Cell-Nuclei-Detection-and-Segmentation

Detecting the location and draw boundary of nuclei from tissue microscopic images (H&E stained). Model is based on U-net with contour enhancement in loss function. Overlap patch based strategy is used to 1) adapt to variant input image size (resize image may stretch features); 2) use random clip and rotation for data augmentation; 3) each region in output mask is determined by combining inference result from multiple patches. More details can be found in [1].

Dependencies

• Tensorflow
• OpenCV
• Scikit-image
• Numpy
• Matplotlib

More

• [x] detection and segmentation model
• [x] consider edge into loss function during training
• [x] morphology operation to calculate center and boundary
• [ ] better color normalization method for preprocess
• [ ] identify overlapping samples with local segmentation model
• [ ] identify tissue types

Reference

[1] K.Chen, N. Zhang, L.S.Powers, J.M.Roveda, Cell Nuclei Detection and Segmentation for Computational Pathology Using Deep Learning, SpringSim 2019 Modeling and Simulation in Medicine, Society for Modeling and Simuation (SCS) International.