The molecular basis for vertebrate color vision is expression of spectrally-selective opsin-based photopigments within distinct cone photoreceptor populations. In several primates including humans, and in numerous fish species, genes encoding duplicated, spectrally-divergent cone opsin proteins may be positioned in tandem arrays. This article reviews studies of the regulation of differential expression of these duplicated cone opsin genes in humans and fish, with an emphasis on knowledge derived from the zebrafish model. In humans, the long wavelength-sensitive/medium wavelength-sensitive (LWS/MWS; red/green) tandemly-duplicated array underlies trichromatic color vision; mutations in this array cause altered color vision and retinal degenerations. A foundational model posited that regulation of human LWS vs. MWS opsin expression is stochastic, whereby an upstream enhancer associates randomly with promoters of the proximal or distal gene, and one of these associations becomes permanent. Discoveries of tandemly-duplicated cone opsin genes in teleost fish, including the versatile zebrafish model, led to their use to further pursue mechanisms of differential expression. The tandemly-duplicated zebrafish lws1 and lws2 genes (encoding red-sensing cone opsins) share a common ancestral gene with human LWS and MWS. Subsequent studies revealed that differential expression of these genes is regulated by the paracrine signal retinoic acid (RA) and the endocrine signal thyroid hormone (TH). Further investigations identified additional cone transcripts distinguishing zebrafish LWS1 vs. LWS2 cones, plasticity of cone opsin expression across the lifespan in response to TH, and a candidate receptor and cis-regulatory element. The findings warrant updates to the foundational model and further consideration of hypotheses for regulation of the human LWS vs. MWS genes.
Cone photoreceptor density is a key quantitative marker of healthy macular microstructure and retinal remodeling related to refractive status and disease. However, absolute density estimates at similar eccentricities remain difficult to compare across studies. We conducted a scoping review of 57 in vivo studies of cone density in healthy adults. Using a three-domain framework comprising Participant/Ocular Factors, Imaging Geometry/Optics, and Analysis Methods/Algorithms, we examined how participant and ocular characteristics, imaging geometry and optics, and analysis methods and algorithms influence cone visibility, detectability, and countability. We synthesized spatial topography, differences among retinal directions, and sources of cross-study dispersion.Cone density was associated with age, refractive status, and axial length, and was further shaped by center definition, scale conversion, optical conditions, frame averaging, image-quality assessment, sampling-window design, counting algorithms, and manual correction rules. Absolute density estimates varied substantially even at similar eccentricities and retinal directions, with the largest dispersion near the fovea, where spatial gradients are steepest. After harmonizing coordinate systems, directional conventions, and spatial units, however, cone topography across retinal directions was highly consistent. Cone density declined continuously with eccentricity in all four cardinal directions, showing stable radial decay; within 0.5–6°, separate exponential models for the four cardinal directions showed robust fits (R2 = 0.899–0.950).Healthy adult cone topography is therefore reproducible and parameterizable, but reliable cross-study comparison requires complete reporting of participant and ocular characteristics, imaging geometry and optical conditions, and analysis methods and algorithms. A minimum reporting and reference framework for the visibility–detectability–countability chain is essential for translating cone density into a comparable structural outcome measure.
Artificial intelligence (AI) in retinal imaging has expanded from image classification to multimodal interpretation, longitudinal prediction, and clinically oriented decision support. This narrative review focuses on four complementary clinical settings: diabetic retinopathy (DR) screening and referral, diabetic macular edema (DME) treatment assessment, neovascular age-related macular degeneration (nAMD) retreatment and longitudinal monitoring, and inherited retinal disease (IRD) diagnosis, genotype-phenotype support, progression modeling, and trial enrichment. We organize the evidence around disease-specific decision points, relevant data modalities, and the requirements for multimodal and longitudinal integration. Evidence is strongest for DR screening, supported by prospective and real-world validation, whereas treatment-oriented applications in DME and nAMD and multimodal diagnostic or prognostic applications in IRDs remain less consistently validated. We highlight the gap between model performance and clinical utility and propose a cautious translational roadmap emphasizing external validation, calibration, uncertainty handling, workflow integration, prospective evaluation, and accountable deployment.
