Deep-learning-based image registration for nano-resolution tomographic reconstruction

Abstract

Nano-resolution full-field transmission X-ray microscopy has been successfully applied to a wide range of research fields thanks to its capability of non-destructively reconstructing the 3D structure with high resolution. Due to constraints in the practical implementations, the nano-tomography data is often associated with a random image jitter, resulting from imperfections in the hardware setup. Without a proper image registration process prior to the reconstruction, the quality of the result will be compromised. Here a deep-learning-based image jitter correction method is presented, which registers the projective images with high efficiency and accuracy, facilitating a high-quality tomographic reconstruction. This development is demonstrated and validated using synthetic and experimental datasets. The method is effective and readily applicable to a broad range of applications. Together with this paper, the source code is published and adoptions and improvements from our colleagues in this field are welcomed.

Towards full-stack deep learning-empowered data processing pipeline for synchrotron tomography experiments

“…For applications using solely DL methods, in 2021, Fu et al. 58 proposed a residual network to correct image jitter present in nano-CT. Unlike using original projection images as network input, this model combines the original projections with the reconstructed back-projection images as input.…”

Multiscale Dense U-Net: A Fast Correction Method for Thermal Drift Artifacts in Laboratory NanoCT Scans of Semi-Conductor Chips

“…The third method uses only the original projections. The method constructs the motion model of the feature points in projections and corrects the drift by the difference between the ideal model and the drift model [ 14 , 15 , 16 , 17 , 18 ]. However, the correction accuracy of the method is limited by the projection truncation and object shape.…”