinal angiomatous proliferation (RAP).Methods: The study included five eyes with HRF representing pigment migration in dry age-related macular degeneration (AMD), five eyes with leaking treatment-naïve RAP, and ten eyes with diabetic hard exudates. We examined flow signal on OCTA cross-sections using PAR, and performed PR-OCTA to study the effect of increasingly stringent projection removal thresholds. Flow signal intensity was analyzed and quantified using imageJ (NIH, Bethesda, MD, USA), by calculating the percentage of red pixels (R) representing flow, compared to green (G) and blue (B) pixels.Results: PAR-OCTA cross sections revealed persistent flow signal in all HRF, including RAP, hard exudates and pigment migration. In RAP, PR-OCTA detected intransigent flow, irrespective of the flow removal threshold. Mean R in the five RAP lesions remained higher than mean G and B at the most stringent PR-OCTA threshold (40.96% vs 29.52 and 29.52%, respectively), denoting persistence of flow. In contrast, increasing the PR-OCTA threshold in pigment migration and hard exudates removed the flow signal, with a statistically significant decrease in mean R with increasing threshold. (p = 0.017 and 0.0029, respectively).Conclusion: Commercial PAR-OCTA is not completely effective at removing artifactual flow in hard exudates and HRF related to pigment migration. Custom built PR-OCTA, using a sliding scale of threshold, allowed us to distinguish true flow in RAP from artifactual flow in avascular HRF. Further studies are needed to validate the optimum threshold for projection artifact removal, which would preserve true flow in RAP and the small intraretinal capillaries.

Project description:Importance:Both parapapillary and macular areas are important in determining the progression of early glaucoma. However, no attempt has been made to assess the progression of glaucoma in images that combine the 2 areas. Objective:To evaluate the potential usefulness of serial analysis of combined wide-field optical coherence tomography (OCT) maps for detection of structural progression in patients with early glaucoma. Design, Setting, and Participants:Retrospective observational study. Patients with early primary open-angle glaucoma with a minimum of 3-year follow-up involving serial spectral-domain OCT measurement were analyzed. Patients were divided into a nonprogressor group (n?=?47) and a progressor group (n?=?47) on the basis of serial stereo disc photography and red-free photography. Serial combined wide-field OCT maps integrating parapapillary retinal nerve fiber layer (RNFL) and macular ganglion cell-inner plexiform layer (GCIPL) maps were generated with the embedded software of serial spectral-domain OCT. Glaucoma specialists then assessed the structural progression detection ability of those serial wide-field OCT maps for early glaucomatous eyes and compared their sensitivity with those of RNFL and GCIPL guided progression analyses (GPAs). Main Outcomes and Measures:The diagnostic ability of the serial wide-field OCT maps for early glaucomatous structural progression. Results:Ninety-four patients (mean [SD] age, 51.4?[12.3] years; 48 [51.1%] women; all Korean) were included. The serial wide-field OCT map analysis showed good agreement for detection of structural progression between the 2 glaucoma graders (wide-field OCT thickness map: ??=?0.649; wide-field OCT deviation map: ??=?0.833). These maps showed early glaucomatous structural progression detection abilities comparable with those of RNFL and GCIPL GPAs (sensitivities of wide-field OCT thickness map, wide-field OCT deviation map, RNFL GPA, and GCIPL GPA?=?63.8%, 83.0%, 83.0%, and 66.0%, respectively, all P?>?.05; specificities of wide-field OCT thickness map, wide-field OCT deviation map, RNFL GPA, and GCIPL GPA?=?93.6%, 95.7%, 84.8%, and 93.6%, respectively, all P?>?.05). Conclusions and Relevance:The serial combined wide-field OCT maps integrating RNFL and GCIPL maps performed well in detecting structural progression in early glaucomatous eyes. Confirmation in an independent prospective study might provide greater confidence in this conclusion.

Dataset Information

Generating retinal flow maps from structural optical coherence tomography with artificial intelligence.

Publications

Generating retinal flow maps from structural optical coherence tomography with artificial intelligence.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets