3D Nuclei Segmentation through Deep Learning
Published in 2023 IEEE Conference on Artificial Intelligence (CAI), 2023
Recommended citation: Rojas, R., Navarro, C. F., Orellana, G. A., Lemus, C. G., & Castañeda, V. (2023). 3D nuclei segmentation through deep learning. Proceedings of the IEEE Conference on Artificial Intelligence (CAI), 309–310. https://doi.org/10.1109/CAI54212.2023.00137 https://ieeexplore.ieee.org/abstract/document/10195067
Co-developed a deep learning pipeline for 3D nuclei segmentation in microscopy data, combining detection and segmentation using specialized 3D U-Net architectures.
The approach separates nucleus localization and instance segmentation into dedicated models, improving robustness and performance in complex volumetric datasets.
The system achieved top-3 performance in the Cell Tracking Challenge (Light Sheet Microscopy dataset), demonstrating strong accuracy in a competitive international benchmark.