ayush14029 / Segmentation-of-Basal-Ganglia-Region-using-Stacked-Sparse-Autoencoder

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Segmentation-of-Basal-Ganglia-Region-using-Stacked-Sparse-Autoencoder

The Basal Ganglia region in the brain comprises of eight parts of the brain including the left and right Caudate, left and right Pallidus. left and right Putamen and the left and right Accumbens. This project proposes a deep learning architecture, Stacked Sparse Autoencoder, to segment the region by a voxel wise classification. A public dataset of 103 images - a mixture of diseased and healthy samples was used for this project. 50 out of these were used to train the autoencoder architecture while 33 of the remaining were used to validate the model. Validation accuracy of above 0.9 as calculated using the dice coefficient has been obtained for all the above four classes for the healthy and diseased dataset.

The project is divided among the following four parts:

  1. Data Preprocessing
  2. Feature Extraction
  3. Training models
  4. Validating the model

About


Languages

Language:MATLAB 100.0%