Estimation of Large Motion in Lung CT by Integrating Regularized Keypoint Correspondences into Dense Deformable Registration

Abstract

We present a novel algorithm for the registration of pulmonary CT scans. Our method is designed for large respiratory motion by integrating sparse keypoint correspondences into a dense continuous optimization framework. The detection of keypoint correspondences enables robustness against large deformations by jointly optimizing over a large number of potential discrete displacements, whereas the dense continuous registration achieves subvoxel alignment with smooth transformations. Both steps are driven by the same normalized gradient fields data term. We employ curvature regularization and a volume change control mechanism to prevent foldings of the deformation grid and restrict the determinant of the Jacobian to physiologically meaningful values. Keypoint correspondences are integrated into the dense registration by a quadratic penalty with adaptively determined weight. Using a parallel matrix-free derivative calculation scheme, a runtime of about 5 min was realized on a standard PC. The proposed algorithm ranks first in the EMPIRE10 challenge on pulmonary image registration. Moreover, it achieves an average landmark distance of 0.82 mm on the DIR-Lab COPD database, thereby improving upon the state of the art in accuracy by 15%. Our algorithm is the first to reach the inter-observer variability in landmark annotation on this dataset.

mlVIRNET: Multilevel Variational Image Registration Network

“…In general, several architectures together with different distance measures, regularizer and penalty terms can be used. However, we focus on a U-Net based architecture, combined with a loss function that has shown good results for the task of pulmonary registration [14]. The second main building block is the embedding into a multilevel approach from coarse to fine.…”

“…Typical examples for the distance measure are, e.g., the squared L 2 norm of the difference image (SSD), cross correlation (CC) or mutual information (MI). In our experiments, we follow the approach of [14] using the edge based normalized gradient fields distance measure (NGF) and second order curvature regularization.…”

An end-to-end deep learning approach for landmark detection and matching in medical images

“…The existing methods [3][4][5][6][7] for obtaining landmark correspondences in medical images are based on large and time-consuming pipelines that involve identifying landmark locations followed by matching local feature descriptors 8 within a restricted neighborhood. These methods rely upon multiple pre-and post-processing steps, multi-resolution search, and manual checking to achieve robustness; each step adding more heuristics and empirical hyperparameters to an already complex pipeline.…”

Closing the Gap Between Deep and Conventional Image Registration Using Probabilistic Dense Displacement Networks

“…For other anatomies in the abdomen, the prostate or lungs, with shape variations of several centimetres, DLIR was mainly applied to less complex cases of intra-patient registration [6,9]. For inhale-exhale lung registration the accuracy of DLIR is still inferior to conventional approaches: ≈ 2.5 mm in [13,15] compared to <1 mm in [12]. When training the state-ofthe-art weakly-supervised DLIR approach Label-Reg [6] on abdominal CT [14] for inter-patient alignment, we reached an average Dice of only 42.7%, which is still substantially worse than the conventional NiftyReg algorithm [10] with a Dice of 56.1% and justifies further research.…”