This post focuses on Deep Learning, how it works, and an example of its application in voxel encoding in medical applications.
Deep Learning is a method of machine learning carrying the difference of actually not being fed with every bit of information, but rather observing different cases of the same subject and identifying patterns by itself. This increased the speed of the learning process but also increased the complexity behind the pattern identification and limits the human understanding. However, one useful feel of applications for deep learning is voxel encoding, especially in the medical field.
A voxel is basically spoken a pixel in a three dimensional way. Voxel representations are for example used when analysing 3D images in medical purposes, for example in brain research.
In order to enable analytics of voxel, deep learning is programmed on a super computer which is having access to a data base of millions of 3D brain pictures. The algorithms of the deep learning application analyse and compare these pictures voxel on voxel (e.g. 3D pixel on 3D pixel) to assess similarities, differences, and patterns. Based on the recognised patterns, the computer is able to interpret these as a symptom for example. This is useful for fast and exact identifications of sicknesses in patients, for example in stroke identification.
Given the importance of this new technological application, deep learning on voxel encoding in the medical field has a bright future with may business cases and deserve attention in nowadays Information Systems Literature.
For more information, see: http://www.jneurosci.org/content/35/48/15769
