Multispectral Convolutional Sparse Coding

Background:

Multispectral Convolutional Sparse Coding is a machine learning algorithm that was initially developed for the classification of distinct regions of microanatomy and histopathology.  Examples can be found as follows,

Description:

This project contains the source code (matlab) for Multispectral Convolutional Sparse Coding that is described in the publication as follows,

  • Yin Zhou*, Hang Chang*, Kenneth Barner, Paul Spellman, and Bahram Parvin. “Classification of Histology Sections via Multispectral Convolutional Sparse Coding.” IEEE Conference on Computer Vision and Pattern Recognition (CVPR 2014), Columbus, Ohio, U.S, June 2014. (*Co-First Authors)

It provides API functions, examplar data and a demo function, with which the learnt fiter banks from separate spectral (corresponding to Haematoxylin and Eosin stain, respectively) are illustrated as follows,

Filters learnt from Nuclear ChannelFilters learnt from Extra Celluar Matrix Channel

After unsupervised convolutional filter learning, the feature extraction pipeline is as follows,

 

Downloads

Project category: 
Source Code / Software