Progressive holistically nested networks (P-HNN) is an adaptation of the HNN model for semantic segmentation (note HNN was orignally called holistically nested edge detection). It is a deep learning solution to semantic segmentation. It adapts HNN by adding progressive multi-path pathways, which helps produce more refined segmentation maps and avoids the output ambiguity of original HNN. We applied P-HNN to pathological lung segmentation.

    For more details, please consult our publication:

    A.P. Harrison, Z. Xu, K. George, L. Lu, R.M. Summers, and D.J. Mollura, "Progressive and Multi-Path Holistically Nested Neural Networks for Pathological Lung Segmentation from CT Images", MICCAI 2017.

    For our work, we implemented P-HNN using the Caffe framework for deep learning. We provide free access to our inference code, trained model, and model specifications for non-commerical interests, e.g., academic and governmental. If you are interested in accessing our code, please sign up for a free gitlab account and send us an email, and include your intended use, institution, and gitlab username. We will then grant you access.

    Questions or comments are very welcome! If you use our code, we ask that you cite the above publication.

    HED Model

    Comparison Images