MicroNucML enables machine learning-based micronuclei segmentation and micronuclei-nuclei association.

Micronuclei (MN) are structures containing small DNA fragments that arise from mitotic errors or failed DNA repair and serve as markers of genome instability. MN are typically quantified manually or with threshold methods, which can be tedious and inaccurate, leading to variable success and throughput. By employing a two-phase labeling approach that uses polygon and brush segmentation, along with SAM2-based refinement, we developed a high-quality MN segmentation tool. Data augmentation capturing heterogeneity in image quality and color diversity enabled us to train a generalizable Mask region-based convolutional neural network (Mask-RCNN) model optimized for small-object detection, achieving state-of-the-art performance in MN detection. Finally, we applied our model to immunofluorescence data from cell lines exposed to DNA damage conditions to gain biological insights into MN dynamics and their role in genome instability. In summary, this work establishes an accessible resource for systematically studying genome instability with greater fidelity and sensitivity, enabling previously unresolved insights into damage biology.