Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper explores the use of Deep Image Prior (DIP), an unsupervised deep learning technique, to improve electron tomography (ET) reconstruction from sparse-view and limited-angle data. The DIP approach leverages a convolutional neural network as a prior to regularize the ill-posed inverse problem of tomographic reconstruction. Results on simulated data show DIP achieves performance comparable to supervised methods, and application to experimental data demonstrates its ability to enable reliable 3D quantification under challenging acquisition conditions.
Achieve high-quality electron tomography reconstructions from sparse and limited-angle data without needing any training data, thanks to a clever application of deep image priors.
Electron tomography (ET) plays an important role in the three-dimensional (3D) characterization of nanomaterials. However, under limited-angle and sparse-view conditions, conventional algorithms produce degraded reconstructions, which compromise the quality and interpretability of resulting 3D data. In this paper, we present deep image prior (DIP), an unsupervised deep learning (DL) approach, for highly degraded tomography acquisitions and demonstrate, using simulated data, that its performance is comparable to that of supervised approaches requiring training datasets, even for tilt ranges as limited as 60° and tilt increments of 10°. We then apply it to experimental data and show that it enables reliable 3D quantification under both sparse-view and limited-angle conditions, highlighting its potential for a wide range of materials and acquisition modalities.