17 November 2017

Our deep learning papers featured in Biomedical Computation Review (Stanford University)

Our deep learning papers featured in Biomedical Computation Review (Stanford University)
Illustration from the Biomedical Computation Review article (via the Internet Archive)
Biomedical Computation Review, published at Stanford University with NIH support, featured our crowdsourcing paper AggNet and other deep learning work in “Deep Learning and the Future of Biomedical Image Analysis” by Alexander Gelfand.

In “Deep Learning and the Future of Biomedical Image Analysis” (17 November 2017), science writer Alexander Gelfand featured our work in Biomedical Computation Review, the magazine published at Stanford University (James H. Clark Center) with support from the NIH.

The article describes how our crowdsourcing approach AggNet combines the ground truth of expert annotations with the “crowd truth” of non-expert ones for mitosis detection in breast cancer histology, using an aggregation layer with weighted majority voting that boosted performance by about 3 percent. It also covers our X-ray in-depth decomposition, CathNets and our playsourcing work.