Abstract
PURPOSE: The purpose of this study was to present a suite of open-source software tools for automated and interactive annotation (detection and segmentation) of retinal cells from adaptive optics (AO) images.
METHODS: Three related software packages were developed, implementing machine learning (ML) approaches for cone photoreceptor detection and segmentation and retinal pigment epithelial (RPE) cell detection in AO images. The software packages were designed to have interactive features allowing multiple users to adjust annotations for systematic grading.
RESULTS: The software packages have been extensively tested for usability to enable reliable annotation of cells in AO images.
CONCLUSIONS: This open-source software suite provides tools for quantitative analysis of retinal cells across different AO modalities, supporting reproducible and standardized measurements across studies and imaging platforms.
TRANSLATIONAL RELEVANCE: By lowering technical barriers to AO image analysis, these tools enable broader adoption of cellular resolution biomarkers in retinal research and clinical studies, including longitudinal monitoring and therapeutic evaluation.