3D Nuclei Segmentation through Deep Learning

This study presents a novel approach to Drosophila 3D nuclei segmentation. Our method involves a pipeline that identifies nuclei centres and segments each detected nucleus individually, employing distinct 3D U-net models for the detection and segmentation steps. Remarkably, our technique emerged as one of the top three performers in the Cell Tracking Challenge segmentation benchmark for the Light Sheet Microscopy Drosophila dataset, achieving an impressive final score of 0.827.