Skip to content

We present QAERTS, a parameter-efficient multi-head model for 3D fetal brain pose estimation from freehand 2D ultrasound videos by leveraging uncertainty across different geometric transformations, for use in low-income settings.

Notifications You must be signed in to change notification settings

jayroopramesh/QAERTS

Repository files navigation

QAERTS: Geometric Transformation Uncertainty for Improving 3D Fetal Brain Pose Prediction from Freehand 2D Ultrasound Videos

Training

python train_end_to_end_vunc8.py [DROPOUT_RATE] [GESTATION_AGE]

Inference

python dummy_epi_utils.py [DROPOUT_RATE] [GESTATION_AGE]

Miscellaneous

  • Trained weights are provided (*.pth) in Google Drive for all the models reported in the paper and supplementary, and their defintions can be found in epi_models_utils.py.
  • Preprocessing functions for volumes and sampled images can be found and modified for your specific data from dataset_end_to_end_vunc1.py
  • All original geometric transformations are implemented in geometry.py, and adapted in epi_models_utils.py.
  • All the necessary/dependent modules are called from these primary scripts. The datasets are not provided, but the original sources are mentioned in the paper should you wish to use them.
  • Python version used is 3.10.8. 3.7+ should work.
  • All additonal dependencies for the conda environment used can be found in requirements.txt.

About

We present QAERTS, a parameter-efficient multi-head model for 3D fetal brain pose estimation from freehand 2D ultrasound videos by leveraging uncertainty across different geometric transformations, for use in low-income settings.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages