Skip to content

Latest commit

 

History

History
25 lines (17 loc) · 1.65 KB

File metadata and controls

25 lines (17 loc) · 1.65 KB

MSD Dataset Task05 Prostate

This repository provides a benchmarking guide and recipe to train the template algorithms, and validation performance, and is tested and maintained by NVIDIA.

Task Overview

The task is the volumetric (3D) segmentation of the prostate central gland and peripheral zone from the multi-contrast MRI (T2, ADC). The segmentation of the prostate region is formulated as the voxel-wise 3-class classification. Each voxel is predicted as either foreground (prostate central gland, peripheral zone) or background. And the model is optimized with a gradient descent method minimizing soft dice loss between the predicted mask and ground truth segmentation. The dataset is from the 2018 MICCAI challenge Medical Image Segmentation (MSD).

  • Target:
    1. Prostate central gland
    2. Prostate peripheral zone
  • Modality: MRI
  • Size: 30 3D volumes (32 Training + 16 Testing)
  • Challenge: MSD MICCAI Challenge
Validation performance: NVIDIA DGX-1 (4x V100 16)

The complete command of Auto3DSeg can be found here. And our validation results are obtained on NVIDIA DGX-1 with (4x V100 16GB) GPUs.

Methods Dimension GPUs Batch size / GPU Fold 0 Fold 1 Fold 2 Fold 3 Fold 4 Avg
SegResNet 3 4 2 0.76004 0.67638 0.68831 0.68003 0.75392 0.71174
DiNTS 3 4 2 0.71309 0.71224 0.73416 0.70840 0.74065 0.72171
SegResNet2d 3 4 2 0.71290 0.65638 0.70347 0.62220 0.70122 0.67923