Tissue Labelling Of MRI Using Recurrent Artificial Neural Networks
John W. Snell, Michael Merickel · 2005
In this paper, a method for tissue labelling of multispectral magnetic resonance images is described. Using a probabilistic, recurrent artificial neural network architecture, both spatial and grey level informtion in an image are used in concert with a priori constraints to achieve consistent tissue labelling. The resulting tissue map of the image provides information regarding structure and pathology.