Model Architecture Selection
========================================

Hyperparameter optimization for the proposed architecture is done using `kerastuner <https://www.tensorflow.org/tutorials/keras/keras_tuner>`_. It provides capability to optimize and select different architectures depending on the application considered. The results for hyperparameter optimizations are shown below.

----------------

.. figure:: hp_cnn.png
   :align:   center

   *Fig 1: Bayesian 3D CNN Hyperparameter Optimization*


----------------

.. figure:: hp_unet.png
   :align:   center

   *Fig 1: Bayesian 3D U-Net Optimization**