English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Zendy Plus

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Zendy Tools

Zendy Open

The use of digitally reconstructed radiographs (DRRs) to solve inverseproblems such as slice-to-volume registration and 3D reconstruction iswell-studied in preoperative settings. In intraoperative imaging, the utilityof DRRs is limited by the challenges in generating them in real-time andsupporting optimization procedures that rely on repeated DRR synthesis. Whileimmense progress has been made in accelerating the generation of DRRs throughalgorithmic refinements and GPU implementations, DRR-based optimization remainsslow because most DRR generators do not offer a straightforward way to obtaingradients with respect to the imaging parameters. To make DRRs interoperablewith gradient-based optimization and deep learning frameworks, we havereformulated Siddon's method, the most popular ray-tracing algorithm used inDRR generation, as a series of vectorized tensor operations. We implementedthis vectorized version of Siddon's method in PyTorch, taking advantage of thelibrary's strong automatic differentiation engine to make this DRR generatorfully differentiable with respect to its parameters. Additionally, usingGPU-accelerated tensor computation enables our vectorized implementation toachieve rendering speeds equivalent to state-of-the-art DRR generatorsimplemented in CUDA and C++. We illustrate the resulting method in the contextof slice-to-volume registration. Moreover, our simulations suggest that theloss landscapes for the slice-to-volume registration problem are convex in theneighborhood of the optimal solution, and gradient-based registration promisesa much faster solution than prevailing gradient-free optimization strategies.The proposed DRR generator enables fast computer vision algorithms to supportimage guidance in minimally invasive procedures. Our implementation ispublically available at https://github.com/v715/DiffDRR.

Fast Auto-Differentiable Digitally Reconstructed Radiographs for Solving Inverse Problems in Intraoperative Imaging