Promising Hyperparameter Configurations for Deep Fully Connected Neural Networks to Improve Image Reconstruction in Proton Radiotherapy.
Sokhna A. YorkAlina M. AliDavid C. LashbrookeRodrigo Yepez-LopezCarlos A. BarajasMatthias K. GobbertJerimy C. PolfPublished in: IEEE BigData (2021)
Keyphrases
- image reconstruction
- fully connected
- neural network
- maximum a posteriori
- super resolution
- emission computed tomography
- emission tomography
- reconstruction method
- image reconstruction algorithms
- respiratory motion
- artificial neural networks
- multilayer perceptron
- sparsely sampled
- positron emission tomography
- activation function
- conditional random fields
- medical images
- fuzzy logic
- high quality
- image processing