Convolutional neural networks to predict brain tumor grades and Alzheimer’s disease with MR spectroscopic imaging data
Fig 4
Example of identification of important spectral regions using stability selection on the output of SL-CNN.
Stability selection is applied to the output of the convolutional layer after the neural network has been trained. Based on the convolutional layer parameters (i.e. number and size of the kernels and stride of the convolution), it is possible to map them to a specific region of the input spectrum.