'AAA' Strategies: Alternative animals, Artificial Intelligence, and Artificial Vision
: Finding New Paths to Treat Degenerative Retinal Diseases
I. Artificial Vision: From Biological Circuitry to Visual Restoration
For advanced stages of retinal degeneration where profound photoreceptor loss occurs, our research bridges the gap between biology and engineering.
Decoding Retinal Circuitry: We analyze the sophisticated signal processing of the biological retina, focusing on center-surround receptive fields, lateral inhibition, and the push-pull mechanics of bipolar and amacrine cells.
Biomimetic Computational Models: By translating these intricate neural mechanisms into mathematical models, we aim to pioneer image-processing algorithms for next-generation retinal prostheses and visual restoration technologies.
II. Alternative Animals: Deciphering Retinal Aging
Our laboratory explores the fundamental mechanisms of degenerative retinal diseases, including age-related macular degeneration (AMD) and retinitis pigmentosa (RP).
Fast-Aging In Vivo Model: We utilize the African Turquoise Killifish, a unique vertebrate model that exhibits rapid, human-like aging phenotypes, to study the progression of retinal degeneration efficiently.
Targeting Cellular Senescence: We investigate senolytic strategies (e.g., Nutlin-3a, ABT-263) to selectively eliminate senescent retinal pigment epithelium (RPE) cells and ameliorate retinal degeneration.
Gene Therapy Innovations: We are developing cell-penetrating siRNA-mediated NRL knockdown strategies to induce rod-to-cone transdifferentiation, aiming to preserve photoreceptor integrity in both RP and AMD environments.
III. Artificial Intelligence: Data-Driven Precision Ophthalmology
We harness the power of Artificial Intelligence to transform complex, multimodal clinical and animal data into actionable insights for personalized patient care.
Classification of Brain Diseases based on Ophthalmic Data: CADASIL, Alzheimer disease severity classification through retinal images.
Automated Segmentation: We develop robust Convolutional Neural Networks (CNNs) to precisely quantify critical imaging biomarkers, such as macular fluid, subretinal hyperreflective material (SHRM), and Haller vessel morphology from OCT and OCTA.
Predicting Long-Term Outcomes: Utilizing Generative Adversarial Networks (GANs), our models successfully predict and generate 12-month post-treatment OCT images from baseline clinical data.
Interpretable Machine Learning: We apply advanced ML techniques (e.g., SHAP) to predict visual acuity outcomes, providing clinicians with transparent, predictive roadmaps for nAMD treatments.