Brain tumor detection from medical images is essential to diagnose earlier and to take decision in treatment planning. Magnetic Resonance Images (MRI) is frequently preferred for detecting brain tumors by the physicians. This paper analyses various Artificial Neural Networks (ANN) training
functions for brain tumor segmentation such as Levenberg-Marquardt (LM), Quasi Newton back propagation (QN), Bayesian regularization (BR), Resilient back propagation algorithm (RP) and Scaled conjugate gradient back propagation (SCG). The training algorithms were employed in different sized
network for segmentation. The results were carefully analyzed and measured using Dice similarity, sensitivity, specificity and accuracy measures.
3D Convolutional Neural Networks for Brain Tumor Segmentation: A Comparison of Multi-resolution Architectures
