Nuclei segmentation is an important task for various computational pathalogy applications like nuclei morphology analysis, cell type classification and cancer grading. Even though Convolution Neural Networks (CNNs) are applied successfully to segment nuclei, its accuracy depends on volume and quality of histopathology data used for training. This model overcomes this problem by utilising conditional GAN. A large dataset of H&E training images with accurate nuclei segmentation labels is generated with the help of unpaired GAN framework. Proposed models trains conditional GAN network with spectral normalization and gradient penalty for multi organ nuclei segmentation. This nuclei segmentation paradigm is validated on publicly available and newly generated datasets.
Input variables : Histopathology image
Output Variables : Segmented Image
Visit Model : github.com
Additional links : arxiv.org
| Model Category | : | Public |
| Date Published | : | October, 2018 |
| Healthcare Domain | : |
Medical Technology
Provider |
| Code | : | github.com |
| Medical Imaging |
| Image Segmentation |