Deep-Learning-Based CT Lung Registration


Logo for Deep-Learning-Based CT Lung Registration

About

Creators:
Image Version:
fb3bf078-ba1e-417b-a138-58745e96c4cc
Last updated:
Sept. 3, 2021, 7:01 a.m.
Associated publication:
Hering A, Häger S, Moltz J, Lessmann N, Heldmann S, van Ginneken B. CNN-based lung CT registration with multiple anatomical constraints. Medical Image Analysis. 2021;72:102139.
Inputs:
  • Fixed Image  (The target image for registration)
  • Moving Image  (The image that will be warped during registration to the fixed image)
  • Fixed Mask  (Mask of the locations in the fixed image to consider)
  • Moving Mask  (Mask of the locations in the moving image to consider)
Outputs:
  • Fixed Image  (The target image for registration)
  • Warped Image  (The moving image warped to the fixed image)
  • Displacement Field  (Displacement vectors to align the moving image to the fixed image)

Model Facts

Summary

This deep-learning-based lung registration developed in [1] uses multiple anatomical constraints to supervise the training. For preprocessing, the lung mask of the inspiration and expiration scan are required. The algorithms will crop the input images to the lung regions and outputs:

  cropped fixed image
  warped moving image
  displacement field in mm

For further questions or comments please contact us at alessa.hering@mevis.fraunhofer.de

[1] Hering, A., Häger, S., Moltz, J., Lessmann, N., Heldmann, S., & van Ginneken, B. (2021). CNN-based Lung CT Registration with Multiple Anatomical Constraints. Medical Image Analysis, 102139.

Mechanism

Validation and Performance

Uses and Directions

This algorithm was developed for research purposes only.

Warnings

Common Error Messages

Information on this algorithm has been provided by the Algorithm Editors, following the Model Facts labels guidelines from Sendak, M.P., Gao, M., Brajer, N. et al. Presenting machine learning model information to clinical end users with model facts labels. npj Digit. Med. 3, 41 (2020). 10.1038/s41746-020-0253-3