This poster is published under an
open license. Please read the
disclaimer for further details.
Keywords:
Computer applications, Bones, CT, Neural networks, CAD, Computer Applications-General, Segmentation, Technology assessment, Osteoporosis, Image verification
Authors:
K. S. Mader1, T. J. Re2, J. Cyriac2, B. Stieltjes2; 1Zurich/CH, 2Basle/CH
DOI:
10.1594/ecr2017/B-0264
Conclusion
The study presents the basis for conducting large scale studies to extract meaningful quantitative information from imaging biobanks.
The use of the latest techniques in machine learning and neural networks brought a substantial improvement over standard image processing methods but suffer from the reputation of being more difficult to understand and very black-box like in nature.
As a beginning of one possible approach for looking into the black box we investigate the different intermediate stages of the convolutional neural network.
The figure shows using a color map (red is a positive activation and blue is an inhibition),
the patterns of activation and relevant features at a number of different scales for the standard axial chest CT slice.
A close examination of these patterns and derivation of the associated textures could provide useful information for developing new approaches for analyzing bone.
For practical usability in biobank settings substantial computing resources are necessary.
For such a bone extraction and analysis tool to run on 80M images 10 days on a cluster (60 nodes) would be required.