Weight Encode Reconstruction Network for Computed Tomography in a Semi-Case-Wise and Learning-Based Way
This work addresses a bottleneck in computed tomography reconstruction for medical imaging, offering a transferable solution with incremental improvements over existing methods.
The authors tackled the problem of inaccurate voxel weight determination in classic algebraic reconstruction technology for computed tomography by proposing WERNet, a semi-case-wise learning-based method, achieving a cosine similarity greater than 0.999 with ground truth and demonstrating denoising capabilities.
Classic algebraic reconstruction technology (ART) for computed tomography requires pre-determined weights of the voxels for projecting pixel values. However, such weight cannot be accurately obtained due to the limitation of the physical understanding and computation resources. In this study, we propose a semi-case-wise learning-based method named Weight Encode Reconstruction Network (WERNet) to tackle the issues mentioned above. The model is trained in a self-supervised manner without the label of a voxel set. It contains two branches, including the voxel weight encoder and the voxel attention part. Using gradient normalization, we are able to co-train the encoder and voxel set numerically stably. With WERNet, the reconstructed result was obtained with a cosine similarity greater than 0.999 with the ground truth. Moreover, the model shows the extraordinary capability of denoising comparing to the classic ART method. In the generalization test of the model, the encoder is transferable from a voxel set with complex structure to the unseen cases without the deduction of the accuracy.