3D U-Net Segmentation - 3.5 English

Vitis AI Library User Guide (UG1354)

Document ID
Release Date
3.5 English

3D U-Net was introduced shortly after U-Net to process volumetric data which is abundant in medical data analysis. It is based on the previous architecture which consists of an encoder part to analyze the whole image and a decoder part to produce full resolution segmentation. 3D U-Net takes 3D volume as inputs and applies 3D convolution, 3D maxpooling and 3D up-convolutional layers unlike 2D U-Net which has an entirely 2D architecture. For more details about 3D U-Net, refer to 3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation.

The following table lists the 3D U-Net model supported by the Vitis AI Library.

Table 1. 3D U-Net Models
No Model Name Framework
1 3D-Unet_pt PyTorch