Providing computing solutions for exascale challenges
CamNET is a deep learning platform for digital histopathology. This 1st layer provides the data pre-processing and patch extraction functionality.
CamNET is a deep learning platform for digital histopathology aimed at detecting tumor tissue in gigapixel scans of slides containing breast lymph nodes tissue. Currently, a 'human' expert needs to inspect these slides in order to find suspected tumors, which is very labor intensive. CamNET attempts to automate this using a deep learning approach. CamNET software is organized into three layers.
This first layer implements the extraction of 224x224 pixel patches from the original gigapixel slides of breast lymph nodes tissue. These patches are randomly sampled from the slide, which is annotated by a physician to point out the tumor areas. Patches belonging to a tumorous region are assigned a ‘tumor’ label (a Boolean variable equals to True). The extracted data are stored in an intermediate dataset with the corresponding labels.