Feature Processing for Automatic Anatomical Landmark Detection Using Reservoir Networks

We present an approach to the combination of an arbitrary number of image features to produce more sophisticated features for anatomical landmark detection. The combination was done using reservoir networks. The results were compared to Gabor wavelet features on single point detection in 2D slices of MR image data to show behavior and potential quality of the combined features.

• Type: Conference
• Year: 2009
• Where: BILDMED
• Authors: Benjamin Roeschies, Susanne Winter
• Keywords: Anatomical Landmark Detection, BILDMED 2007, BILDMED 2009, Single Point Detection, Sophisticated Features