3D-UNet Architecture Using Separable 2D Convolutions

In this decade the main challenge facing in the entire treatment sketch and the evaluation is how vast a brain tumor.one of the most dangerous reason for cancer. Accuracy in quantitative analysis and segmentation of brain are crucial for the treatment sketch. Even though many manual segmentations and magnetic resonance image has emerged they are highly time consuming and error prone.2D and 3D convolutions using neural networks cannot satisfy the whole treating plans of brain tumors even though if possible they are highly expensive in cost of its computation and the demand in its memory .Here we propose 3D UNet architecture using separable 2D convolutions.