NOA: a versatile, extensible tool for AI-based organoid analysis

TL;DR

NOA (Napari Organoid Analyzer) addresses the accessibility and interoperability gap in AI-based organoid image analysis by providing a general-purpose, open-source napari plugin that unifies detection, segmentation, tracking, feature extraction, annotation, and ML-based prediction. It leverages four pre-trained detectors plus a SAM-based segmentation approach, supports multi-channel and timelapse data, and enables in-tool ML model training with a modular pipeline. The authors validate NOA with three mouse intestinal organoid case studies—morphology profiling, phototoxicity assessment, and phenotype prediction—demonstrating practical utility and extensibility. The work offers a flexible, extensible framework that can be adapted to diverse organoid models and tasks, bridging automated analysis and domain-specific curation for reproducible research.

Abstract

AI tools can greatly enhance the analysis of organoid microscopy images, from detection and segmentation to feature extraction and classification. However, their limited accessibility to biologists without programming experience remains a major barrier, resulting in labor-intensive and largely manual workflows. Although a few AI models for organoid analysis have been developed, most existing tools remain narrowly focused on specific tasks. In this work, we introduce the Napari Organoid Analyzer (NOA), a general purpose graphical user interface to simplify AI-based organoid analysis. NOA integrates modules for detection, segmentation, tracking, feature extraction, custom feature annotation and ML-based feature prediction. It interfaces multiple state-of-the-art algorithms and is implemented as an open-source napari plugin for maximal flexibility and extensibility. We demonstrate the versatility of NOA through three case studies, involving the quantification of morphological changes during organoid differentiation, assessment of phototoxicity effects, and prediction of organoid viability and differentiation state. Together, these examples illustrate how NOA enables comprehensive, AI-driven organoid image analysis within an accessible and extensible framework.

Figures (4)

• Figure 1: NOA workflow: 1) Loading of organoid images and timelapses. 2) Detection using pre-trained models. Previous results are cached and can be restored. 3) Filtering and manual editing of detections. For timelapses, tracking can align unique detection IDs across frames. 4) Instance segmentation with SAM. Added signal images are also segmented. 5) Computation of organoid features via 5.1) ML models, trained with a built-in training pipeline or 5.2) built-in annotation tools. 6) Data export.
• Figure 2: Example of the NOA interface: napari viewer controls (left panel), interactive image view displaying detections/segmentation (center), and NOA's organoid analysis instruments (right panel). The image demonstrates detection and segmentation results (red boxes and multi-colored masks) within the user-defined ROI (green box).
• Figure 3: Morphological changes of organoids due to differentiation. A) Examples of the annotated differentiation stages and features (red: wall thickness, blue: bud length). B) Geometric features (size and roundness) over time. C) Manually annotated features distinguish differentiation stages.
• Figure 4: Phototoxicity effect measured through mean signal intensity from brightfield, GFP, and cl.Casp3-stained images. Scalebars are 500 µm.