Deep-Hipo (deep learning for HIstoPathology) can accurately analyze histopathological images by learning multi-scale morphological patterns from various magnification levels of patches in a WSI simultaneously. Deep-Hipo computes a cancer probability on each pixel, sliding multi-scale windows over a histopathological image, and then visualizes the probability scores to locate cancerous regions. The set of probability scores generated by Deep-Hipo on a histopathological image can be used for cancer classification, detection, and segmentation.

The manuscript is currently under review.

Source code:

Docker image: