The goal was to segment, as well as register ( align the images taken at different times ), the white & grey matter from MRI images, and finally reconstruct the brain 3D model from layers of MRI images.

We used a modified level set method, which jointly does segmentation as well as registration.

Some results are shown below:

The above picture shows one image of the raw data, where the boundary of the brain was registered with other raw data images.

The above images show the segmentation results of grey matter of T1-weighted MRI map.

The above images show the segmentation results of white matter of T1-weighted MRI map.

The above image shows the 3D reconstruction of one brain's grey matter and white matter

The above image shows the 3D reconstruction of another brain's grey matter and white matter

Publication:

• P. Chen, R. G. Steen, A. Yezzi, and H. Krim, ''Brain MRI T1-Map and T1-weighted Image Segmentation in a Variational Framework'', Acoustics, Speech, and Signal Processing, IEEE International Conference on, pp. 417-420, 2009 IEEE International Conference on Acoustics, Speech and Signal Processing, 2009. [pdf]
• P. Chen, R. G. Steen, A. Yezzi, and H. Krim, ''Joint Brain Parametric T1-Map Segmentation and RF-inhomogeneity Calibration,'' International Journal of Biomedical Imaging, 2009. [pdf]