2012 Winter Project Week:PairwiseLF
- Ramesh Sridharan
- Christian Wachinger
- Polina Golland
Our goal is to improve the performance of label fusion for high-variability datasets in which registration algorithms may not align subjects very well. We will investigate two directions for improving label fusion:
- We will examine the potential of using parameters learned over the label fusion training set to improve the quality of segmentation
- We will examine pairwise interactions between images in the training set for label fusion.
Our approach is to learn intensity and label prediction parameters over pairs of images in the training set.
Our plan for the Project Week is to apply our method to MR and CT data, including the Head and Neck Cancer dataset.
- Discussed data and motivation with fellow project week participants
- Discussed registration issues with data
- Continued to apply method to synthetic data